Mercury User`s Manual

Mercury User`s Manual
Digital X-ray Processor
User’s Manual
Models Mercury and Mercury-4
With
Prospect Software
version 1.0.x
XIA LLC
31057 Genstar Road
Hayward, CA 94544 USA
Tel: (510) 401-5760; Fax: (510) 401-5761
http://www.xia.com/
Information furnished by XIA LLC is believed to be accurate and reliable. However, no
responsibility is assumed by XIA LLC for its use, nor for any infringements of patents or other
rights of third parties which may result from its use. No license is granted by implication or
otherwise under any patent or patent rights of XIA LLC. XIA LLC reserves the right to change
specifications at any time without notice. Patents have been applied for to cover various aspects
of the design of the DXP Digital X-ray Processor (DXP).
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Safety .............................................................................................................................. v
Specific Precautions ............................................................................................... v
Power Source .................................................................................................. v
User Adjustments/Disassembly ....................................................................... v
Servicing and Cleaning .................................................................................... v
Manual Conventions ....................................................................................... vi
End Users Agreement ................................................................................................. vii
Contact Information: ............................................................................................. vii
8/13/2010
1
Introduction.............................................................................................................. 1
1.1 Mercury Features ............................................................................................. 1
1.2 Data Acquisition Modes ................................................................................... 2
1.2.1 MCA Mode .......................................................................................... 2
1.2.2 MCA Mapping Mode ........................................................................... 2
1.2.3 SCA Mapping Mode ............................................................................ 3
1.3 System Requirements:..................................................................................... 4
1.3.1 Host Computer .................................................................................... 4
1.3.2 Detector/Preamplifier .......................................................................... 4
1.3.3 DXP Mercury Power Supplies ............................................................ 5
1.3.4 Cabling ................................................................................................ 5
1.4 Software and Firmware Overview.................................................................... 5
1.4.1 User Interface: ProSpect .................................................................... 6
1.4.2 Device Driver: Handel ......................................................................... 6
1.4.3 Firmware and FDD Files ..................................................................... 6
1.4.4 Initialization File .................................................................................. 6
1.5 Support ............................................................................................................ 7
1.5.1 Software and Firmware Updates ........................................................ 7
1.5.2 Related Documentation ...................................................................... 7
1.5.3 Technical Support ............................................................................... 7
1.5.4 Feedback ............................................................................................ 8
2
Installation.............................................................................................................. 10
2.1 Software Installation....................................................................................... 10
2.1.1 Running the Installer ......................................................................... 10
2.1.2 File Locations .................................................................................... 10
2.1.3 Support ............................................................................................. 11
2.2 Configuring the Analog Signal Conditioner .................................................... 11
2.2.1 Input Attenuation: JP100 .................................................................. 11
2.3 Making Connections ...................................................................................... 12
2.3.1 Signal Connections ........................................................................... 12
2.3.2 GATE/SYNC Connection .................................................................. 13
2.4 Starting the System........................................................................................ 13
2.4.1 DXP Mercury Driver Selection .......................................................... 13
3
System Configuration ........................................................................................... 14
3.1 Initialization Files ............................................................................................ 14
3.1.1 Starting ProSpect Without an INI File ............................................... 14
3.2 The Configuration Wizard .............................................................................. 15
3.2.1 General Settings ............................................................................... 15
3.2.2 Hardware Synchronization Settings.................................................. 18
3.2.3 Mapping Mode Settings .................................................................... 19
3.2.4 Completing the Configuration ........................................................... 20
3.3 Loading and Saving Initialization Files........................................................... 20
i
DXP Mercury / Prospect User Manual
3.3.1
3.3.2
8/13/2010
Prospect 1.0.x
Loading an INI file ............................................................................. 20
Saving an INI file ............................................................................... 20
4
Using ProSpect with the Mercury ........................................................................ 22
4.1 A Quick Tour of ProSpect .............................................................................. 22
4.1.1 Channel Selection ............................................................................. 22
4.1.2 Settings Sidebar................................................................................ 23
4.1.3 Main Window .................................................................................... 23
4.2 Detector and Preamplifier Settings ................................................................ 24
4.2.1 Pre-Amplifier Polarity ........................................................................ 24
4.2.2 Reset Interval .................................................................................... 25
4.2.3 Preamp Gain ..................................................................................... 25
4.2.4 Preamp Risetime .............................................................................. 25
4.2.5 Saving the Configuration File ............................................................ 25
4.3 Normal Spectrum Mode Data Acquisition ...................................................... 26
4.3.1 Starting a Run ................................................................................... 26
4.3.2 Skipping Channels ............................................................................ 27
4.3.3 Spectrometer Settings ...................................................................... 27
4.3.4 Setting Regions of Interest (ROIs) .................................................... 30
4.3.5 Gain Calibration ................................................................................ 31
4.3.6 Saving and Loading INI Files ............................................................ 34
4.3.7 Output Statistics ................................................................................ 35
4.3.8 Single Channel Analyzer (SCA) ........................................................ 36
4.3.9 Saving and Loading Data ................................................................. 37
4.4 Run Control .................................................................................................... 38
4.4.1 Run Presets (Automatic Run Termination) ....................................... 38
4.4.2 The GATE Function .......................................................................... 39
4.4.3 Resume Run: Clear or Retain MCA Data ......................................... 39
4.5 Display Controls ............................................................................................. 40
4.5.1 MCA Auto Update / Refresh Rate..................................................... 40
4.5.2 Graphical Display Tools .................................................................... 40
4.6 Optimizations ................................................................................................. 43
4.6.1 Throughput (OCR) ............................................................................ 43
4.6.2 Pileup Rejection ................................................................................ 45
4.6.3 Energy Resolution............................................................................. 46
4.7 Diagnostics .................................................................................................... 48
4.7.1 The ADC Panel (Oscilloscope) ......................................................... 48
4.7.2 The Baseline Panel ........................................................................... 55
4.7.3 DSP Parameters ............................................................................... 57
4.7.4 Submitting a problem report: ............................................................ 58
5
Mapping Mode ....................................................................................................... 60
5.1 Pixel Advance Settings .................................................................................. 60
5.1.1 Pixel Advance on GATE Edge .......................................................... 60
5.1.2 Pixel Advance using SYNC Clock .................................................... 61
5.1.3 Pixel Advance under Host Control .................................................... 62
5.2 Mapping Mode Data Acquisition .................................................................... 62
5.2.1 The Mapping Panel ........................................................................... 62
5.2.2 Mapping Mode: MCA or SCA .......................................................... 62
5.2.3 Total Number of Pixels ..................................................................... 63
5.2.4 Buffer Control .................................................................................... 63
5.2.5 Mapping Mode Data Acquisition ....................................................... 63
5.3 Mapping Mode Data....................................................................................... 64
5.3.1 Mapping Data Options ...................................................................... 64
5.3.2 Mapping Data Format ....................................................................... 64
ii
DXP Mercury / Prospect User Manual
5.3.3
Prospect 1.0.x
Single Buffer Format ......................................................................... 65
6
Digital Filtering: Theory of Operation and Implementation Methods .............. 66
6.1 X-ray Detection and Preamplifier Operation: ................................................. 66
6.1.1 Reset-Type Preamplifiers ................................................................. 66
6.1.2 RC-Type Preamplifiers ..................................................................... 67
6.2 X-ray Energy Measurement & Noise Filtering: .............................................. 68
6.2.1 Digital Filtering Theory ...................................................................... 68
6.2.2 Trapezoidal Filtering ......................................................................... 70
6.3 Trapezoidal Filtering in the DXP: ................................................................... 71
6.3.1 Comparing DXP Performance .......................................................... 71
6.3.2 Decimation and Peaking Time Ranges ............................................ 71
6.3.3 Time Domain Benefits of Trapezoids................................................ 72
6.4 Baseline Issues: ............................................................................................. 73
6.4.1 The Need for Baseline Averaging ..................................................... 73
6.4.2 Raw Baseline Measurement ............................................................. 75
6.4.3 Baseline Average Settings and Recommendations ......................... 75
6.4.4 Why Use a Finite Averaging Length? ............................................... 76
6.5 X-ray Detection & Threshold Setting: ............................................................ 76
6.6 Peak Capture Methods .................................................................................. 77
6.6.1 Setting the Gap Length ..................................................................... 78
6.6.2 Peak Sampling vs. Peak Finding ...................................................... 78
6.7 Energy Measurement with Resistive Feedback Preamplifiers ...................... 80
6.8 Pile-up Inspection: ......................................................................................... 83
6.9 Input Count Rate (ICR) and Output Count Rate (OCR): ............................... 85
6.10
Throughput: ............................................................................................. 86
6.11
Dead Time Corrections: .......................................................................... 88
7
DXP Mercury Hardware Description .................................................................... 89
7.1 DXP Mercury Overview.................................................................................. 89
7.1.1 The Digital X-ray Processor (DXP) ................................................... 89
7.1.2 Rapid Data Readout ......................................................................... 90
7.2 Timing and Synchronization Logic ................................................................. 91
7.2.1 GATE Function: MCA Mode ............................................................. 91
7.2.2 GATE Function: Mapping Mode ....................................................... 91
7.2.3 SYNC Function: Mapping Mode ....................................................... 93
7.3 The Analog Signal Conditioner (ASC): .......................................................... 94
7.4 Analog to Digital Converter ............................................................................ 95
7.5 The Filter, Pulse Detector, & Pile-up Inspector (FiPPI): ................................ 95
7.5.1 FiPPI Configuration ........................................................................... 96
7.5.2 FiPPI Version and Variants ............................................................... 96
7.5.3 FiPPI Decimation .............................................................................. 96
7.5.4 Digital Trapezoidal Filtering .............................................................. 96
7.5.5
Statistics ............................................................................................ 97
7.6 The Digital Signal Processor (DSP): .............................................................. 98
7.6.1 Event Processing .............................................................................. 98
7.6.2 Statistics ............................................................................................ 98
7.7 System FPGA ................................................................................................ 98
7.7.1 Basic 32-bit MCA Data Acquisition ................................................... 99
7.7.2 Full Spectrum 16-bit MCA Mapping/Scanning Mode...................... 100
7.7.3 Other Data Acquisition Modes ........................................................ 100
Appendices ................................................................................................................. 101
8/13/2010
iii
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Appendix A. Accessing the Circuit Board in Bench-Top Models ....................... 101
Appendix B. Mercury Revision C Circuit BoardDescription ............................... 102
B.1. Jumper Settings.................................................................................... 103
B.2. LED Indicators ...................................................................................... 103
B.3. Connectors ........................................................................................... 105
Appendix C. Mercury-4 Revision A Circuit Board Description ........................... 107
C.1 Jumper Settings .................................................................................... 108
C.2. LED Indicators ...................................................................................... 108
C.3 Connectors ............................................................................................ 110
Appendix D. Specification for ROI outputs on the Mercury and Mercury4 Auxiliary
Port............................................................................................................... 113
D.1 Signal Assignment................................................................................. 113
D.2 Signal Descriptions................................................................................ 115
D.2.1. ROI Outputs ...................................................................................... 115
D.2.2. Trigger and Live Time Outputs.......................................................... 116
D.3. Register Definitions .............................................................................. 117
D.3.1. FiPPI Registers ................................................................................. 117
D.3.2. SysFPGA Registers .......................................................................... 117
Appendix E. Mapping Buffer Specification ......................................................... 118
E.1. Buffer Header ....................................................................................... 118
E.2. Pixel Data Block ................................................................................... 119
E.2.1. Mapping Mode 1: Full Spectrum Mapping ........................................ 120
E.2.2. Mapping Mode 2: Multiple SCA Mapping .......................................... 121
E.2.3. Mapping Mode 3: List Mode Mapping ............................................... 122
8/13/2010
iv
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Safety
Please take a moment to review these safety precautions. They are
provided both for your protection and to prevent damage to the digital x-ray
processor (DXP) and connected equipment. This safety information applies to
all operators and service personnel.
Specific Precautions
Observe all of these precautions to ensure your personal safety and to
prevent damage to either the DXP Mercury or equipment connected to it.
Power Source
The DXP Mercury is intended to operate from a set of DC voltage
supplies specified in section 1.3.3. To avoid damage to the DXP Mercury
ensure that the power supply meets these specifications before attempting to
power on. For the DXP Mercury bench-top models, all DC voltages necessary
for the operation of the signal processor are generated internally, and AC
voltage is supplied to the rear panel, as specified in section 1.3.3.1.
User Adjustments/Disassembly
To avoid personal injury, and/or damage, always turn off power before
accessing the Mercury.
Servicing and Cleaning
The DXP hardware is warranted against all defects for 1 year. Please
contact the factory or your distributor before returning items for service. To
avoid personal injury, and/or damage to the DXP Mercury, do not attempt to
repair or clean the unit.
8/13/2010
v
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Manual Conventions
Through out this manual we will use the following conventions:
Convention
»
Description
The » symbol leads you
through nested menu
items and dialog box
options.
Bold
Bold text denotes items
that you must select or
click on in the software,
such as menu items,
and dialog box options.
Bold text within [ ]
denotes a command
button.
Items in this font
denote text or
characters that you
enter from the
keyboard, sections of
code, file contents, and
syntax examples.
Text in quotation refers
to window titles, and
quotations from other
sources
Italic text denotes a new
term being introduced ,
or simply emphasis
[Bold]
monospace
“window”
Italics
<Key>
<Shift-Alt-Delete>
or <Ctrl+D>
Bold italic
CAPITALS
8/13/2010
Angle brackets denote a
key on the keybord (not
case sensitive).
A hyphen or plus
between two or more
key names denotes that
the keys should be
pressed simultaneously
(not case sensitive).
Warnings and
cautionary text.
CAPITALS denote
DSP parameter names
Example
The sequence
File»Page Setup»Options
directs you to pull down the
File menu, select the Page
Setup item, and choose
Options from the sub menu.
...click on the MCA tab.
[Start Run] indicates the
command button labeled
Start Run.
Setup.exe refers to a file
called “setup.exe” on the host
computer.
“Options” indicates the
window accessed via
Tools»Options.
peaking time refers to the
length of the slow filter.
...it is important first to set
the energy filter Gap so that
SLOWGAP to at least one
unit greater than the
preamplifier risetime...
<W> indicates the W key
<Ctrl+W> represents holding
the control key while pressing
the W key on the keyboard
CAUTION: Improper
connections or settings can
result in damage to system
components.
SLOWLEN is the length of
the slow energy filter
vi
DXP Mercury / Prospect User Manual
Prospect 1.0.x
End Users Agreement
XIA LLC warrants that this product will be free from defects in
materials and workmanship for a period of one (1) year from the date of
shipment. If any such product proves defective during this warranty period,
XIA LLC, at its option, will either repair the defective products without charge
for parts and labor, or will provide a replacement in exchange for the defective
product.
In order to obtain service under this warranty, Customer must notify
XIA LLC of the defect before the expiration of the warranty period and make
suitable arrangements for the performance of the service.
This warranty shall not apply to any defect, failure or damage caused
by improper uses or inadequate care. XIA LLC shall not be obligated to furnish
service under this warranty a) to repair damage resulting from attempts by
personnel other than XIA LLC representatives to repair or service the product;
or b) to repair damage resulting from improper use or connection to
incompatible equipment.
THIS WARRANTY IS GIVEN BY XIA LLC WITH RESPECT TO
THIS PRODUCT IN LIEU OF ANY OTHER WARRANTIES, EXPRESSED
OR IMPLIED. XIA LLC AND ITS VENDORS DISCLAIM ANY IMPLIED
WARRANTIES OF MERCHANTABILITYOR FITNESS FOR A
PARTICULAR PURPOSE. XIA’S RESPONSIBILITY TO REPAIR OR
REPLACE DEFECTIVE PRODUCTS IS THE SOLE AND EXCLUSIVE
REMEDY PROVIDED TO THE CUSTOMER FOR BREACH OF THIS
WARRANTY. XIA LLC AND ITS VENDORS WILL NOT BE LIABLE FOR
ANY INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL
DAMAGES IRRESPECTIVE OF WHETHER XIA LLC OR THE VENDOR
HAS ADVANCE NOTICE OF THE POSSIBILITY OF SUCH DAMAGES.
Contact Information:
XIA LLC
31057 Genstar Rd.
Hayward, CA 94544 USA
Telephone:
Downloads:
Hardware Support:
Software Support:
8/13/2010
(510) 401-5760
http://xia.com/DXP_Mercury_Download.html
[email protected]
[email protected]
vii
DXP Mercury / Prospect User Manual
Prospect 1.0.x
1 Introduction
The Mercury Digital X-ray Processor (DXP) is a high rate, digitallybased, multi-channel analysis spectrometer designed for energy dispersive x-ray
or γ-ray measurements and is available as a single or four-channel board, (the
Mercury-4). Amplifier and spectrometer controls including gain, filter peaking
time, and pileup inspection criteria are under computer control. Features include
a high-speed USB 2.0 interface, a customizable auxiliary bus and digital logic
controls, and support for x-ray timing and scanning applications. The Mercury
and Mercury-4 may be purchased as a high performance OEM printed circuit
board for emerging embedded applications, or as an enclosed bench-top box
with built-in power supply. In the following sections of the operating manual the
term “Mercury” refers to all variants of the Mercury and Mercury-4, unless
specifically stated otherwise.
1.1 Mercury Features
8/13/2010
•
4 MB of high-speed memory allows ample storage for timing
applications such as mapping with full spectra or multiple ROI's.
•
Peak USB 2.0 transfer rates exceed 15 MB/sec.
•
Auxiliary bus with 24 customizable digital I/O lines.
•
Peaking time range: 0.1 to 164 microseconds
•
Maximum throughput up to 1,000,000 counts/sec per channel.
•
Digitization: 14 bits at 50 MHz
•
Low noise front end offers excellent resolution, and provides
excellent performance in the soft x-ray region (110 - 1500 eV).
•
Operates with virtually any x-ray detector. Preamplifier interface is
computer controlled.
•
16 bit gain DAC and input offset are computer controlled.
•
Pileup inspection criteria are computer selectable.
•
Accurate ICR and livetime for precise deadtime correction and
count rate linearity.
•
Multi-channel analysis allows optimal use of data.
•
Facilitates automated gain setting and calibration to simplify
tuning array detectors.
•
External GATE and SYNC inputs allow data acquisition on all
channels to be synchronized.
•
Normal MCA mode allows for simultaneous full spectrum and
multiple SCA acquisition.
•
Mapping modes provide for time-resolved data acquisition, i.e. one
spectrum or set of SCA windows per pixel or scan point.
1
DXP Mercury / Prospect User Manual
Prospect 1.0.x
1.2 Data Acquisition Modes
The Mercury currently supports three data acquisition modes: static
single-spectrum 'Normal' acquisition and two time-resolved 'Mapping'
acquisition modes: Full spectrum MCA mapping, and fast SCA mapping. Note:
Normal and Mapping acquisition modes use different memory architectures and
thus require different firmware code to be downloaded.
1.2.1
MCA Mode
In Multi-Channel Analyzer (MCA) mode a data acquisition run
produces a single energy spectrum and associated run statistics. Data
acquisition runs can be started and stopped manually, or can be stopped
automatically according to a preset real time, live time or number of input or
output events.
Spectrum size ranges from 256 bins to 16384 bins. Each spectral bin is
stored as a 32-bit value, allowing for up to 4,294,967,295 events per bin per run.
Data is stored in on-board memory, and can be read by the host at any time
during or after the run. The memory is normally cleared at the beginning of a
run, but can instead be preserved, allowing for 'pause and resume' functionality.
Data acquisition can be halted system-wide according to a user provided
TTL/CMOS GATE signal, e.g. to achieve a synchronous run start.
Figure 1.1: Data flow diagram for MCA mode.
1.2.1.1 SCA Feature in MCA Mode
The Single-Channel Analyzer (SCA) feature allows for up to 32
regions of the spectrum (SCA windows) to be defined and for which output
counts are individually summed. The sums are organized into a table stored in
memory, in addition to the MCA data and statistics. The SCA table can be
accessed directly for fast readout of critical data.
1.2.2
MCA Mapping Mode
This mode supports x-ray scanning applications where multiple spectra
are generated as an x-ray beam is scanned across a sample; each spectrum
corresponds to a scan point, or pixel. This mode also supports XAFS
spectroscopy, where each spectrum corresponds to the beam energy, or
monochrometer setting.
8/13/2010
2
DXP Mercury / Prospect User Manual
Prospect 1.0.x
A data acquisition run produces multiple energy spectra, each with
associated run statistics, for each DXP processing channel. Typically a userprovided TTL/CMOS timing signal is used to advance from one spectrum to the
next during the run. Data acquisition runs can be started and stopped manually,
or can be stopped automatically according to a preset number of spectra.
Spectrum size ranges from 256 bins to 16384 bins. Each spectral bin is
stored as a 16-bit value, allowing for up to 65,535 events per bin. On-board
memory is configured as two devices, memory A and memory B, each
accessible to either the host or the on-board DSP. Continuous operation is
achieved by reading memory A while the DSP writes memory B, and vice-versa.
The data readout speed and spectrum size place a limit on the minimum pixel, or
dwell, time.
The external logic (LEMO) input can be configured to control the pixel
advance function, which creates a new spectrum corresponding to a new pixel.
Figure 1.2: Data flow diagram for mapping modes.
1.2.3
SCA Mapping Mode
The Single-Channel Analyzer (SCA) mapping mode allows for up to
64 regions of the spectrum (SCA windows) to be defined and for which output
counts are individually summed. Instead of entire spectra, only the tables of
SCA sums are stored in memory. Compressing the data in this way allows for
faster readout times, or, conversely shorter dwell times.
8/13/2010
3
DXP Mercury / Prospect User Manual
Prospect 1.0.x
1.3 System Requirements:
The digital spectroscopy system considered here consists of a remote
host computer, a DXP Mercury, and an x-ray detector/preamplifier with
appropriate power supplies.
1.3.1
Host Computer
The DXP Mercury communicates with a host computer via the USB
2.0 interface. The host computer that runs XIA’s Handel and/or ProSpect
software must have the following minimum capabilities:
 300 MHz or greater processor speed running most Microsoft
Windows Operating systems (2000, XP, Vista).

1.3.2
Preamplifier signal
specifications must be
verified.
At least one available USB 2.0 port.
Detector/Preamplifier
The DXP Mercury accommodates nearly all preamplifier signals. The
two primary capacitor-discharge topologies, pulsed-reset and resistive-feedback,
are both supported. The input voltage range of the DXP analog circuitry results
in the following constraints:
Parameter
X-ray pulse-height
(w/ input attenuator)
Input voltage range
(w/ input attenuator)
Table 1.1:
Maximum
375 mV
(1.50 V)
+/-4 V
(+/-8V)
Typical
25 mV
+/-3 V
-
Analog input signal constraints for pulsed-reset preamplifiers.
Parameter
X-ray pulse-height
(w/ input attenuator)
Input voltage range
(w/ input attenuator)
Decay time τ
Table 1.2:
Minimum
250 µV
(1 mV)
-
Minimum
250 µV
(1 mV)
100 ns
Maximum
625 mV
(2.50 V)
+/-4 V
(+/-8V)
infinity
Typical
100 mV
+/-3 V
50 µs
Analog input signal constraints for resistive-feedback
preamplifiers.
1.3.2.1 Preamp Power Supplies
If possible, we recommend using local power to generate DC voltages
for the preamplifier.
 The XPPS, manufactured by XIA, provides linear power for up to
20 NIM-standard preamplifiers.
If you decide to use your own supplies, expect to spend some time
experimenting with ground connections. A low-impedance connection between
preamplifier and detector supplies chassis’ and the DXP Mercury chassis are
almost always necessary.
8/13/2010
4
DXP Mercury / Prospect User Manual
Prospect 1.0.x
1.3.3
DXP Mercury Power Supplies
DC Voltage requirements for single-channel Mercury card:
Nominal voltage Acceptable range
Current
+ 12 V
+12V to +15V
50 mA
- 12V
-12V to -15V
50 mA
+6V
+5.5V to +6.0V
1.0 A
DC Voltage requirements for Mercury-4 card:
Nominal voltage Acceptable range
Current
+ 12 V
+12V to +15V
100 mA
- 12V
-12V to -15V
100 mA
+6V
+5.5V to +6.0V
3.0 A
1.3.3.1 AC Power
In the case of the DXP Mercury bench-top models all DC voltages
necessary for operation of the signal processor are generated internally. Use the
provided IEC certified power cable to connect the line voltage source to the rear
panel. AC voltage may be in the range 100-240VAC, 50/60Hz.
AC Line Voltage/Frequency:
115 V/60 Hz
230 V/50 Hz
Maximum Current Draw:
250 mA
250 mA
Supply voltage fluctuations are not to exceed 10% of the nominal value.
Use only a 250V/1A, 5x20mm Time-Lag fuse. The fuse is accessed by sliding
out the fuse holder below the 3-pin AC connector.
1.3.4
Cabling
1.3.4.1 Analog Inputs
The DXP Mercury uses a BNC connector to accept the preamplifier
signal.
1.3.4.2 TTL/CMOS Logic Inputs
The DXP Mercury also uses BNC connectors for timing and
synchronization logic.
1.4 Software and Firmware Overview
Two levels of software are employed to operate the DXP Mercury: a
user interface for data acquisition and control, and a driver layer that
communicates between the host software and the USB 2.0 interface. In
addition, separate firmware code is downloaded to and runs on the DXP
Mercury itself.
8/13/2010
5
DXP Mercury / Prospect User Manual
1.4.1
Prospect 1.0.x
User Interface: ProSpect
The user interface communicates with and directs the DXP Mercury via
the driver layer, and displays and analyzes data as it is received. As such XIA
provides ProSpect as a general-purpose data acquisition application. ProSpect
features full control over the DXP Mercury, intuitive data visualization,
unlimited ROI’s (regions of interest) Gaussian fitting algorithms and the
exporting of collected spectra for additional analysis. Please refer to Chapter 4
of this manual for instructions on using ProSpect with the DXP Mercury. Many
users will employ ProSpect for configuration and system optimization, but will
want to develop their own software to acquire data.
1.4.2
Device Driver: Handel
XIA provides source code and documentation for the Handel driver
layer to advanced users who wish to develop their own software interface. XIA
recommends using Handel for almost all advanced applications. Handel is a
high-level device driver that provides an interface to the DXP hardware in
spectroscopic units (eV, microseconds, etc...) while still allowing for safe,
direct-access to the DSP. ProSpect uses the Handel driver, and thus also serves
as a development example. Installation files and user manuals for Handel are
available online at
http://www.xia.com/DXP_Software.html.
1.4.3
Firmware and FDD Files
Firmware refers to the DSP (digital signal processor) and FPGA (Field
Programmable Gate Array) configuration code that is downloaded to the DXP
Mercury itself. Typically two System FPGA files (one each for normal and
mapping acquisition modes), one DSP file and up to four FiPPI (Filter-PulsePileup-Inspector FPGA) files are necessary to acquire spectra across the full
range of peaking times with a given detector/preamplifier. For simplicity XIA
provides complete firmware sets in files of the form “firmware_name.fdd”. This
file format is supported by Handel, XIA’s digital spectrometer device driver,
and is the standard firmware format used in ProSpect. Two standard firmware
files are available, one for pulsed-reset type preamplifiers and one for RCfeedback type preamplifiers. Updates to the firmware are available online at:
www.xia.com/DXP_Resources.html.
The System FGPA, DSP and FiPPI are discussed in Chapter 7.
1.4.4
Initialization File
Handel (and thus ProSpect) uses an initialization (INI) file to store all
necessary configuration information, including the path and filename of the
firmware file on the host computer, detector characteristics and spectrometer
settings, and timing and synchronization logic functions used.
8/13/2010
6
DXP Mercury / Prospect User Manual
Prospect 1.0.x
1.5 Support
A unique benefit of dealing with a small company like XIA is that the
technical support for our sophisticated instruments is often provided by the same
people who designed them. Our customers are thus able to get in-depth technical
advice on how to fully utilize our products within the context of their particular
applications. Please read through this brief chapter before contacting us.
XIA LLC
31057 Genstar Rd.
Hayward, CA 94544 USA
Telephone:
Downloads:
Hardware Support:
Software Support:
Check for firmware and
software updates at:
http://www.xia.com/DXP_
Resources.html
1.5.1
(510) 401-5760
http://xia.com/DXP_Mercury_Download.html
[email protected]
[email protected]
Software and Firmware Updates
It is important that your DXP unit is using the most recent
software/firmware combination, since most problems are actually solved at the
software level. Please check http://xia.com/DXP_Mercury_Download.html for
the most up to date standard versions of the DXP software and firmware. Please
contact XIA at [email protected] if you are running semi-custom or proprietary
firmware code. (Note: It a good practice to make backup copies of your existing
software and firmware before you update).
1.5.2
Related Documentation
As a first step in diagnosing a problem, it is helpful to consult most
recent data sheets and user manuals for a given DXP product, available in PDF
format from the XIA web site. Since these documents may have been updated
since the DXP unit has been purchased, they may contain information that may
actually help solving your particular problem. All manuals, datasheets, and
application notes, as well as software and firmware downloads can be found at
http://xia.com/ DXP_Mercury _Download.html. In order to request printed
copies, please send an e-mail to [email protected], or call the company directly.
In particular, we recommend that you download the following user manuals:
1.5.3

ProSpect User Manual – All users

Handel User Manual – Users who wish to develop their own user
interface
Technical Support
The Mercury comes with one year of e-mail and phone support.
Support can be renewed for a nominal fee. Please call XIA if your support
agreement has expired.
The XIA Digital Processors (DGF & DXP) are digitally controlled,
high performance products for X-ray and gamma-ray spectroscopy. All settings
can be changed under computer control, including gains, peaking times, pileup
inspection criteria, and ADC conversion gain. The hardware itself is very
8/13/2010
7
DXP Mercury / Prospect User Manual
Prospect 1.0.x
reliable. Most problems are not related to hardware failures, but rather to setup
procedures and to parameter settings. XIA's DXP software includes several
consistency checks to help select the best parameter values. However, due to the
large number of possible combinations, the user may occasionally request
parameter values which conflict among themselves. This can cause the DXP unit
to report data which apparently make no sense (such as bad peak resolution or
even empty spectra). Each time a problem is reported to us, we diagnose it and
include necessary modifications in the new versions of our DXP control
programs, as well as adding the problem description to the FAQ list on our web
site.
1.5.3.1 Submitting a problem report:
XIA encourages customers to report any problems encountered using
any of our software via email. In most cases, the XIA engineering team will
need to review bug information and run tests on local hardware before being
able to respond.
All software-related bug reports should be e-mailed to
[email protected] and should contain the following information, which
will be used by our technical support personnel to diagnose and solve the
problem:
 Your name and organization

Brief description of the application (type of detector, relevant
experimental conditions...etc.)

XIA hardware name and serial number

Version of the library (if applicable)

OS

Description of the problem; steps taken to re-create the bug

Full Error Report (see section 4.7.4.1) plus additional data:
o
Saved MCA data, if relevant (see section 4.7.4.2)
o
Saved Baseline data, if relevant (see section 4.7.4.3)
o Saved Trace data, if relevant (see section 4.7.4.4)
Please compress the Error Report into a ZIP archive and attach the support
request email.
1.5.4
Feedback
XIA strives to keep up with the needs of our users. Please send us your
feedback regarding the functionality and usability of the Mercury and ProSpect
software. We are also interested in hearing about improvements to the hardware
and software. In particular, we are considering the following development
issues:
8/13/2010
8
DXP Mercury / Prospect User Manual
Prospect 1.0.x
1.5.4.1 Export File Formats
We would like to directly support as many spectrum file formats as
possible. If we do not yet support it, please send your specification to
[email protected]
1.5.4.2 Calibration
Currently the hardware gain of the Mercury is modified during energy
calibration to produce a spectrum with a user defined bin scale, i.e. an integer
electron-volts-per-bin value. The drawback is that the calibration process often
takes several iterations. Another approach to calibration is re-interpreting the
bins. This is not difficult to do, but may produce confusion for the novice user.
We are considering supporting this feature in future ProSpect releases.
8/13/2010
9
DXP Mercury / Prospect User Manual
Prospect 1.0.x
2 Installation
CAUTION: Improper
connections or settings can
result in damage to system
components. Such damage
is not covered under the
DXP Mercury warranty.
Please carefully follow these instructions. It is important that you
follow the steps in order: Install ProSpect and drivers, connect Mercury
hardware, run ProSpect software.
2.1 Software Installation
Do not attempt to install the Mercury hardware until after the software
and drivers have been installed. ProSpect operates on Windows XP, 2000 and
Vista machines. Updates to ProSpect are available online at:
www.xia.com/DXP_Mercury_Download.html
The update installation file is a executable, or .EXE file.
2.1.1
Running the Installer
1) Please close all applications that are currently running.
2) Insert the CD into the CD-ROM drive or, if your copy was
delivered electronically, double-click the setup.exe program. If
the CD installation does not start immediately, follow the
instructions in steps (3) and (4).
3) Click the Start button and select the Run command.
4) Type X:\Setup.exe and click [OK], where X is the letter of
your CD-ROM drive.
5) After setup has completed, shut down your computer and complete
the hardware configuration described in sections 2.2 through 2.3
before restarting.
2.1.2
•
The ProSpect 0.1.x installation will create a new directory:
"C:\Program Files\xia\ProSpect 0.1".
•
A new Start Menu > Program group will be created.
•
A shortcut to the ProSpect executable is created on your desktop.
•
Necessary drivers will be installed
File Locations
The ProSpect default installation folder is:
C:\Program Files\XIA\ProSpect 0.1
This directory contains program files, libraries, log files and configuration, or
INI, files. The “firmware” folder is:
~\ProSpect 0.1\firmware
8/13/2010
10
DXP Mercury / Prospect User Manual
Prospect 1.0.x
This directory contains the normal and mapping firmware, or FDD,
files (see section 1.4.3). Updates to the firmware are available online at:
www.xia.com/DXP_Mercury_Download.html
2.1.3
Support
For the latest documentation, please refer to XIA’s website at
www.xia.com/DXP_Mercury_Download.html
XIA values all of the feedback it receives from customers. This feedback is an
important component of the development cycle and XIA looks to use this
feedback to improve the software. All bug fixes and feature suggestions should
be directed to [email protected] Please be sure to include as much
information as possible when submitting a bug report. For further instructions
please refer to section 1.5.
2.2 Configuring the Analog Signal Conditioner
The term ‘jumper’ is used in this section. Jumpers are placed on 3-pin
headers, connecting the center pin to one or the other peripheral pin, similar to a
single-pole-double-throw (SPDT) switch.
2.2.1
Input Attenuation: JP100
Attenuation may be necessary if the preamplifier gain or output voltage
range is excessive and/or high-energy x-rays are to be processed. Pulses up to
several hundred milliVolts in size and a voltage range of +/- 4 Volts can be
accommodated without attenuation. The default position for jumper JP100,
labeled ‘0dB’ (see Figure 2.1, single channel Mercury), passes the signal
directly. If larger signals must be processed, set JP100 to the ‘-6dB’ position to
reduce the input signal by a factor of two. The equivalent jumpers are also
provided on the Mercury-4 board.
8/13/2010
11
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 2.1: The DXP Mercury printed circuit board. The input attenuation jumper JP100 is highlighted in
blue.
2.3 Making Connections
It is possible to damage the DXP Mercury and/or connected equipment
if the instructions below are not followed. All electronic connections are made
at the front panel of the Mercury. We recommend using cables under three
meters in length for signal connections to the preamplifier.
2.3.1
Signal Connections
The DXP Mercury uses BNC connectors for convenience, reliability
and signal quality. Fasten a BNC cable to the signal input and connect the other
end to the detector/preamplifier output.
8/13/2010
12
DXP Mercury / Prospect User Manual
Prospect 1.0.x
2.3.2
GATE/SYNC Connection
Each Mercury includes BNC GATE or SYNC timing inputs for
synchronization and time-resolved spectroscopy applications.
For now, do not make GATE or SYNC connections. The
Configuration Wizard utility will help you make decisions about which
function(s) to use, and how to make the proper connections. See sections 3.2
and 3) for more details.
2.4 Starting the System
Make sure of the following before proceeding:

Your system satisfies the requirements outlined in section 1.3
above.

The ProSpect software and drivers have been installed.

DXP Mercury module has been installed.

Detector and preamplifier are connected and powered.

A low-to-moderate intensity x-ray source is available for
calibration and system verification.
Turn on the DXP Mercury...
2.4.1
DXP Mercury Driver Selection
Windows should automatically find the new hardware and start the
Found New Hardware Wizard. The driver file ‘xia_usb2.inf’ is located in the
‘drivers’ sub-folder within the Prospect install folder.
1) The first screen asks whether Windows can connect to Windows
Update to search for the driver. Select “No, not this time” and press
[Next] to proceed to the "Install Hardware Device Drivers" page.
2) Select "Install from a list or specific location (Advanced)" option and
press [Next] to proceed to the "Locate Driver Files" page.
3) Select "Search for the best driver in these locations", check “Include
this location in the search:”, and enter “C:\Program Files\XIA\ProSpect
0.1\drivers”. Then press [Next].
4) Windows should find the suitable driver. Press [Next] to complete the
driver installation..
Note: Driver selection can be changed at any time via the Windows
Device Manager. To open the Device Manager, right-click on the "My
Computer" icon and select "Manage". Now click on "Device Manager" in the
left-pane of the Computer Management window. Mercury cards can be found
under "Other Devices".
8/13/2010
13
DXP Mercury / Prospect User Manual
Prospect 1.0.x
3 System Configuration
At this point the ProSpect software and drivers should have been
installed, and the Mercury hardware should be powered on and identified by
Windows. This chapter will guide you in using the ProSpect Configuration
Wizard utility.
3.1 Initialization Files
After power up the Mercury's DSP and programmable logic are in an
unknown state. Program code, or firmware, for these devices must first be
downloaded via the USB before data can be acquired. After the devices are
operational, user settings are downloaded.
Handel (and thus ProSpect) uses an initialization (INI) file to store all
necessary configuration information, including the path and filename of the
firmware file on the host computer, detector characteristics and spectrometer
settings, and timing and synchronization logic functions used. In order to start
properly, ProSpect needs to have the following information:

The location of the Mercury FDD firmware file (DSP and FPGA
code that runs on the board, included in the installation package).

Various properties of the detector preamplifier including type,
polarity and gain.

Which timing and synchronization functions are to be used.
Master and slave modules will be designated automatically.
Note: section 3) describes timing and synchronization logic.
INI files can be updated at any time, i.e. after the spectrometer settings
have been optimized, and existing INI files can be loaded at any time. If you
have previously run with ProSpect, your registry settings will point to the most
recently used INI file, and ProSpect will automatically run with these settings
upon startup.
3.1.1
Starting ProSpect Without an INI File
Start ProSpect via the Start menu: Start > Programs > ProSpect 0.1 >
ProSpect. The first time ProSpect starts up, the ProSpect Configuration File
Error panel will appear, because a valid configuration file has not been selected.
Press the "Generate New File" button to launch the Configuration Wizard,
which guides the user step-by-step to create an INI file.
8/13/2010
14
DXP Mercury / Prospect User Manual
Prospect 1.0.x
3.2 The Configuration Wizard
The Configuration Wizard utility can be launched at any time from the
"Tools" menu in ProSpect. First select the appropriate instrument (“mercury” or
“mercury4” from the drop-down list and press [OK].
3.2.1
General Settings
These basic settings are the bare minimum necessary to run the
Mercury in normal single-spectrum mode.
1) Welcome to the Configuration Wizard
The first panel of the Configuration Wizard is simply a
welcome screen with some information about the utility.
Press [Next] in the Mercury Configuration Panel.
2) Firmware
The firmware file contains all program code for the programmable
devices on the Mercury. Press the [FDD File…] button to browse,
or type "C:\Program Files\xia\ProSpect
0.1\firmware\mercury\mercury_reset.fdd" and press [Next]. If you
have updated your firmware since ProSpect was installed, be sure
to select the new file. Note that different firmware files are
required for pulsed-reset and RC-feedback type preamplifiers.
Updates to the firmware are available online at:
http://www.xia.com/DXP_Mercury_Download.html
Figure 3.1 The firmware file contains program code for the Mercury's
programmable devices.
3) Detector Configuration
Select the appropriate detector type. For Reset type, enter the
8/13/2010
15
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Reset Interval. This is the time in microseconds that the
preamplifier takes to reset and settle, and should be set
conservatively to prevent associated voltage transients from
entering the spectrum. If you don't know the reset time enter 10
(microseconds). For RC Feedback enter the RC Decay Time in
microseconds. Press [Next].
Figure 3.2: The Detector Configuration settings.
4) Hardware Configuration
This panel displays all located Mercury modules, At this point it is
possible to add or disable modules. Clicking the Rescan button
will detect connected devices if changes have been made.
8/13/2010
16
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 3.3: The Hardware Configuration panel for a two -module Mercury4 system .
5) Detector Element Property
Each Mercury processing channel includes a programmable gain
analog stage to compensate for the detector gain. Initially the same
polarity and gain should be used for all channels (Mercury-4).
During the calibration process the gain can be fine-tuned for each
channel and the INI file updated.
Click in the Polarity column: "-" if X-ray steps generate a negative
voltage step; "+" if X-ray steps generate a positive voltage step. If
you don't know the polarity, keep the default (negative) setting.
Enter the gain in [mV/keV]. If you don't know the gain, keep the
default of (3 mV/keV). Press [Next].
8/13/2010
17
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 3.4: Detector Element Property panel for an 8-channel system
3.2.2
Hardware Synchronization Settings
If you intend only to use the Mercury in single-spectrum mode without
the synchronization features, press [Finish] to skip ahead to save the generated
configuration file.
6) Hardware Timing Synchronization
If you want to use the synchronization features and/or use the
Mercury in mapping mode, select Configure hardware for
synchronized run, press [Next]. Otherwise press [Finish] to skip
ahead to save the generated configuration file. (Proceed to 0.
7) GATE Function
The GATE function is used to selectively halt data acquisition
during a run according to a user-provided TTL/CMOS logic signal.
Please review section 7.2.1 for a complete description of this
feature.
The Enable GATE setting reserves the left-most available module
in each PCI bus segment as the GATE master, i.e. it accepts the
front-panel GATE connection. The "GATE Polarity" setting
determines whether data is halted when the GATE logic signal is
LO or HI. Make your selections and press [Next].
8/13/2010
18
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 3.5: For this system GATE is enabled, set to halt acquisition when LO.
3.2.3
Mapping Mode Settings
The remaining settings relate to the mapping mode wherein multiple
spectra are acquired for each processing channel in a single run, e.g. to produce
an elemental map of the sample in x-ray scanning applications. Please review
section 1.2.1.1 for a description of the mapping modes.
8) Mapping Configuration
If you want to use the Mercury in mapping mode, select "Continue
with mapping configuration" and press [Next]. Otherwise select
"Don't use mapping mode" and press [Next].
9) Pixel Advance Mode
The Pixel Advance triggers the change to a new spectrum in
multiple-spectrum data acquisition. Typically the Pixel Advance is
controlled by a user-provided logic signal. In GATE mode each
leading edge transition generates a Pixel Advance instruction. See
section 7.2.2 for a description of this mode. In SYNC mode a
Pixel Advance instruction is generated every N LO-to-HI
transitions. See section 7.2.3 for a description of this mode. In
User mode the Pixel Advance is triggered by a command from the
host computer.
10) The next panel depends on the Pixel Advance Mode selection:
a)
8/13/2010
GATE Pixel Advance Options
As described in section 7.2.2.3, in GATE Pixel Advance mode the
GATE signal by default also halts data acquisition. If "Pixel
advance only" is selected on this panel, data will be written to the
new spectrum immediately after each leading edge transition
regardless of the pulse-width.
Note: The polarity selection made in step 7) above is used for both
the normal and mapping modes, e.g. if "LO = halt acquisition" was
selected, the pixel advance occurs on the HI-to-LO transition.
19
DXP Mercury / Prospect User Manual
Prospect 1.0.x
b) SYNC Pixel Advance Options
The SYNC pixel advance occurs after the selected "Number of
cycles" is detected, on the desired "Trigger Edge".
Note: Selecting SYNC mode reserves the left-most available
module in each PCI bus segment as the SYNC master, i.e. it
accepts the front-panel SYNC connection.
c)
User Pixel Advance Options
There are no options for the this mode; the utility skips to next
panel.
11) Timing Mode Run Control Options
The Mercury can automatically stop the data acquisition run after a
prescribed number of pixels. The "Number of Pixels Per Run"
setting can easily be modified later in ProSpect. If it is set to
minus one (-1) the run continues until the user stops the run.
A dual memory architecture is used to achieve continuous
operation in mapping mode. Each memory device is 1,048,576
words in size. The "Number of Pixels Per Readout" is slightly less
than the total device size divided by the individual spectrum size.
If zero, or a number greater than acceptable, is selected the largest
number that can be used is automatically calculated. This setting
can easily be modified later in ProSpect.
3.2.4
Completing the Configuration
Save Completed Configuration
The INI file you have created can now be saved. Select a unique name
for the file, e.g. "C:\Program Files\xia\ProSpect
0.1\mercury_myconfig.ini". Press [Finish] to save the INI file and exit
the Configuration Wizard. Note: In Windows Vista and above, the
INI file should be saved in a folder outside of “Program Files” to avoid
saving and retrieval problems. Note that if you did not start the system
above, you must load the INI file to enact your changes.
3.3 Loading and Saving Initialization Files
INI files can be updated at any time, i.e. after the spectrometer settings
have been optimized, and existing INI files can be loaded at any time. If you
have previously run with ProSpect, your registry settings will point to the most
recently used INI file, and ProSpect will automatically run with these settings
upon startup.
3.3.1
Loading an INI file
Select "Load Configuration" from the File menu. Browse to and select
an INI file that you just created and press "Open". ProSpect will download
firmware and initialize the Mercury modules in your system.
3.3.2
Saving an INI file
INI files can be updated at any time, i.e. after the spectrometer settings
have been optimized, by selecting "Save Configuration", or "Save Configuration
As…" from the "File menu. You may find it useful to maintain several INI files,
e.g. for operating with different detectors, or with different spectrometer
8/13/2010
20
DXP Mercury / Prospect User Manual
Prospect 1.0.x
settings.
Note: In Windows Vista and above, the INI file should be saved in a folder
outside of “Program Files” to avoid saving and retrieval problems
8/13/2010
21
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4 Using ProSpect with the Mercury
At this point the ProSpect software and drivers should be installed, the
Mercury hardware should be powered on and identified by Windows, and a
valid initialization file should have been created and loaded. This chapter will
guide you in using ProSpect with the Mercury module.
4.1 A Quick Tour of ProSpect
ProSpect is a PC-based application that provides for the setup,
optimzation and failure diagnosis of the instrument, and allows for the reading
out, displaying, analyzing and exporting of acquired energy spectra. When you
start the program and an initialization file has been loaded, the ProSpect main
window should be displayed as in Figure 4.1.
Figure 4.1: The ProSpect main window upon startup, after hardware initialization.
4.1.1
Channel Selection
This panel is not displayed for the single-channel Mercury. Each
Mercury-4 module provides four (4) digital x-ray processing channels. The
8/13/2010
22
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Channel Selection global control sets the channel for which settings and data
are displayed in ProSpect. The "Apply to All" checkbox applies to settings
only: If the checkbox is checked, any change to settings will be applied to all
channels simultaneously.
4.1.2
Settings Sidebar
The tabbed Settings sidebar provides easy access to all hardware and
firmware settings. It is intended to be the primary interface for setup and
optimization. The Acquisition tab contains spectrometer settings such as
peaking time and thresholds. The Detector tab contains detector and
preamplifier settings such as polarity and gain. The SCA tab displays Single
Channel Analyzer data, if applicable, based on user-entered regions of interest.
4.1.3
Main Window
The tabbed Main Window contains the MCA, Baseline, oscilloscope,
and system calibration. The MCA tab is used for normal mode spectrum
acquisition. The Baseline tab displays the baseline histogram. The ADC tab
contains the oscilloscope tool for displaying ADC, filter output, and baseline
data. The GainMatch tab contains the system-wide gain calibration tool. The
Mapping panel is used for time-resolved multi-spectrum and multi-SCA data
acquisition.
8/13/2010
23
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4.2 Detector and Preamplifier Settings
If the Configuration Wizard was followed correctly as described in
section 3.2, the system should be nearly ready for data acquisition. Before
taking a spectrum, however, we recommend verifying the Detector and
Preamplifier settings. After settings have been optimized, the configuration
(INI) file should be saved.
Figure 4.2: An ADC trace displayed in the ADC panel oscilloscope tool.
Notice that the displayed x-ray events (11 total) are voltage steps
with rising edges, thus the polarity is set correctly.
Select the ADC tab in the Main Window to display the oscilloscope
tool (see Figure 4.2). Select ADC from the drop-down list, set the Sampling
Interval to "1.000" µs and press the Get Trace button to display a 4096-point
raw ADC data set.
Select the Detector tab of the Settings panel. The Polarity setting
enables or disables a digital inverter depending on the signal polarity of the
preamplifier. The Reset Interval is the settling time, in microseconds, of the
preamplifier reset. The Preamp Gain is the gain, in milli-Volts per kiloelectron-Volt of the charge sensitive preamplifier. The Apply button downloads
the adjusted setting(s) to the Mercury hardware. For a thorough discussion of
oscilloscope diagnostic tool, please review section 4.7.1.
Note: Do not confuse
detector bias polarity with
the polarity of the
preamplifier signal; they
are not necessarily related.
8/13/2010
4.2.1
Pre-Amplifier Polarity
Preamplifier polarity denotes the polarity of the raw preamplifier
signal, NOT the detector bias voltage polarity. A positive polarity preamplifier
produces a positive step, defined as a voltage step with a rising edge, response to
an incident x-ray. The digital filters in the Mercury expect an input signal with
positive steps. An optional input inverter is employed to correct the signal
24
DXP Mercury / Prospect User Manual
Prospect 1.0.x
polarity for negative polarity preamplifiers. If the polarity has been set
correctly, the ADC oscilloscope trace should display positive steps.
If the ADC trace displays positive steps (as in Figure 4.2), the polarity
has been set correctly. If not, change the Polarity setting and press the Apply
button. Acquire a new trace to verify that the polarity setting is correct.
Please read through section 4.7.1 for a thorough description and figures
relating to the preamplifier signal polarity.
4.2.2
Reset Interval
The Reset Interval is the period of time after each preamplifier reset
that the Mercury waits before re-enabling data acquisition. The delay is set
based on the settling time of the preamplifier reset transient waveform, typically
ranging from hundreds of nanoseconds to hundreds of microseconds. If you are
unsure, enter "10" µs. Setting the delay shorter than the transient settling time
may introduce ‘reset artifact’ events into the spectrum. Setting the delay longer
than necessary introduces additional processor dead time, which will reduce the
data throughput at high count rates.
4.2.3
Preamp Gain
The Preamp Gain setting, in combination with the dynamic range
setting, controls the Mercury’s variable gain amplifier such that the input
requirements of the ADC are satisfied, given the gain of the preamplifier. If you
know the gain of your preamplifier, enter that value. Otherwise we recommend
using the default value of 3mV/keV. This setting is normally adjusted
automatically during energy calibration. In cases of extremely low or high
preamplifier gain, it may be necessary to adjust the nominal gain before taking a
spectrum. If the displayed x-ray steps are less than 50 ADC units in height,
reduce the Preamp Gain setting. If the displayed x-ray steps are greater than
2,000 ADC units in height, increase the Preamp Gain setting.
4.2.4
Preamp Risetime
This is an advanced setting, accessible by pressing the [Edit Filter
Parameters] in the Acquisition settings tab. The preamplifier rise-time should
be measured and the Minimum Gap Time set accordingly. This setting is
described in detail in section 4.6.1.2. See section 4.7.1.2 for details on using
ProSpect to measure the rise-time for your system and section 6.3 for a
theoretical discussion of the issues involved in trapezoidal filtering.
4.2.5
Saving the Configuration File
This is a good time to save your configuration file. From the File
menu, select “Save Configuration” to update the currently-used INI file. Or,
select “Save Configuration As…” to create a new INI file.
8/13/2010
25
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4.3 Normal Spectrum Mode Data Acquisition
4.3.1
To begin data collection:
Select the MCA tab and
press the [Start Run]
button.
Starting a Run
Once the detector/preamplifier settings have been verified you are
ready to collect a sample spectrum. Place a known X-ray source, for example an
55Fe source that produces Mn K line at 5895 eV, such that x-rays strike the
α
detector’s active area at a moderate to low rate, i.e. less than 10,000 x-rays
absorbed per second.
Select the MCA tab and press the [Start Run] button in the data
display panel to begin data collection. An uncalibrated energy spectrum should
appear. Figure 4.3 shows a sample uncalibrated 55Fe spectrum. Proceed to
section 4.3.2 if a spectrum is displayed.
No spectrum?
Check your hardware
setup, e.g. x-rays present?
 Check your
initialization settings,
e.g. preamplifier type
and polarity correct?

Troubleshoot the
signal using the
Oscilloscope tool, as
described in section
4.7.1.
Figure 4.3: An uncalibrated 55Fe spectrum.
Press the [Update] button to manually read out the MCA data, or check
the Continuous checkbox to automatically refresh the spectrum. A horizontal
line at zero on the y-axis indicates that no output events have been acquired,
although the run is active. This can result from a hardware setup problem, e.g.
x-rays not hitting detector; detector not powered, etc. Or it can result from
incorrect configuration settings. The most common problem is incorrect
detector/preamplifier settings. To troubleshoot these settings please refer to the
Diagnostics section 4.7.
8/13/2010
26
DXP Mercury / Prospect User Manual
4.3.2
Prospect 1.0.x
Skipping Channels
The manual or automatic MCA data readout operates on all active
processing channels, i.e. though only one spectrum is displayed, data from all
channels is locally accessible via the Channel Selection control.
Processing channels can be disabled, or skipped, from the readout
operation. Select the System settings tab to display the Channel Selection
Detail. Click in the Skip Session column to de-select individual channels.
Note: Channel skipping also applies to system-wide gain matching as described
in section 4.3.5.2.
Figure 4.4: The Channel Selection Detail control in the System settings tab.
In the system shown, MCA data readout and system-wide gain
matching would be skipped for channels 10 and 11.
To change common
acquisition settings select
the Acquisition tab of the
settings panel.
4.3.3
Spectrometer Settings
The primary spectrometer settings are immediately accessible via the
Acquisition tab in the settings panel.
4.3.3.1 Peaking Time (Energy Filter)
Figure 4.5: The Peaking Time is displayed in the Acquisition tab of the
Settings panel.
8/13/2010
27
DXP Mercury / Prospect User Manual
Note: The energy filter
peaking time is widely
referred to as “peaking
time”, whereas the fast
filter peaking time is
referred to as “fast
peaking time”
Making a plot of energy
resolution versus peaking
time provides a useful
future reference.
Prospect 1.0.x
The energy filter peaking time is one of the primary user controls.
Generally speaking, a longer peaking time produces better energy resolution at
the cost of increased dead time and thus lower output count rate. In practice, the
user may set the peaking time to a shorter than optimal value in order to increase
data throughput, making up for degraded energy resolution with improved
statistics. Most detectors also have an upper limit above which the energy
resolution gets worse. HPGe detectors typically have optimal peaking times
between 16µs and 32µs. Silicon drift detectors often produce the best resolution
at 10µs or less. Some SiLi detectors show resolution improvements out to 80µs
or longer.
The [Edit Filter Parameters] button accesses additional filter
parameters, including the energy gap time, the fast, or trigger, filter settings and
pileup rejection parameters. The default filter settings reflect a compromise
between robustness and performance and typically do not need to be changed.
In some cases energy resolution for a given peaking time can be improved
significantly if these settings are optimized as described in section 0.
4.3.3.2 Trigger Threshold
The trigger, or fast, filter threshold sets the low-energy limit for the fast
filter, which is used primarily for pileup inspection. If the baseline threshold is
employed, the detection of x-rays actually extends to energies significantly
below the trigger threshold (see section 4.3.3.3). For this reason it is not
necessary to set the trigger threshold aggressively, i.e. setting the threshold as
low as possible will derive little benefit. If set too low, the trigger threshold will
introduce a zero energy noise peak into the spectrum. In extreme cases it will
halt data throughput entirely.
To optimize the fast filter threshold, set the Baseline Threshold to zero
(so that output events are generated by fast filter triggers only), edit the Trigger
Threshold value and press [Apply]. Typical values range from 600eV to
1500eV. A good procedure is to initially set the value too high, reduce it until
the zero energy noise peak starts to become significant, and then raise it again
until the noise peak is eliminated.
The fast filter length is independent of the energy filter length, or
peaking time, thus the trigger threshold does NOT need to be optimized every
time the peaking time is changed. All thresholds must be readjusted if the gain
changes significantly.
4.3.3.3 Baseline Threshold
Note: The baseline threshold is not available for decimation 0, i.e.
peaking times less than or equal to 2.0 µs.
The baseline threshold sets the low-energy limit for the intermediate, or
baseline, filter, which is used for both baseline acquisition and low-energy x-ray
detection. To optimize the baseline filter threshold, first optimize the trigger
threshold as described above, then edit the Baseline Threshold value and press
[Apply]. Typical values range from 150 eV to 1000 eV.
The baseline filter length is linked to the energy filter length, or
peaking time, thus the baseline threshold should be optimized every time the
peaking time is changed. All thresholds must be readjusted if the gain changes
significantly.
8/13/2010
28
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Please review section 6.4 for a detailed description of baseline
acquisition and averaging. Section 4.7.2.1 describes the empirical optimization
of the baseline threshold
4.3.3.4 Energy Threshold
CAUTION: In almost all
cases the Energy
Threshold should be set to
zero. An error term in the
counting statistics is
introduced when the
Energy Threshold is
enabled. For this reason it
should only be enabled at
low data rates.
The energy threshold sets the low-energy limit for the slow, or energy,
filter, which is used primarily for measuring the pulse-height, i.e. energy, of xray voltage steps. Triggering on the energy (slowest) filter can extend the
detection range down to the lowest energies for a given detector, however, in
most cases we recommend setting the Energy Threshold to zero. This is
because the dead time associated with x-rays detected by the energy filter can
not be directly measured. It remains available primarily for two special cases:
A non-zero energy threshold is appropriate for ultra-soft x-ray detection at
very low input count rates.
A non-zero energy threshold may be used to extend the detection range for
decimation 0, i.e. peaking times under 2 µs. Dead time and count rate
statistics will however be distorted.
4.3.3.5 Dynamic Range
The dynamic range setting combines with the detector gain setting to
determine the variable analog gain of the Mercury. The variable gain is set such
that an x-ray with energy equal to the dynamic range value produces a voltage
step of the maximum allowable amplitude at the ADC input. X-rays with
energies exceeding the dynamic range value cannot be processed correctly. The
presence of such x-rays can result in a significant reduction in the output count
rate. The Dynamic Range setting should be set above the largest x-ray energy
present in the system. Typical values range from 40keV to 100keV. Edit the
Dynamic Range value and press [Apply].
4.3.3.6 MCA Number of Bins and MCA Bin Width
The size and granularity of the spectrum can be adjusted. The number
of spectrum bins sets the granularity of the acquired spectrum. The eV/Bin
setting determines the size of each MCA bin in electron Volts. Together, these
settings determine the energy span of the MCA: The spectrum ranges from zero
to a maximum energy equal to the number of spectrum bins multiplied by the
MCA bin width (e.g. a 40.96keV spectrum results from 2048 bins at 20eV/bin).
Note that these digital spectrum controls are independent of the
Preamp Gain and Dynamic Range settings that control the variable analog
gain. If the MCA energy range is less than the dynamic range, the entire
spectrum will be free of distortion. If the MCA energy range exceeds the
dynamic range setting, the spectrum will be distorted: Higher energy x-ray data
will be attenuated or cut off. For this reason the product of Number of Bins and
MCA Bin Width, i.e. the MCA energy range, should be less than the dynamic
range. Edit the Number of Bins and MCA Bin Width values and press
[Apply]. Start a new run.
8/13/2010
29
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4.3.3.7 Baseline Average
The baseline is the output of the energy filter in the absence of x-rays.
A running average of baseline samples, acquired between x-ray events, is
subtracted from the x-ray peak samples to arrive at the true energy of incident xrays. A perfect detector and preamplifier would produce a constant baseline,
however, in the real world the actual baseline varies. The number of Baseline
Average Samples can strongly affect performance. More samples improve
noise reduction but slow the reaction time to actual changes in the baseline. In
most cases a value between 64 and 512 will produce the best results.
Please review section 6.4 for a detailed description of baseline
acquisition and averaging. Section 4.7.1.4 describes the empirical optimization
of the number of samples in the baseline average.
4.3.4
Setting Regions of Interest (ROIs)
A region of interest (ROI) is a user-defined energy range of the MCA
spectrum for which separate statistics are displayed in the ROI Table. Typically
an ROI corresponds to an energy peak. ROIs are used for energy calibration and
SCA acquisition.
4.3.4.1 Adding ROIs
The Region Of Interest table is located below the spectrum. A single
ROI is displayed by default. If you cannot see the ROI table, slide the panel
separator up or press increase the size of the entire Prospect window. Click on
Add New Row to manually add an ROI. The three leftmost columns in the
table control the display of the ROIs. The first column indicates which ROI is
active, and its color. Only one ROI is active at a time; click on a row to make it
active. The second column locks the ROI. The third column toggles the fill
mode. The Lower and Upper bounds of the ROI can be entered manually in the
table, or automatically created for a given peak using the Auto ROI function.
4.3.4.2 Auto ROI
The Auto ROI function generates lower and upper bounds for the
active ROI about a selected energy peak. Place the mouse pointer over the
spectrum peak of interest. Right-click and select Place Cursor. Drag the cursor
to the center of the calibration peak, right-click and select Auto ROI. A region
of interest should automatically appear on the peak. In some cases, where few
events have been collected, the Auto ROI feature will not properly enclose the
peak. In these cases, the ROI can be adjusted directly in the Spectrum Window.
8/13/2010
30
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 4.6: The Auto ROI function (found in the cursor context menu)
automatically defines a region of interest around the peak selected
with the active cursor. The cursor context menu is displayed by
right-clicking on a cursor.
4.3.5
Gain Calibration
This process modifies an analog variable-gain stage on the Mercury
hardware, and correspondingly modifies the Preamp Gain software setting.
Due to the analog nature of the variable gain amplifier that is used, the precise
analog gain following a hardware gain modification is unknown until it is
measured. For this reason, calibration is an iterative process that must be
executed any time acquisition values are changed that require a hardware gain
modification, e.g. if the Dynamic Range is increased.
Section 4.3.5.1 describes how to directly calibrate a single processing
channel using the active ROI. This approach is appropriate for the singlechannel Mercury, but should also be used during initial setup of the Mercury-4,
to propagate a nominal calibration to all channels. Section 4.3.5.2 describes
how to use the GainMatch tool to calibrate multiple processing channels
simultaneously.
Once calibration is complete, the modified configuration settings can
be saved to the configuration file, each channel with a unique Preamplifier
Gain, so that calibration is maintained the next time ProSpect is started.
4.3.5.1 ROI-Based Gain Calibration
At this point you should have an energy peak bounded by a region of
interest. Please review section 4.3.4 if you have not created an ROI. The ROI
table displays the mean energy and width of the peak in the ROI, as well as the
ROI upper and lower limits.
Select the Detector settings tab and make note of the Preamp Gain
value. To calibrate: First make sure the ROI containing the selected peak is
active, as indicated by a green checkmark in the leftmost column, then enter the
peak’s known energy into the Calib. (keV) field of the ROI (i.e. for an 55Fe
Kα line enter 5.895 keV). Then press the [Calibrate ROI] button, and start a
new run. The spectrum should now appear with the peak properly calibrated.
For the best accuracy it is often necessary to run the calibration through a few
8/13/2010
31
DXP Mercury / Prospect User Manual
Prospect 1.0.x
iterations. If the initial spectrum was badly out of calibration, the resulting
change in gain may cause the peak to jump partially or fully out of its ROI. In
this case, readjust the ROI so that it centers on the peak before repeating the
calibration.
When you are satisfied with the calibration, note that the Preamp Gain
value has been modified. For the Mercury-4, before using the GainMatch tool
it is good practice to propagate this nominal value to all the processing channels.
Make sure the Apply to All checkbox is checked. Cut-and-paste, or simply retype the Preamp Gain value and press [Apply]. Note that all channels have
now been set to the new value.
Figure 4.7: A calibrated 55Fe spectrum.
4.3.5.2 Multi-Channel Gain Calibration
Multi-element detectors often have significant channel-to-channel gain
variations. Further, the nominal gain specification provided by the manufacturer
can be off by 20% or more. Given the tedium of calibrating one channel at a
time as described above, Prospect includes a tool to calibrate multiple channels
simultaneously. The GainMatch tool automates the gain calibration for all
specified processing channels, according to user constraints. The iterative
routine acquires data for the user-specified Acquisition Time, looks for a peak
within the Calibration Peak Range, compares the measured peak centroid to
the Calibration Energy and adjusts the gain as necessary. The process repeats
until to the Number of Interations has expired, or the % of Calibration
Energy has been reached.
4.3.5.3 Skipping Channels
The GainMatch tool operates on all active processing channels.
Individual processing channels can be disabled, or skipped, from the calibration
8/13/2010
32
DXP Mercury / Prospect User Manual
Prospect 1.0.x
macro. Select the System settings tab to display the Channel Selection Detail.
Click in the Skip Session column to de-select individual channels.
Note: Channel skipping also applies to MCA data readout as described in
section 4.3.2.
Figure 4.8: The Channel Selection Detail control in the System settings tab.
In the system shown, MCA data readout and system-wide gain
matching would be skipped for channels 10 and 11.
4.3.5.4 Running the Calibration Macro
Select the GainMatch tab in the main window. This tool only works if
the nominal gain has been set such that the energy peak falls within the
Calibration Peak Range. The best practice is to calibrate a single channel,
then propagate the nominal gain to all channels as described in section 4.3.5.1
above.
Depending on source intensity the Acquisition Time should typically
range between 1 and 10 seconds. Setting the % of Calibration Energy less
than 0.1 may result in failure to converge. It is best to experiment with the
settings to get a feel for the utility.
Note that if the calibration routine fails to find a peak in a given
channel, that channel will automatically be disabled. You may want to re-enable
channels afterwards in the System settings tab.
8/13/2010
33
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 4.9: The GainMatch panel after one iteration. Note the significant
channel-to-channel gain variation.
Figure 4.10: The GainMatch panel after five iterations. Note the near perfect
channel-to-channel calibration.
Note that if energy calibration results in a significant change in gain, it
may be necessary to adjust thresholds.
 Save your (modified) INI
file to a unique filename:
Select Save Configuration
As… from the File menu.
8/13/2010
4.3.6
Saving and Loading INI Files
Completion of the gain calibration is the final step in the verification of
basic settings. The settings should now be saved to an INI file such that they
will automatically reload whenever ProSpect is started. Because calibration is
required any time spectrometer settings are changed, we recommend creating
separate INI files for each commonly used peaking time. Once the settings for a
34
DXP Mercury / Prospect User Manual
Prospect 1.0.x
peaking time have been optimized and the system calibrated, the entire setup can
be restored by loading the INI file.
4.3.6.1 Saving an INI File
Select File » Save Configuration As… to open the Save
Configuration File dialog. Enter a unique filename and press the [Save]
button.
4.3.6.2 Loading an INI File
Select File » Load Configuration… to open the Open Configuration
File dialog. Select a file and press the [Open] button.
4.3.6.3 Creating an INI File
The Configuration Wizard utility should be used to generate new INI
files. Select Tools » Configuration Wizard… to open the Save Configuration
File dialog. The utility allows for a previously used INI file to be used as a
template. Please refer to section 3.2.
4.3.7
Output Statistics
Global statistics, such as ICR, OCR and deadtime fraction are
displayed along the top of the main window. Statistics for defined regions of
interest are displayed in the ROI table.
4.3.7.1 Real Time
This is simply the time elapsed between the Start Run and Stop Run
operations, measured in the DSP itself every 500µs with 800 ns accuracy.
Intermediate values read out during the run will therefore have the lower
accuracy, but the value reported at the end of the run will be fully accurate
4.3.7.2 Trigger Live Time
This is the measured live time of the Fast Trigger Filter, i.e. the time
that the Fast Trigger Filter output stays below threshold. This measurement is
used to accurately calculate the input count rate.
Note: The displayed Live
Time does not express a
relationship between the
OCR and ICR. The Dead
Time Percentage display
does relate OCR to ICR.
4.3.7.3 Energy Live Time
This DSP-level output is the computed live time of the Energy Filter.
This value is also accurate to 800 ns at end of run but is only computed at 10 ms
intervals during the run.
4.3.7.4 Input Count Rate (ICR)
The measured input count rate (ICR) is displayed in units of thousands
of counts per second [kcps]. The DSP applies internal correction procedures so
that the measured ICR is very close to the true ICR, especially for longer
peaking time settings. Please see section 6.9 for a discussion of this issue.
8/13/2010
35
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4.3.7.5 Output Count Rate (OCR)
The output count rate is also displayed in units of thousands of counts
per second [kcps]. The OCR is simply the total number of detected events that
did not pile up divided by the real time elapsed. Detected events that do not pile
up, but whose measured energy falls outside the spectrum upper and lower
limits, are called overflows and underflows, respectively. Both overflows and
underflows are included in the OCR.
4.3.7.6 Dead Time %
The Dead Time Percentage (%) is calculated as the OCR divided by the
ICR multiplied by the real time.
4.3.7.7 ROI Statistics
The peak centroid and width measurements are displayed in the ROI
table. The number of events and the OCR are also displayed for each ROI.
4.3.8
Single Channel Analyzer (SCA)
An SCA window is similar to a region of interest (ROI): It is a userdefined range of the energy spectrum. While an ROI is a software-level
construct used in analyzing a spectrum, an SCA window is processed at the
hardware level. The DSP counts all events within the window and stores the
result, the SCA Counts, in memory. SCA windowing thus allows for faster
readout. Instead of reading the entire spectrum, it is only necessary to read a
single word for each SCA. Select the SCA tab in the settings panel to view and
modify SCAs.
4.3.8.1 Creating SCAs
SCA windows can be transferred from the ROI table. Simply select the
desired ROI from the table (such that the green check-mark appears on the
desired row), then press the [Insert Active ROI] button. All defined ROIs can
be transferred in one step by pressing the [Insert All ROI] button. If necessary,
the upper and lower limits of each SCA can be adjusted manually by typing in
the appropriate box under “SCA Settings”.
4.3.8.2 Running with SCAs
Acquire data by starting and stopping an MCA run. Notice that when
the run is complete, the SCA table is updated with the sum of events in the SCA
window. Please refer to the Handel User Guide for help reading out SCAs
directly via the Handel device driver. The Mercury also provides a hard-wired
TTL output to register an event falling within an SCA, which may be connected
to a legacy counting system. Up to 16 (M-1) and 32 (M-4) ROI outputs are
available on a multi-pin connector, plus Trigger and Live Time In the case of
the Mercury bench-top models these signals are brought to 25-way connectors
on the front panel. See Appendix D for more information.
8/13/2010
36
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4.3.8.3 Trigger and Live Time Outputs
The SCA tab also displays a “Signal Output Control” panel that sets
Trigger and Live Time output signals on the multi-way connector. For each
channel, an Input Count Rate (Trigger) TTL signal may be selected from the
Fast, Baseline, or Energy Filters. Similarly, a Live Time signal can be selected
from the Fast, Baseline, or Energy filters, or all filters combined. In the
Mercury bench-top models the ICR and Live Time outputs are brought to the
25-way connectors on the front panel.
4.3.9
Saving and Loading Data
All data is stored in ASCII format and are easily readable using a text
editor. The default format for MCA data includes bin scaling and other basic
operating parameters and is date/time stamped. We would like to directly
support as many formats as possible—please let us know if your format is not
supported.
4.3.9.1 MCA Data
To Save or Load a
spectrum:
• Press the [Save MCA
Data] button in the
MCA Window, or
• Select Save MCA
Data… or Open MCA
Data… from the File
menu.
Spectra can be saved for later display or for analysis in another
program. Acquire a spectrum, then select Save MCA Data… from the File
menu. Saved MCA data can be displayed at any time. There are two methods
of displaying saved MCA data:
 If you want to compare a saved spectrum to currently acquired
spectrum, the saved data can be displayed as an overlay. First
acquire a spectrum, then select Open MCA Data… from the File
menu. Notice that the statistics and ROI table operate on the
current data only. Note also that if new data is acquired the stored
data is preserved. Zooming and panning operate by default on the
current data. To zoom and pan the saved data right-click in the
display area, select Active Plot and select the saved waveform. To
clear the stored data right-click in the data display area and select
Remove Ref.
 If you would like to analyze a stored spectrum, first right-click in
the data display area and select Clear Graph. Now select Open
MCA Data… from the File menu. Notice that the statistics, ROI
table and graphical tools now operate on the stored data. Note also
that if new data is acquired the stored data is automatically cleared.
4.3.9.2 Baseline Histogram Data
The baseline histogram can be saved for later display. Acquire a
baseline histogram, then select Save Baseline… from the File menu. A saved
baseline histogram can be displayed at any time. Again, there are two methods
of displaying a saved baseline histogram:
 If you want to compare a saved baseline histogram to currently
acquired data, the saved data can be displayed as an overlay. First
acquire a baseline histogram, then select Open Baseline… from
the File menu. Notice that the Baseline Mean and Baseline
FWHM operate on the current data only. Note also that if new
data is acquired the stored data is preserved. Zooming and panning
8/13/2010
37
DXP Mercury / Prospect User Manual
Prospect 1.0.x

operate by default on the current data. To zoom and pan the saved
data right-click in the display area, select Active Plot and select
the saved waveform. To clear the stored data right-click in the data
display area and select Remove Ref.
If you would like to analyze a stored baseline histogram, first rightclick in the data display area and select Clear Graph. Now select
Open Baseline… from the File menu. Notice that the Baseline
Mean and Baseline FWHM now operate on the stored data. Note
also that if new data is acquired the stored data is automatically
cleared.
4.3.9.3 Trace Data
Oscilloscope traces can be saved for later display. Acquire a trace, then
select Save Trace… from the File menu. A saved trace can be displayed at any
time. Again, there are two methods of displaying a saved baseline histogram:
 If you want to compare a saved trace to currently acquired data, the
saved data can be displayed as an overlay. First acquire a trace,
then select Open Trace… from the File menu. Note that if new
data is acquired the stored data is preserved. Zooming and panning
operate by default on the current data. To zoom and pan the saved
data right-click in the display area, select Active Plot and select
the saved waveform. To clear the stored data right-click in the data
display area and select Remove Ref.
 If you would like to analyze a stored trace, first right-click in the
data display area and select Clear Graph. Now select Open
Trace… from the File menu. Notice that zooming, panning and
cursor operations now operate on the stored data. Note also that if
new data is acquired the stored data is automatically cleared.
4.3.9.4 DSP Parameters
The DSP Parameters panel allows for the export of displayed data to a
file. Select Tools»DSP Parameters…, then press the [Export to File…]
button.
4.4 Run Control
This section covers run synchonization. Automatic run termination is
achieved by setting run Presets. Runs can be disabled in realtime according to a
user-provided logic signal using the GATE function. After stopping a run, the
run can be resumed without first clearing the data.
Run Control settings are
accessed in the MCA
Panel – select a preset type
from the drop-down list
and enter a value to select
the criteria for which the
current run is stopped.
8/13/2010
4.4.1
Run Presets (Automatic Run Termination)
Run presets determine the duration of the data acquisition run. The
Mercury can end the run when a specified preset real or live time has elapsed, or
when a specified number of events have been detected or processed.
38
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 4.11: The Run Preset settings in the MCA panel.
4.4.1.1 Run To Preset [ choice1 ]
Preset Value = choice2
Select the MCA tab in the Main Window, select a Preset Run type
from the drop-down list select the criteria for which the current run is stopped:
•
None - run ends when user presses “Stop” button.
•
Fixed Real Time – run ends after specified real time elapses.
•
Fixed Energy Live Time – run ends after specified live time
elapses.
•
Fixed MCA Events – run ends after specified number of valid
events have been processed into the MCA.
•
Fixed Input Triggers – run ends after specified number of input
events have been detected.
Now enter a value in the box below, e.g. 30 seconds, or 1000 counts.
Start a new data acquisition run and note that the run automatically stops when
the run preset criteria have been met.
4.4.2
The GATE Function
The external logic (BNC) input can be configured to halt data
acquisition simultaneously on all DXP channels according to an external logic
signal. The so-called GATE function supports TTL/CMOS levels. Please
review section 7.2.1 for a complete description of this feature, and run the
Configuration Wizard again if you'd like to enable the GATE function.
4.4.3
Resume Run: Clear or Retain MCA Data
When a new run is started the data from the previous run can either be
cleared or retained. This setting is accessed in the MCA window via the
Resume checkbox. If the box is checked the [Start/Stop] button has start/pause
functionality: new data accumulates with data from the previous run. If the box
is unchecked, data from the previous run is first cleared.
8/13/2010
39
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4.5 Display Controls
This section briefly describes tools and options related to the display of
acquired data.
4.5.1
MCA Auto Update / Refresh Rate
The MCA display can be set to refresh with a user specified interval. If
the Continuous box is checked, ProSpect will automatically refresh the display.
To set the refresh interval, select Options… from the Tools menu, and select the
Misc tab. Enter the desired interval in milliseconds. If the Continuous box is
unchecked the display is updated only when the [Update] button is pressed or
when the run ends.
4.5.2
Graphical Display Tools
ProSpect features a wide array of display controls, some or all of which
are accessible in each graphics window (i.e. MCA, Baseline and ADC panels).
Most of these controls can be accessed by right-clicking, as shown in the figures
below. These tools are intuitive and redundant.
4.5.2.1 Panning and Zooming
•
•
Left-click and drag in the plot window to zoom.
Left-click and drag on the x-axis or y-axis to pan.
The first two drop-down lists in the upper-left corner of the plot
window can override panning and zooming in the x-axis and yaxis, respectively:




Fixed – Locks the axis; prevents the y-axis from re-scaling
according to new data or user zoom or pan operations..
Floating – Un-locks the axis; prevents the y-axis from rescaling according to new data but allows user to zoom and
pan.
Auto – Auto-scales the axis; sets the axis dynamically such
that input data is fully displayed, but can be overridden by
user zoom or pan operations.
Fixed Auto – Auto-scales and locks the axis; sets the y-axis
dynamically such that input data is fully displayed, and
disables user zoom or pan operations
Figure 4.12: Panning and zooming are overridden by the settings in the dropdown list at the upper-left corner of the plot window.
8/13/2010
40
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4.5.2.2 Basic Tools and Options
Right-click in the plot window to view the basic display tools and
options. Some of these items are also accessible in the drop-down list items
above the plot area.
 Full Scale – Scales both axes such that all data is displayed.
 Set to Logarithmic (Set to Linear) – Sets the y-axis to
logarithmic or linear scale.
 Place Cursor – Select to place one of two cursors. Notice
that dX and dY values appear above the plot if both cursors
are placed.
 Save Spectrum – Save the current spectrum to an SPC file
(also under the File menu).
 Set as Reference – Save the current spectrum to a Reference
waveform, displayed in an alternate color. Subsequent data
acquired will display simultaneously with the fixed Reference
data. Note that you can also open saved waveforms from the
File menu; they will be displayed as a reference waveform.
 Reference on Top – Brings the Reference waveform to the
foreground.
 Use Reference Axis – Displays the Reference Y-axis to the
right of the plot window.
 Zoom Reference – Active only if the Reference axis is
displayed. If un-checked, zoom operations operate on both
waveforms in the x-axis, but only the active waveform in the
y-axis. If checked, zoom operations operate on both
waveforms in the x-axis, but only the Reference waveform in
the y-axis.
 Remove Reference – Clears a reference waveform from the
display.
 Clear Spectrum – Clears all data and the cursors.
Figure 4.13: Right-click in the Main display window to view the basic display
tools and options.
8/13/2010
41
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Alternatively, a default mode, e.g. zoom, ROI, or pan, can be selected
using the drop-down list menus above the plot area. The default mode applies
when you left-click in the display area. Left-click on either axis to pan. Most of
the tools are intuitive and redundant.
4.5.2.3 Cursor Tools and Options
After placing a cursor, right-click on the cursor itself to view the cursor
tools and options.
 Auto ROI (MCA panel only) – If the cursor was placed on an
energy peak this will automatically set the bounds of the
active ROI to enclose the peak.
 Crosshair – The default cursor is a vertical line. This option
sets the cursor to an x-y crosshair.
 Delete – Deletes the cursor.
Figure 4.14: After placing a cursor, right-click on the cursor itself to view the
cursor tools and options.
8/13/2010
42
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4.6 Optimizations
This section describes various parameter optimizations for the best
performance in throughput, pileup rejection and energy resolution.
4.6.1
Throughput (OCR)
The OCR depends only on the ICR and the dead time per event τd:
OCR = ICRt * exp -
(ICRt τd)
,where τd = 2*(tp + tg).
To increase the OCR at a given ICR, the dead time per event must be
reduced. The obvious first step is to reduce the energy peaking time tp. Further
improvements can be made by reducing the gap time tg.
4.6.1.1 Peaking Time (Energy Filter)
Making a plot of energy
resolution versus peaking
time provides a useful
future reference.
The peaking time is the energy filter length, or integration time, i.e. the
ramping interval of the trapezoid. It is the primary setting in determining the
balance between count rate performance and energy resolution.
Figure 4.15: The output response of the Energy Filter with peaking time
(ramping time) = 20.16µs and gap time (flattop time) = 0.96µs.
The trapezoid is the response to an x-ray.
You will generally find it useful, after making a first attempt to
optimize settings, to capture a set of spectra over a wide range of peaking times,
preferably over the full range that the Mercury supports and generate a plot of
energy resolution versus peaking time. This will serve two purposes: first to
serve as a standard of comparison, so that you can tell if further parameter
adjustments are helping or not; and second, to provide you with some feedback
8/13/2010
43
DXP Mercury / Prospect User Manual
Prospect 1.0.x
about whether your spectroscopy system is behaving properly. Later, when
everything is optimized and all the noise sources have been suppressed, you can
go back and repeat these measurements to provide hard data for use in selecting
the best peaking time for a given input count rate.
To change advanced
acquisition settings such as
the Minimum Gap Time:
Press the [Edit Filter
Parameters] button in the
Reducing the gap time can
significantly increase the
data throughput at a given
peaking time.
4.6.1.2 Gap Time (Energy Filter)
The gap time of the energy filter sets the flattop length of the output
trapezoid. Because the gap time directly affects the dead time, it is
advantageous to set the gap time as short as possible. The gap time is subject to
several constraints.
Generally the gap time should be set to a value that exceeds the 0 – 100
% preamplifier rise-time in response to a detected x-ray. As long as this
constraint is met, the trapezoid peak is tolerant of variations of the x-ray arrival
time relative to the ADC clock. The digital filter architecture further constrains
the gap time to an integer between 3 and 64 decimated clock intervals. In
ProSpect, the user sets the Minimum Gap Time slightly larger than the
measured preamplifier rise-time, and ProSpect automatically maintains the gap
time based on the decimation-dependent filter constraints. Please refer to
section 4.7.1.2 for details on using ProSpect to measure the rise time for your
system, and section 6.3.2 for a discussion of decimation and decimated clock
periods.
Figure 4.16: The Edit Filter Parameters panel.
8/13/2010
44
DXP Mercury / Prospect User Manual
Prospect 1.0.x
To edit the Minimum Gap Time press the Edit Filter Parameters
button in the Configuration settings panel to open a dialog. Enter the desired
value for Minimum Gap Time and press [OK].
Normally the Minimum Gap Time should be set to a value that
exceeds the preamplifier rise-time in response to a detected x-ray, however,
there is one exception. At very high count rates, where resolution is less of a
concern, it can be advantageous to set the Minimum Gap Time to a smaller
value, even to zero. This setting will only have an effect for decimation 0, i.e.
for peaking times less than 2.0 µs. For other decimations the gap time will be
set to the minimum value of 3 decimated clock cycles. Note that you may have
to adjust other settings as a result:
1. Peak Sample Offset – Because you are reducing or even
eliminating the flattop section of the trapezoid, performance
becomes more sensitive to the energy sampling time. Refer to
section 4.6.3.6.
2. Gain Calibration – A consequence of setting the gap time less
than the preamplifier rise-time is ballistic deficit: The peak value
of the trapezoid is reduced. As a result you will almost certainly
have to increase the gain after the gap time has been changed. See
section 4.3.5.
4.6.2
Pileup Rejection
Pileup inspection is described in detail in section 6.8. These settings
should only be modified by users with a good understanding of the principles of
pileup inspection.
Figure 4.17: Slow, or Energy, filter output waveform diagram.
CAUTION: When set too
low, the MAXWIDTH
criterion can reject nonpiled-up x-rays, resulting
in attenuation at higher
energies.
8/13/2010
4.6.2.1 Maximum Width Constraint
The DSP parameter MAXWIDTH sets the maximum acceptable time
that the fast trigger output can stay above threshold for a single event. Properly
45
DXP Mercury / Prospect User Manual
Prospect 1.0.x
set, this constraint detects fast pileup (event separation on the order of 100ns).
See section 6.8 for more information. The Max Width setting is accessible in
the Edit Filter Parameters panel. By default it is set to 400ns, allowing for a
preamplifier rise-time up to 200ns. MAXWIDTH should be at least twice the
fast peaking time plus the preamplifier’s 1% settling time:
Max Width > 2*Fast Peaking Time + Preamp Rise-time
4.6.2.2 Peak Interval
The DSP parameter PEAKINT sets the minimum acceptable time that
the slow energy filter needs to process a single event, i.e. the interval between
peaks that can be properly sampled. This constraint detects slow pileup (event
separation on the order of the energy peaking time). See section 6.8 for more
information.
The optimum peak interval is usually fixed relative to the sum of the
peaking time and gap time. The Peak Interval Offset is measured forwards in
time from this sum, i.e. measured forwards from the end of the flattop period
(see Figure 4.17):
Peak Interval = Peaking Time + Gap Time + Peak Interval Offset
The Peak Interval setting is accessible in the Edit Filter Parameters panel.
In most cases it should be left at zero. Larger values will result in a more
conservative pileup inspection at the cost of increased deadtime-per-event.
4.6.2.3 Reducing the Fast Peaking Time
At very high rates the fast
filter peaking time may be
reduced, to maintain good
pileup inspection.
The default fast peaking time of 100 ns should be used in most cases.
Generally speaking, a longer fast filter peaking time produces a lower pileup
inspection threshold at the cost of a longer pileup inspection time interval. Little
if any real benefit is derived from increasing the fast peaking time unless the
preamplifier signal is extremely noisy. For good pileup rejection, the fast filter
peaking time should be much shorter than the energy filter, which becomes a
problem when the shortest energy filter peaking times are used. In these cases,
some improvements in pileup rejection may be possible if the fast filter peaking
time is reduced, e.g. to 60ns. We don’t recommend using a fast gap time other
than zero.
Open the Edit Filter Parameters panel and enter a new value for the
Fast Trigger Filter Peaking Time and press [Apply]. Note that you may have
to adjust other settings as a result:
1. Trigger Threshold – Because of the zero gap time, the Fast
Trigger Filter normally produces some ballistic deficit. Reducing
the trigger peaking time can heighten this effect. For best results
the threshold be checked, as described in section 4.3.3.2.
2. Max Width – The time over threshold is directly related to the
filter length. If you previously optimized Max Width, i.e. the
maximum time over threshold, you may need to re-optimize. See
section 4.6.2.1.
4.6.3
Energy Resolution
There are many possible reasons for poor energy resolution. This
section points to the most common issues.
8/13/2010
46
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4.6.3.1 Proper Peaking Time Selection
The first step in improving energy resolution is, of course, to optimize
the Peaking Time. Use your plot of energy resolution versus peaking time to
select a peaking time where you get good energy resolution before making these
adjustments.
4.6.3.2 Baseline Acquisition
Capturing good baseline values and proper averaging are vital to
achieving good energy resolution. The Baseline Threshold and Baseline
Average Samples settings must be set properly for a given peaking time. See
section 4.7.2 for making adjustments in ProSpect, and section 6.4 for more
detailed explanations of baseline issues.
4.6.3.3 Eliminate Noise Pickup
Noise pickup can destroy performance. It is very important to identify
and eliminate excess noise in the hardware. Typically this involves eliminating
ground loops, removing switching power supplies in close proximity and
improving shielding. Please refer to section 4.7.1 below for a brief introduction
to using the ADC panel to identify noise issues.
4.6.3.4 Sufficient Gain to Sample Noise
If the signal gain is such that noise is not properly digitized at the ADC,
energy resolution will not be optimal. This would result from a Preamp Gain
setting that is too high (resulting in a Mercury variable gain setting that is too
low). Set the gain so that the noise is sufficiently digitized – see section 4.7.1.1.
4.6.3.5 Sufficient Gap Time
If the gap time is too short, the trapezoid peak sample (the energy
sample) becomes dependent on the arrival time of the x-ray relative to the ADC
clock. Make sure that the Minimum Gap Time is longer than the preamplifier
rise-time as described above in section 4.6.1.2.
4.6.3.6 Peak Sampling Time
The optimum sampling time of the energy filter is usually fixed relative
to the sum of the peaking time and gap time. The Peak Sample Offset is
measured backwards in time from this sum, i.e. measured backwards from the
end of the flattop period (see Figure 4.17):
Sampling Time = Peaking Time + Gap Time – Peak Sample Offset
The Peak Sample Offset setting is accessible in the Edit Filter Parameters
panel. In most cases it should not be edited. An exception is when running
with a very short gap time at decimation 0 (see section 4.6.1.2 above). In this
case the Peak Sample Offset should be reduced empirically. Please see section
6.6.2 for a full discussion before attempting this procedure.
8/13/2010
47
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4.7 Diagnostics
The Mercury and ProSpect provide several diagnostic tools for
identifying and resolving functional and performance issues.
To open the Traces panel,
click on the Traces tab in
the main window.
4.7.1
The ADC Panel (Oscilloscope)
The ADC Panel displays waveforms captured at various stages of the
DXP's digital filter, such as the preamplifier signal as seen at the Mercury’s
analog-to-digital-converter (ADC). Each 16-bit output is plotted over 4096
sample points, with a user settable sampling interval. It can be a useful
diagnostic tool for checking preamplifier polarity and gain, measuring the risetime, and for tracking down noise pickup and baseline irregularities. Note that
signed waveforms are shifted such that zero is displayed at the midpoint value
32,768.
To acquire and view a waveform:
1. Click the drop down list to select the desired waveform:
 ADC – The digitized (unsigned) ADC waveform, i.e. the input
signal to the digital filters.
 Baseline History – The (signed) baseline running average.
These are the actual values that are subtracted from the energy
filter.
 Trigger Filter – The (signed) raw fast filter output.
 Baseline Filter – The (signed) raw intermediate filter output.
For decimation 0 this is the same as the Energy filter.
 Energy Filter – The (signed) raw slow filter output.
 Baseline Samples – The (signed) intermediate filter output
baseline samples selected for use in the running average.
 Energy Samples – The (unsigned) slow filter output peak
samples selected for the energy spectrum.
2. Enter a value ranging 0.020 - 1000 microseconds in the Sampling
Interval field. This is the time between each displayed sample.
3. Press [Get Trace].
8/13/2010
48
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 4.18: The ADC panel is a useful diagnostic tool.
The preamplifier Gain and
Polarity settings are
accessed in the Detector
tab of the Settings panel:
8/13/2010
4.7.1.1 Determining the Preamplifier Polarity and Gain
A common configuration error involves setting either the preamplifier
signal polarity or gain incorrectly. Note: The preamplifier type, i.e. pulsed-reset
or RC-feedback, is determined by the firmware file that is downloaded to the
hardware (see section 3.2.1)
The Preamplifier Polarity configuration setting determines whether the
ADC code is inverted prior to the digital filter pipeline, which expects x-ray
pulses with a rising edge. The ADC panel displays the digital signal after this
optional inversion. If the x-ray pulses are displayed with a falling edge, as
shown in Figure 4.19, then the polarity setting is incorrect; if pulses are
displayed with a rising edge, as in Figure 4.20, then the polarity setting is
correct.
The Preamp Gain setting in the Detector panel, in combination with
the Dynamic Range setting in the Configuration panel, determines the
Mercury analog variable gain setting. The variable gain is set such that an x-ray
with energy equal to the dynamic range value produces a voltage step of the
maximum allowable amplitude at the ADC input. X-rays with energies
exceeding the dynamic range value cannot be processed correctly. The
Dynamic Range setting should thus be set above the largest x-ray energy
present in the system.
In order to get the best energy resolution the gain must be set such that
electronic noise is digitized sufficiently that it can be properly filtered. In
practice this means that the noise should span 20 or more vertical units in the
display. In Figure 4.19 the noise is contained in less than 10 displayed vertical
units, indicating that the hardware gain setting is too low. This could be due
49
DXP Mercury / Prospect User Manual
For best results the noise
should span 20 or more
vertical units in the
Oscilloscope Panel display.
Prospect 1.0.x
either to a Preamp Gain configuration setting that is too high or to the
Dynamic Range being set too large. The noise displayed in Figure 4.20 spans
approximately 40 vertical display units, indicating that the Preamplifier Gain is
set correctly and the spectrum is properly sized.
To adjust your own system, first select the ADC tab in the main
window and acquire a few traces, until you have recognizable x-ray events
displayed. Try increasing the Sampling Interval to acquire longer traces.
Compare the polarity and noise amplitude to the figures. If necessary, change
the Polarity and Preamp Gain. You may also need to modify the Dynamic
Range.
Figure 4.19: An ADC trace of a reset-type detector with the Mercury
configured with the wrong polarity and a gain setting that is too
low. X-ray steps displayed in this panel should have a rising edge,
and noise should span 20 or more vertical units.
8/13/2010
50
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 4.20: An ADC trace with correct polarity and a typical gain. Note that
the noise is well digitized at roughly 50 vertical units.
4.7.1.2 Measuring the Preamplifier Rise-time
The ADC panel is useful for measuring the preamplifier signal risetime, which should be done before modifying the Minimum Gap Time as
described in section 4.6.1.2.
As mentioned earlier, the minimum sampling interval in the display is
20ns—the actual ADC sampling period. Acquire an ADC trace at the minimum
sampling interval of 0.020µs that includes at least one well separated x-ray
event. Use the zoom tool (accessed via the right click menu or through the
display controls at the graph’s upper left) to expand the horizontal axis about the
selected event. Place Cursor 1, by right-clicking in the display area and
selecting “Place Cursor 1”, immediately before the x-ray pulse. Similarly,
place Cursor 2 immediately after the signal has settled following the pulse. The
dX field of the cursor data area in the upper right hand corner should now
display the 0 – 100% preamplifier rise-time in µs.
8/13/2010
51
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 4.21: Use the zoom function and cursors to measure the preamplifier
rise-time. The rise-time here is approximately 0.5µs (as displayed
in the dX field at the upper-right).
4.7.1.3 Measuring the RC Decay Time τ (RC-Feedback Preamplifiers
only)
The ADC panel is also useful for measuring the decay time for RCfeedback preamplifiers.
Acquire an ADC trace that includes at least one well separated x-ray
event. Use the zoom tool (accessed via the right click menu or through the
display controls at the graph’s upper left) if necessary to expand the horizontal
axis about the selected event such that the entire decay time is displayed. Place
Cursor 1, by right-clicking in the display area and selecting “Place Cursor 1”,
at the peak value of the x-ray pulse. Similarly, place Cursor 2 immediately
before the x-ray pulse such that a baseline value is selected. Record the dY
value from the cursor data display—this is the pulse height. Now move Cursor
1 to the point on the decay curve that produces a new dY value that is 1/e times
the measured pulse height:
dY’ = (1/e) · dY ~ 0.37 · dY
The cursors should now be separated by the time constant τ, displayed in µs in
the dX field.
4.7.1.4 Optimizing the Baseline Average Length
The ADC panel is also useful for optimizing the number of samples in
the baseline average. Please first review section 6.4 for a thorough discussion of
baseline acquisition.
The Baseline Average Length refers to the number of samples of the
Baseline Filter raw output in the running average of baseline samples, which
should be:
•
8/13/2010
Large enough to average out electronic noise at the higher frequencies.
52
DXP Mercury / Prospect User Manual
Prospect 1.0.x
•
Small enough to track low frequency fluctuations in the front end, i.e.
detector dark current.
The precise noise distribution and the low-frequency signal
characteristics of the detector and preamplifier together yield an optimum
number of samples in the average. Too small a value will not allow for proper
filtering of electronic noise. Too large a value will not allow proper tracking of
low frequency signals, e.g. due to EMI, that can be cancelled out with double
correlation.
First select Baseline Filter from the drop-down list, set the Sampling
Interval to 1.000µs and press [Get Trace] to view the raw output of the
baseline filter.
Figure 4.22: The raw output of the Baseline Filter. Approximately 10 x-ray
events are visible as trapezoid, as well as a preamplifier reset
spike. The goal of baseline acquisition is to sample the baseline
noise.
Now select Baseline History from the drop-down list, set the
Sampling Interval to 10.000µs and press [Get Trace] to view the baseline
running average. You want to achieve a waveform similar to that shown in
Figure 4.24, where noise is filtered out but the average still tracks real
variations. If you see something more like Figure 4.23 or Figure 4.25, adjust
the Baseline Average Samples setting in the Configuration panel and press
[Apply]. Acquire another trace and adjust as necessary. In most cases the
values 128 and 256 yield the best results. At high rates it may be advantageous
to reduce the number of samples as low as 16. For near perfect preamplifiers the
average can be increased to 1024 or more. In any case the optimization is not
complete until you acquire a spectrum and verify the energy resolution has
improved.
8/13/2010
53
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 4.23: The baseline average with the number of samples set too low (16
samples in the average). Notice that there is still a lot of noise, as
well as some real variations in the baseline.
Figure 4.24: The baseline average with the number of samples set properly (256
samples in the average). Notice that there is virtually no noise, but
that real baseline variations are tracked. In this case the downward
variation is due to some curvature in the preamplifier output
following a reset.
8/13/2010
54
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 4.25: The baseline average with the number of samples set too high
(4096 samples in the average). Notice that although there is no
noise, the average lags behind a real downward variation in the
baseline.
To open the Baseline
panel, click on the Baseline
tab in the main window.
8/13/2010
4.7.2
The Baseline Panel
Baseline measurements are continually updated samplings of the output
of the energy filter when no event is being processed. A running average of
these baseline samples is then made to reduce the noise in this measurement and
the result is subtracted from instantaneous raw pulse-height measurements to
determine their true amplitudes. Please first review section 6.4 for a thorough
discussion of baseline acquisition.
The Baseline panel displays a histogram of the instantaneous baseline
samples. The histogram, in combination with the Baseline History and
Baseline Filter traces (see section 4.7.1.4 above) are powerful tools for
diagnosing electronic noise and common nonlinearities in the detector and
preamplifier. Select the Baseline tab and press [Get Baseline] to acquire a
baseline histogram. You should see a Gaussian peak with few, if any, outliers,
as in Figure 4.26. If there are many outliers to the right of the peak, as in
Figure 4.27, the threshold is set too high. If the right side of the peak is
attenuated, non-Gaussian as in Figure 4.28, the threshold is set too low.
55
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 4.26: A good baseline histogram: the shape of the noise peak is
Gaussian with no outlying data points.
4.7.2.1 The Baseline Threshold
Section 4.3.2 includes a discussion about setting thresholds based on
visual feedback in the MCA panel. Threshold settings also affect baseline
acquisition: Baseline acquisition is enabled only when the Fast Trigger Filter
output is below Trigger Threshold and the Baseline Filter output is below the
Baseline Threshold. It is assumed that the Trigger Threshold is set
conservatively, so that baseline acquisition is dominated by the Baseline
Threshold. Note: The baseline threshold is not available for decimation 0, i.e.
peaking times less than or equal to 2 µs.
Edit the Baseline Threshold value in the Configuration tab and press
[Apply]. Typical values range from 150 eV to 1000 eV. The baseline filter
length is linked to the energy filter length, or peaking time, thus the baseline
threshold should be optimized every time the peaking time is changed. All
thresholds must be readjusted if the gain changes significantly. For this reason it
is useful to save INI files for commonly used peaking times after optimizations
are complete.
8/13/2010
56
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 4.27: A baseline histogram with the threshold too low. Notice that the
right side of the noise peak is attenuated. The rest of the noise
peak will show up in the energy spectrum.
Figure 4.28: A baseline histogram with the threshold too high. The baseline
samples to the right of the noise peak are partial energy events that
should be in the energy spectrum.
Warning: Changing DSP
parameters without
understanding them is
discouraged..
8/13/2010
4.7.3
DSP Parameters
The DSP Parameters panel, accessible via the Tools menu, provides a
diagnostic display of all DSP's internal parameters for the currently selected
processing channel. The Hex and Decimal radio buttons determine whether
parameters are displayed in hexadecimal or decimal format. Press the [Update]
button to refresh the display. Note that various active parameters will change
57
DXP Mercury / Prospect User Manual
Prospect 1.0.x
every time the [Update] button is pressed. The Channel selector immediately
loads the new processing channels parameters.
4.7.3.1 Generating a Diagnostic DSP Parameters File
The [Export to File…] button generates a ASCII text file containing
all of the parameters for the currently selected processing channel. You may be
asked to generate this file by technical support. More likely you will be asked to
generate a full Error Report, as described in section 4.7.4.
Figure 4.29: The DSP Parameters panel. Do not modify these values unless as
instructed by XIA support staff.
4.7.3.2 Modifying DSP Parameters
In some cases, as directed by the XIA support staff, it may be necessary
to modify the DSP’s operating parameters directly. To edit a parameter select
the field using the mouse, enter the new value and press [Return]. If you do not
press [Return] the parameter will return to its unmodified value when another
item is selected. Changing parameters in this panel without a deep
understanding of XIA’s DSP processors may produce exotic and unpredictable
results. We recommend doing so only under the guidance of XIA support staff.
4.7.4
Submitting a problem report:
XIA encourages customers to report any problems encountered using
any of our software via email. In most cases, the XIA engineering team will
need to review bug information and run tests on local hardware before being
able to respond.
All software-related bug reports should be e-mailed to
[email protected] and should contain the following information, which
will be used by our technical support personnel to diagnose and solve the
problem:
 Your name and organization
8/13/2010
58
DXP Mercury / Prospect User Manual
Prospect 1.0.x

Brief description of the application (type of detector, relevant
experimental conditions...etc.)

XIA hardware name and serial number

Version of the library (if applicable)

OS

Description of the problem; steps taken to re-create the bug

Full Error Report (see section 4.7.4.1) plus additional data:
o
Saved MCA data, if relevant (see section 4.7.4.2)
o
Saved Baseline data, if relevant (see section 4.7.4.3)
o Saved ADC data, if relevant (see section 4.7.4.4)
Please compress the Error Report into a ZIP archive and attach the support
request email.
4.7.4.1 Generating a Full Error Report
If you are unable to solve your problem using the diagnostic tools in
ProSpect, the last resort is to send as much information as possible to XIA
support. This task is facilitated by the Generate Error Report… command
under the Help menu. This feature creates a sub-directory titled
"ProSpect_xxxxxx_Rpt" (where "xxxxxx" is the date) in the ProSpect
installation directory that contains all DSP parameters, ProSpect registry
settings, INI and log files, Handel log files, etc.
4.7.4.2 Saving MCA Data
If you are having difficulty acquiring a spectrum, or the spectrum looks
strange, please save and submit a sample MCA file with the Error Report. In
the MCA tab, acquire a spectrum, then press [Save MCA Data]. Save the file
in the "ProSpect_xxxxxx_Rpt" sub-directory in the ProSpect installation
directory.
4.7.4.3 Saving Baseline Data
If you are having difficulty acquiring a good baseline histogram, or the
spectrum looks strange, please save and submit a sample baseline histogram file
with the Error Report. In the Baseline tab, acquire a histogram, then press
[Save Baseline]. Save the file in the "ProSpect_xxxxxx_Rpt" sub-directory in
the ProSpect installation directory.
4.7.4.4 Saving ADC Data
If the ADC or filter output traces look strange, please save and submit
a sample trace file with the Error Report. In the ADC tab, acquire a trace, then
press [Save Trace]. Save the file in the "ProSpect_xxxxxx_Rpt" sub-directory
in the ProSpect installation directory.
8/13/2010
59
DXP Mercury / Prospect User Manual
Prospect 1.0.x
5 Mapping Mode
Mapping mode supports a number of time-resolved experiments, but is
primarily intended for x-ray scanning applications wherein the x-ray beam is
systematically scanned over a surface in order to produce a pixilated spectral
map of the sample. Data is stored sequentially at each 'pixel advance'
instruction, which can be provided by the host computer or triggered by an
external logic signal, i.e. GATE or SYNC.
Mapping mode utilizes a different System FPGA configuration, i.e.
firmware must be downloaded when switching between acquisition modes. A
dual-memory architecture allows the Mercury to continuously acquire data in
this mode. Host readout must be synchronized with the hardware in order to
sustain continuous operation.
5.1 Pixel Advance Settings
The pixel advance settings are included in the Configuration Wizard
utility, described in section 3.2. To run the utility select Configuration
Wizard… from the Tools menu.
At the beginning of a run the pixel number starts at zero, corresponding
to the x-ray beam initial position. The pixel number advances in several
possible ways, either using digital hardware lines for real time applications or by
computer control. These methods are described in detail below.
5.1.1
Pixel Advance on GATE Edge
The recommended method for advancing the pixel is to use the GATE
digital input, where the pixel advance occurs on every trailing edge of the signal
(the transition from active data acquisition to the inactive state). By default the
GATE signal also halts data acqusition when it is LO and the pixel advances on
every falling edge, as in Figure 5.1.
Figure 5.1: GATE pixel advance, with data acquisition halted during the LO periods. The pixel advance
occurs at each falling edge.
5.1.1.1 GATE Polarity
The interpretation of the user-provided GATE signal can be inverted
such that data acquisition is halted when the signal is HI (see Figure 6.2) and
the pixel advance occurs on rising edges.
8/13/2010
60
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 5.2: GATE pixel advance, with data acquisition halted during the HI periods. The pixel advance
occurs at each rising edge.
5.1.1.2 GATE Ignore Setting
Normally the GATE signal is used both to advance the pixel and to halt
data taking. In some cases the user may prefer continuous operation through the
pixel advance operations. With the GATE Ignore option selected, the pixel
advance occurs but data acquisition is not halted. Note that the data acquisition
is halted at the beginning of the run until GATE is released the first time, i.e. run
start synchronization is still available.
Figure 5.3: GATE pixel advance, with LO polarity and the Gate Ignore enabled. Acquisition is halted
initially until GATE is released for the first time. As before, the pixel advance occurs at each
falling edge, but acquisition continues during the LO periods.
5.1.2
Pixel Advance using SYNC Clock
The SYNC signal can also be used to generate the pixel advance. Using
this method, the pixel will advance for every N positive pulses on the SYNC
line, where N can be set from 1 to 65535. Note that the pulses must be greater
than 40 ns wide to be guaranteed to be recognized.
Figure 5.4: SYNC pixel advance with GATE used to synchronize the run start. Acquisition is halted
initially until GATE is released for the first time. The pixel advance occurs every N cycles of
the SYNC clock.
8/13/2010
61
DXP Mercury / Prospect User Manual
5.1.3
Prospect 1.0.x
Pixel Advance under Host Control
It is also possible to advance the pixel using Handel. Manually
advancing the pixels is slower and unwieldy because the command must be
issued separately to each module, but it does provide an easy way to test
mapping operations.
5.2 Mapping Mode Data Acquisition
Mapping mode acquisition is automatically enabled when the Mapping
panel is active. Conversely, normal mode acquisition is automatically enabled
when the MCA tab is active. In both cases the corresponding System FPGA
firmware is downloaded (as evidenced by LED pattern flashing).
5.2.1
The Mapping Panel
Figure 5.5: The Mapping panel.
Click the Mapping tab to display the Mapping panel. Please be
patient while the System FPGA firmware is downloaded, during which period
the controls will be grayed out. The layout and functionality are similar to the
MCA panel. The [Start Mapping] button starts and stops mapping data
acquisition runs. Output statistics are displayed as before along the top right
section of the panel. Notice that data and statistics do not update in real-time.
This is as a result of the dual memory architecture. The display is refreshed only
after a memory buffer swap.
5.2.2
Mapping Mode: MCA or SCA
The MCA and SCA mapping modes are distinct and exclusive of one
another: In MCA mode full spectra are stored in memory; In SCA mapping
mode only the tables of SCA counts are stored in memory. Click the MCA or
8/13/2010
62
DXP Mercury / Prospect User Manual
Prospect 1.0.x
SCA radio button to select the desired mode. Section 0 details the data formats
for MCA and SCA mapping modes.
5.2.3
Total Number of Pixels
A data acquisition run can be ended manually via the host at any time.
In most cases the length of a data acquisition run corresponds to the number of
pixels in the map. The Total Number of Pixels setting instructs the DSP to
automatically end the run after that number of pixel advances have been
detected. If Indefinite Run is checked, the run ends only upon the "End Run"
host command.
5.2.4
Buffer Control
The Mercury uses a dual memory architecture to achieve continuous
data acquisition: The Mercury writes into one 'active' buffer while the host reads
the other 'inactive' buffer. The size of each buffer is 2MB, organized as
1Mword by 16 bits. The Mercury must periodically swap buffers. The swap
must occur after the host has completed reading the inactive buffer and before
the active buffer becomes full.
The Mapping Pixels Per Buffer setting controls when the swap takes
place. The DSP also calculates the maximum number of pixels that can be
stored in a memory buffer based upon the MCA Number of Bins. If the
Maximum Allowed checkbox is checked, or if a greater number is entered, the
Mercury will override the Mapping Pixels Per Buffer and use this calculated
maximum number instead.
The effects of this setting are generally qualitative:
 Changes the structure but not the content of the data stream
binary file (see section 0).
 The ProSpect display updates after each buffer, thus the
number of pixels per buffer sets the visual refresh rate.
5.2.5
Mapping Mode Data Acquisition
Press the [Start Mapping] button to begin a data acquisition run. If
you are using GATE or SYNC for the pixel advance, you should see the
Progress Bar begin to fill. If it does not, please review sections 5.1 and 3),
adjust your hardware and use the Configuration Wizard (see section 3.2.3) to
correct your settings if necessary. Alternatively, press the [Next Pixel] button to
manually advance pixels. When the incremental number of processed pixels
exceeds the Mapping Pixels Per Buffer setting, the buffer will be read and
spectrum (or SCA data) and statistics corresponding to the first pixel from the
buffer will be displayed. When the total number of pixels processed exceeds the
Total Number of Pixels setting, the display will again refresh and the run will
automatically end. Alternatively, the [Stop Mapping] button can be pressed at
any time to end the run.
8/13/2010
63
DXP Mercury / Prospect User Manual
Prospect 1.0.x
5.3 Mapping Mode Data
In mapping mode, the Mercury uses two completely separate memory
buffers, enabling the system to take data into one buffer while the other buffer
can be read out by the host. The capacity of each buffer is 2MB, organized as
1Mword by 16 bits, however, the portion that is used can be significantly
smaller, depending on the Mapping Pixels Per Buffer setting (see section
5.2.4). Dealing with all this data in real time is a challenge. ProSpect streams
the buffer data to a binary file for offline analysis. For visual feedback, the
display is refreshed with data from the first pixel of each buffer that is read out.
5.3.1
Mapping Data Options
The name of the output file is automatically generated with the user
specified File Prefix. For diagnostic purposes, the user can elect not to save
output data. If the Save Mapping Data checkbox is un-checked, the output data
will be lost.
The default location of the output file is:
C:\Program Files\xia\ProSpect 0.1\data
To change the folder, select Options from the Tools menu, then select the
Mapping Data tab to display the Mapping Data File Options. For the present
release each data acquisition run generates a single binary file in the specified
folder, with a unique filename with the specified prefix. Note: When using
Windows Vista and above, the mapping data should be stored outside the
Program Files folder to avoid problems with data storage and retrival.
5.3.2
Mapping Data Format
Currently the data file created is a binary concatenation of all the
buffers read out during the run. Table 5.1 shows the structure of the file
created. The following sections specify the buffer level formatting.
Buffer 0: Module 0
Buffer 0: Module 1
Buffer 0: Module 2
Buffer 0: Module 3
Buffer 1: Module 0
Buffer 1: Module 1
Buffer 1: Module 2
Buffer 1: Module 3
…
8/13/2010
64
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Buffer i-1: Module 0
Buffer i-1: Module 1
Buffer i-1: Module 2
Buffer i-1: Module 3
Table 5.1: The Mapping mode binary data output file format of a 4 Mercury
module system for a data acquisition run comprising 'i' buffers.
5.3.3
Single Buffer Format
The buffer starts out with a buffer header, containing general
information about the data contained in the memory block, followed by pixel
data. For each pixel there is a pixel data block containing a header and spectra
for all 4 channels. The pixel data block format depends on the mapping mode:
MCA, SCA or List Mode. Table 5.2 shows the buffer level structure. The
following sections describe the header and pixel data blocks in detail.
Buffer Header
Pixel X Data Block
Pixel X+1 Data Block
…
Pixel X+j-1 Data Block
Table 5.2: A single buffer for a data acquisition run with Mapping Pixels Per
Buffer = 'j'.
8/13/2010
65
DXP Mercury / Prospect User Manual
Prospect 1.0.x
6 Digital Filtering: Theory of Operation and
Implementation Methods
This chapter provides an in-depth discussion of x-ray pulse-processing
theory both generally and as implemented in the DXP Mercury. The topics
include x-ray detection how, digital trapezoidal filter basics, thresholds,
baselines, peak sampling, pileup inspection, and input and output count rates.
Topics are covered to illustrate the theoretical issues, practical implementation,
and how to adjust parameters to obtain best performance.
The acronym DXP stands for “Digital X-ray Processor” and refers to a
digital processing technology, for which XIA has received several US and
International patents.
6.1 X-ray Detection and Preamplifier Operation:
Energy dispersive detectors, which include such solid state detectors as
Si(Li), HPGe, HgI2, CdTe and CZT detectors, are generally operated with
charge sensitive preamplifiers. When an x-ray is absorbed in the detector
material it releases an electric charge Qx = Ex/ε, where the material constant ε is
the amount of energy needed to form an electron-hole pair. Qx is integrated onto
the preamplifier’s feedback capacitor Cf, to produce the voltage Vx = Qx/Cf =
Ex/(εCf). Measuring the energy Ex of the x-ray therefore requires a
measurement of the voltage step Vx in the presence of the amplifier’s noise σ.
Figure 6.1 and Figure 6.3 depict reset-type and RC-type charge sensitive
amplifiers, respectively. In both figures the detector D is biased by voltage
source HV (either positive or negative) and connected to the input of amplifier
A. Note that the signal polarity must be distinguished from the bias voltage
polarity. The signal polarity is positive if the voltage step Vx is a rising edge, as
displayed in Figure 6.1. Whether signal polarity is positive or negative depends
upon the preamplifier’s design and does not depend upon bias voltage polarity,
which is specified on the detector and is determined by its design.
6.1.1
Reset-Type Preamplifiers
Figure 6.1a is a simplified schematic of a reset-type preamplifier,
wherein Cf is discharged through the switch S from time to time when the
amplifier’s output voltage gets so large that it behaves nonlinearly. Switch S
may be an actual transistor switch, or may operate equivalently by another
mechanism. In pulsed optical reset preamps light is directed at amplifier A’s
input FET causing it to discharge Cf. In transistor reset preamps, the input FET
may have an additional electrode which can be pulsed to discharge Cf. The
output of a reset-type preamplifier following the absorption of an x-ray of
energy Ex in detector D is a voltage step of amplitude Vx. Two x-ray steps are
shown in Figure 6.3
8/13/2010
66
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 6.1: a) Reset-type charge sensitive preamplifier with a negatively biased detector; b) Output on
absorption of x-ray rays. Note that the steps have a rising edge, so that the signal polarity is
positive.
Figure 6.2 depicts the large-signal sawtooth waveform that results from
successive x-ray steps followed by the reset. Note that the units here are Volts
and milliseconds vs. millivolts and microseconds in the previous figure.
Figure 6.2: The large-signal reset waveform for a reset-type preamplifier with positive signal polarity, as
displayed on a real oscilloscope. Note that the large signal character of the DXP Mercury
diagnostic ADC readout, used in ProSpect’s ADC panel, looks quite different because of the
dynamic range reduction carried out in the ASC, as described in section 0.
6.1.2
RC-Type Preamplifiers
Figure 6.3a is a simplified schematic of an RC-type preamplifier,
wherein Cf is discharged continuously through feedback resistor Rf. The output
of an RC-type preamplifier following the absorption of an x-ray of energy Ex in
detector D is, again, a voltage step of amplitude Vx. The continuous discharge
of Cf through Rf results in an exponential voltage decay after the x-ray step, with
decay constant τ, where:
8/13/2010
67
DXP Mercury / Prospect User Manual
Prospect 1.0.x
τ = Rf Cf
Equation 6-1
In practice the decay time may depend on subsequent circuitry, i.e. if a
pole-zero cancellation circuit is used, thus τ may not be directly related to the
feedback elements of the front-end. The point of this simplified model is that
the resulting waveform is a single-pole RC decay. The discussion in section 6.2
through section 6.6.2 assumes a reset-type preamplifier, but is mostly applicable
to RC-type preamplifiers. section 6.7 describes the few key differences in the
processing of RC-type preamplifier signals.
Figure 6.3: a) RC-type charge sensitive preamplifier with a positively biased detector; b) Output on
absorption of an x-ray. Note that the step has a falling edge, thus the signal polarity is
negative.
6.2 X-ray Energy Measurement & Noise Filtering:
Reducing noise in an electrical measurement is accomplished by
filtering in the frequency, or conversely, the time domain. When discussing
digital pulse-processor filters it's more straightforward to use the time domain.
Traditional analog pulse-processing filters use combinations of a differentiation
stage and multiple integration stages to convert the preamp output steps, such as
shown in Figure 6.1b, into either triangular or semi-Gaussian pulses whose
amplitudes (with respect to their baselines) are then proportional to Vx and thus
to the x-ray’s energy.
6.2.1
Digital Filtering Theory
Digital filtering proceeds from a slightly different perspective. Here
the signal has been digitized and is no longer continuous, but is instead a string
of discrete values, such as shown in Figure 6.4. The data displayed are actually
just a subset of Figure 6.3b, which was digitized by a Tektronix 544 TDS digital
oscilloscope at 10 MHz (10 million per second). Given this data set, and some
kind of arithmetic processor, the obvious approach to determining Vx is to take
some sort of average over the points before the step and subtract it from the
value of the average over the points after the step. That is, as shown in Figure
8/13/2010
68
DXP Mercury / Prospect User Manual
Prospect 1.0.x
6.4, averages are computed over the two regions marked “Length” (the “Gap”
region is omitted because the signal is changing rapidly here), and their
difference taken as a measure of Vx. Thus the value Vx may be found from the
equation:
Vx,k = – Σ wivi +
i (before)
Σ
i (after)
wi vi
Equation 6-2
where the values of the weighting constants wi determine the type of
average being computed. The sums of the values of the two sets of weights
must be individually normalized.
Preamp Output (mV)
4
2
Length
Gap
0
Length
-2
Digitized Step 960919
-4
20
22
24
26
28
30
Time ( µs)
Figure 6.4: Digitized version of one of the x-ray steps of Figure 6.3b.
The primary differences between different digital signal processors lie
in two areas: what set of weights {wi} is used and how the regions are selected
for the computation of Equation 6-2. Thus, for example, when the weighting
values decrease with separation from the step, then the equation produces “cusplike” filters. When the weighting values are constant, one obtains triangular (if
the gap is zero) or trapezoidal filters. The concept behind cusp-like filters is
that, since the points nearest the step carry more information about its height,
they should be more strongly weighted in the averaging process. How one
chooses the filter lengths results in time variant (the lengths vary from pulse to
pulse) or time invariant (the lengths are the same for all pulses) filters.
Traditional analog filters are time invariant. The concept behind time variant
filters is that, since the x-rays arrive randomly and the lengths between them
vary accordingly, one can make maximum use of the available information by
adjusting Length on a pulse by pulse basis.
8/13/2010
69
DXP Mercury / Prospect User Manual
Prospect 1.0.x
In principal, the very best filtering is accomplished by using cusp-like
weights and time variant filter length selection. There are serious costs
associated with this approach however, both in terms of computational power
required to evaluate the sums in real time and in the complexity of the
electronics required to generate (usually from stored coefficients) normalized
{wi} sets on a pulse by pulse basis. A few such systems have been produced but
typically cost about $13K per channel and are count rate limited to about 30
Kcps. Even time invariant systems with cusp-like filters are still expensive due
to the computational power required to rapidly execute strings of multiply and
adds. One commercial system exists which can process over 100 Kcps, but it
too costs over $12K per channel.
6.2.2
Trapezoidal Filtering
The DXP processing system developed by XIA takes a different
approach because it was optimized for very high speed operation and low cost
per channel. It implements a fixed length filter with all wi values equal to unity
and in fact computes this sum afresh for each new signal value k. Thus the
equation implemented is:
L Vx , k =
–Σ
k –L – G
i = k – 2L – G + 1
vi +
Σ
k
i = k – L+ 1
vi
Equation 6-3
where the filter length is L and the gap is G. The factor L multiplying Vx,k
arises because the sum of the weights here is not normalized. Accommodating
this factor is trivial for the DXP’s host software. The operations are carried out
using hardwired logic in a field programmable gate array (FPGA) that is called
the FiPPI because is implements Filtering, Peak capture, and Pileup Inspection.
In the FiPPI, Equation 6-3 is actually implemented by noting the recursion
relationship between Vx,k and Vx,k-1, which is:
L Vx,k = L Vx,k-1+ vk - vk-L - vk-L-G + vk-2L-G
Equation 6-4
While this relationship is very simple, it is still very effective. In the
first place, this is the digital equivalent of triangular (or trapezoidal if G = 0)
filtering which is the analog industry’s standard for high rate processing. In the
second place, one can show theoretically that if the noise in the signal is white
(i.e. Gaussian distributed) above and below the step, which is typically the case
for the short shaping times used for high signal rate processing, then the average
in Equation 6-4 actually gives the best estimate of Vx in the least squares sense.
This, of course, is why triangular filtering has been preferred at high rates.
Triangular filtering with time variant filter lengths can, in principle, achieve
both somewhat superior resolution and higher throughputs but comes at the cost
of a significantly more complex circuit and a rate dependent resolution, which is
unacceptable for many types of precise analysis. In practice, XIA’s design has
been found to duplicate the energy resolution of the best analog shapers while
approximately doubling their throughput, providing experimental confirmation
of the validity of the approach.
8/13/2010
70
DXP Mercury / Prospect User Manual
Prospect 1.0.x
6.3 Trapezoidal Filtering in the DXP:
From this point onward, we will only consider trapezoidal filtering as it
is implemented in the DXP according to Equation 6-3 and Equation 6-4. The
result of applying such a filter with Length L = 20 and Gap G = 4 to the same
data set of Figure 6.4 is shown in Figure 6.5. The filter output Vx is clearly
trapezoidal in shape and has a rise-time equal to L, a flat top equal to G, and a
symmetrical fall-time equal to L. The base-width, which is a first-order measure
of the filter’s noise reduction properties, is thus 2L+G.
6.3.1
Comparing DXP Performance
This raises several important points in comparing the noise
performance of the DXP to analog filtering amplifiers. First, semi-Gaussian
filters are usually specified by a shaping time, which is roughly half of the
peaking. Their pulses typically are not symmetric so that the base-width is
about 5.6 times the shaping time or 2.8 times their peaking time. Thus a semiGaussian filter typically has a slightly better energy resolution than a triangular
filter of the same peaking time because it has a longer filtering time. This is
typically accommodated in amplifiers offering both triangular and semiGaussian filtering by stretching the triangular peaking time a bit, so that the true
triangular peaking time is typically 1.2 times the selected semi-Gaussian
peaking time. This also leads to an apparent advantage for the analog system
when its energy resolution is compared to a digital system with the same
nominal peaking time. A valid energy resolution comparison must start with
filters that have equal base-widths, and thus equal throughput, e.g. The energy
resolution of an analog system with shaping time of 1 µs should be compared to
that of a DXP with a peaking time of 2.8 µs.
6.3.2
Decimation by N means to
pre-average sequential
sums of length D = 2N.
Decimation and Peaking Time Ranges
A practical limitation on the implementation of Equation 6-4 is that two
FIFO memories are required, one of length L and one of Length L+G. Since
memory space is limited in FPGAs, we have restricted our designs to values of
L+G less than 128. The DXP Mercury samples at 50 MHz (20 ns clock period),
so this corresponds to a maximum peaking time of 2.56 µs. XIA overcomes this
limitation by first pre-averaging the data stream from the ADC by performing
sequential sums of D data points, where D = 2N. We refer to this pre-averaging
procedure as “Decimating by N”. By feeding the decimated data in an Equation
6-4 filter, we now obtain peaking times that are extended to L*D. It is important
to understand that no data are lost in this procedure, we have merely rearranged
the order of the summations represented in Equation 6-3. By extension, a
“Decimation N FiPPI” is one that decimates the data by N before applying the
energy filter. The common decimation values in the DXP Mercury are 0, 2, 4,
and 6, corresponding to averaging times of 20 ns (no averaging), 80 ns, 320 ns,
and 1.28µs, respectively.
Decimation
N
0
2
8/13/2010
ADC Clock
Period ∆t
20 ns
20 ns
#ADC Samples
in Average 2N
1
4
Decimation
Period ∆t *2N
20 ns
80 ns
Peaking Time
Range*
80 ns – 2.58µs
320 ns – 10.32 µs
71
DXP Mercury / Prospect User Manual
Prospect 1.0.x
4
20 ns
16
320 ns
1.28 µs – 41.28 µs
6
20 ns
64
1.28 µs
5.12 µs – 163.84 µs
*Experience has shown that an absolute minimum slow filter length of 4 should
be used.
Table 6.1: FiPPI decimation details.
In practice it is important to realize that implementing an energy filter
in a Decimation N FiPPI sets certain limitations on the flat-top lengths that can
be obtained in trapezoidal filters. Because the decimation process is
uncorrelated with the arrival of x-rays, the gap G must be 3 or greater to assure
that the filter’s peak truly represents the x-ray’s energy. Therefore, the
minimum Decimation N gap time is G*2N*∆t, where ∆t is the ADC’s sampling
interval. With the DXP Mercury’s ∆t = 20 ns sampling interval, for instance,
the smallest useful flat-top in Decimation 6 is 3*1.28 µs = 3.84 µs.
Given the significant overlap in peaking time ranges, it is generally
better to choose a lower decimation value, such that a shorter gap time can be
used. Decimation 0 has other limitations, i.e. no intermediate baseline filter, and
is thus an exception to this rule. The FDD firmware file defines the actual, i.e.
non-overlapping, peaking time ranges used.
6.3.3
Time Domain Benefits of Trapezoids
One extremely important characteristic of a digitally shaped trapezoidal
pulse is its extremely sharp termination on completion of the base-width 2L+G.
This may be compared to analog filtered pulses which have tails which may
persist up to 40% of the peaking time, a phenomenon due to the finite bandwidth
of the analog filter. As we shall see below, this sharp termination gives the
digital filter a definite rate advantage in pileup free throughput.
8/13/2010
72
DXP Mercury / Prospect User Manual
Prospect 1.0.x
6
Filtered Step S.kfig 960920
Output (mV)
4
2
0
L
L+G/2
2L+G
-2
Preamp Output (mV)
Filter Output (mV)
-4
24
26
28
Time ( µs)
30
32
Figure 6.5: Trapezoidal filtering the Preamp Output data of Figure 6.4 with
L = 20 and G = 4.
6.4 Baseline Issues:
6.4.1
The Need for Baseline Averaging
Figure 6.6 shows the same event as is Figure 6.5 but over a longer time
interval to show how the filter treats the preamplifier noise in regions when no
x-ray pulses are present. As may be seen, the effect of the filter is both to
reduce the amplitude of the fluctuations and reduce their high frequency content.
This signal is termed the baseline because it establishes the reference level or
offset from which the x-ray peak amplitude Vx is to be measured. The
fluctuations in the baseline have a standard deviation σe which is referred to as
the electronic noise of the system, a number which depends on the peaking time
of the filter used. Riding on top of this noise, the x-ray peaks contribute an
additional noise term, the Fano noise, which arises from statistical fluctuations
in the amount of charge Qx produced when the x-ray is absorbed in the detector.
This Fano noise σf adds in quadrature with the electronic noise, so that the total
noise σt in measuring Vx is found from
σt = sqrt( σf2 + σe2)
Equation 6-5
8/13/2010
73
DXP Mercury / Prospect User Manual
Prospect 1.0.x
6
Filtered Step L.kfig 960920
σt
Output (mV)
4
Vx
2
σe
0
-2
Filter Output (mV)
-4
5
10
15
20
25
30
Time ( µs)
35
40
45
Figure 6.6: The event of Figure 6.5 displayed over a longer time period to
show baseline noise.
The Fano noise is only a property of the detector material. The
electronic noise, on the other hand, may have contributions from both the
preamplifier and the amplifier. When the preamplifier and amplifier are both
well designed and well matched, however, the amplifier’s noise contribution
should be essentially negligible. Achieving this in the mixed analog-digital
environment of a digital pulse processor is a non-trivial task, however.
In the general case, the mean baseline value is not zero. This situation
arises whenever the slope of the preamplifier signal is not zero between x-ray
pulses. This can be seen from Equation 6-3. When the slope is not zero, the
mean values of the two sums will differ because they are taken over regions
separated in time by L+G, on average. Such non-zero slopes can arise from
various causes, of which the most common is detector leakage current.
When the mean baseline value is not zero, it must be determined and
subtracted from measured peak values in order to determine Vx values
accurately. If the error introduced by this subtraction is not to significantly
increase σt, then the error in the baseline estimate σb must be small compared to
σe. Because the error in a single baseline measurement is σe, by definition, this
means that multiple baseline measurements will have to be averaged. This
number, NB is the Baseline Average. For example, if NB = 128 measurements
are averaged then the total noise will be as shown in Equation 6-6.
8/13/2010
74
DXP Mercury / Prospect User Manual
Prospect 1.0.x
σt = sqrt( σf2 + (1+1/128)σe2)
Equation 6-6
This results in less than 0.5 eV degradation in resolution, even for very
long peaking times, when resolutions of order 130 eV are obtained.
6.4.2
Applying a Baseline Cut
can improve performance
when the Baseline
Histogram is nonGaussian. Outlying data
points are‘cut’ from the
running Baseline Average
(though still included in the
histogram)
Raw Baseline Measurement
The output of the energy filter (or a derivative of the energy filter, the
intermediate filter) is sampled periodically in the explicit absence of an x-ray
step, defined by a baseline threshold. In practice, the DXP initially makes a
series of NB baseline measurements to compute a starting baseline mean. It then
makes additional baseline measurements at quasi-periodic intervals to keep the
estimate up to date. These values are stored internally and can be read out to
construct a spectrum of baseline noise, referred to as the Baseline Histogram.
This is recommended because of its excellent diagnostic properties. When all
components in the spectrometer system are working properly, the baseline
spectrum should be Gaussian in shape with a standard deviation reflecting σn.
Deviations from this shape indicate various pathological conditions which may
also cause the x-ray spectrum to be distorted and therefore have to be fixed.
The situation is remedied by removing (“cutting”) outlying samples
from the baseline average described below. If the maximum in the baseline
distribution lies at E0, then captured baseline values that deviate from E0 by
more than ∆E+ and ∆E-, respectively, are not included in the running baseline
average. Note that all captured baseline values are included in the Baseline
Histogram, however, so that it is always a valid representation of the system’s
behavior.
6.4.3
Baseline Average Settings and Recommendations
A FIR running average of baseline measurements is computed, which is
then subtracted from sampled peak values to compute the energy of
corresponding incident x-rays. The number of baseline samples averaged is set
in ProSpect as “Baselines Average Samples”. In the DSP this is converted into
the parameter BLAVGDIV according to the equation:
# baseline samples averaged = 2(BLAVGDIV +1)
Decimation
# Baseline Samples to
Average
BLAVGDIV
(DSP Parameter)
0
2
4
6
64
128
256
256
5
6
7
7
Table 6.2:
8/13/2010
Typical values used for baseline averaging. The best value for
each decimation should be determined empirically, though the
general trend illustrated in the table, i.e. larger number to average
for higher decimations, should be followed.
75
DXP Mercury / Prospect User Manual
Prospect 1.0.x
6.4.4
Why Use a Finite Averaging Length?
Physically, the baseline is a measure of the instantaneous slope
(volts/sec) for a pulsed-reset detector, and a measure of the DC offset for an RCfeedback preamplifier. The variation in leakage current of the detector and
offset drift and 1/f noise of the preamplifier often contribute to a baseline with
significant low-frequency (i.e. relative to the energy filter cutoff) noise. These
variations pass through the energy filter, and thus should also pass through the
baseline averaging stage to achieve good cancellation when the baseline average
is subtracted from the energy filter sample. The goal is to produce a baseline
average that has a sufficient number of samples to average out the high
frequency noise, but which still reflects the ‘local’ instantaneous baseline upon
which the x-ray step ‘rides’. Generally speaking, the number of baseline
samples in the average is set to achieve the best energy resolution performance
over the desired range of input count rate. There are two considerations worth
emphasizing:
1.
Excess detector/preamplifier noise and pickup (all decimations): The values
in the table above implicitly assume a flat noise spectrum from the
preamplifier. A high-frequency noise peak can result in poor relative
performance at the corresponding ‘resonant’ peaking time. Often this
problem can be mediated, though not eliminated, by increasing the number
of baseline samples in the average for the affected peaking times. On the
other hand, excess low-frequency noise, i.e. wandering, can be remedied by
reducing the number baseline samples in the average.
2.
High rate performance (decimation 0): At higher rates, i.e. > 50%
deadtime, the slow filter returns less and less often to baseline, thus the time
between baseline samples grows longer. This is the primary cause of
degraded energy resolution at high rates. Decimation 2,4 and 6 firmware
now employs a proprietary circuit that virtually eliminates this problem,
resulting in industry-leading count rate stability. This improvement cannot
however be implemented in the decimation 0 firmware. The resolution can
nonetheless be improved in most cases by reducing the number of baseline
samples in the average.
6.5 X-ray Detection & Threshold Setting:
Before capturing a value of Vx we must first detect the x-ray. X-ray
steps (in the preamp output) are detected by digitally comparing the output of a
trapezoidal filter to a threshold.
In the DXP up to three trapezoidal filters are implemented: fast,
intermediate and slow; each with a threshold that can be individually enabled or
disabled. A fast filter very quickly detects larger x-ray steps. A slow (energy)
filter averages out the most noise and can thus detect smaller x-ray steps, but has
a response that is much slower. An intermediate filter (used in decimations 2, 4
and 6 only) is a derivative of the slow filter that provides a balance between the
speed of the fast filter and the noise reduction of the slow filter.
The fast filter is used solely for x-ray detection, i.e. a threshold crossing
initiates event processing. Its short base-width (2L+G) means that successive
pulses that would ‘pile-up’ in the slow filter can be resolved in the fast filter and
rejected from the spectrum (see Figure 6.11 below). Conversely, little noise
8/13/2010
76
DXP Mercury / Prospect User Manual
Prospect 1.0.x
reduction is achieved in the fast filter, thus the fast threshold cannot be set to
detect particularly low x-ray energies.
The intermediate filter is used for reset-type preamplifiers, in
decimations 2, 4 and 6 only. Its threshold is automatically set by the DSP and
applied as part of the baseline acquisition circuitry, i.e. baseline measurements
are taken when the signal is below this threshold. Intermediate threshold
crossings by default also trigger event processing, extending the detectable
energy range significantly below the fast filter threshold.
After an x-ray has been detected, the step height is measured at the
slow filter output. The slow filter’s excellent noise reduction also allows for
detection of the very lowest energy x-rays however its slow response precludes
accuracy both in the determination of pulse pileup and the measurement of
deadtime. The intermediate filter, which does not suffer this loss of accuracy,
typically provides sufficient low energy detection. When present the
intermediate threshold is enabled by default, and should be used in most cases.
The slow threshold should be used cautiously, and only at low rates.
6.6 Peak Capture Methods
As noted above, we wish to capture a value of Vx for each x-ray
detected and use these values to construct a spectrum. This process is also
significantly different between digital and analog systems. In the analog system
the peak value must be “captured” into an analog storage device, usually a
capacitor, and “held” until it is digitized. Then the digital value is used to
update a memory location to build the desired spectrum. During this analog to
digital conversion process the system is dead to other events, which can severely
reduce system throughput. Even single channel analyzer systems introduce
significant deadtime at this stage since they must wait some period (typically a
few microseconds) to determine whether or not the window condition is
satisfied.
Digital systems are much more efficient in this regard, since the values
output by the filter are already digital values. All that is required is to capture
the peak value – it is immediately ready to be added to the spectrum. If the
addition process can be done in less than one peaking time, which is usually
trivial digitally, then no system deadtime is produced by the capture and store
operation. This is a significant source of the enhanced throughput found in
digital systems.
Once an active threshold is exceeded, the DXP Mercury employs one
of two methods to capture the slow energy filter output such that the best
measure of Vx results. In the first method the slow filter output is monitored
over a finite interval of time in the region of its maximum, and the maximum
value within that interval is captured. This method is referred to as “peak
finding” or “peak sensing”. Alternatively the slow filter can be sampled at a
fixed time interval after the pulse is detected by the fast filter. This method is
referred to as “peak sampling”. There is a panel in the Acquisition/Edit Filter
Parameters tab to allow selection of either method.
After describing in section 6.6.1 below how to set the Gap parameter
so that there will be a quality value of the energy filter to capture, we describe
the two methods in detail in section 6.6.2.
8/13/2010
77
DXP Mercury / Prospect User Manual
Prospect 1.0.x
6.6.1
Setting the Gap Length
When starting with a new detector, it is important first to set
SLOWGAP to a minimum of 3 decimated clock cycles, and at least one
decimated clock cycle greater than the entire preamplifier rise-time, per section
4.7.1.2. (See Table 6.3).
Decimation
# ADC
Samples
averaged
Decimated
Clock
frequency
Decimated
Clock cycle
interval
Minimum Gap
Time
0
2
4
6
1
4
16
64
50 MHz
12.5 MHz
3.125 MHz
781.25 kHz
20 ns
80 ns
320 ns
1.28 µs
60 ns
240 ns
960 ns
3.84 µs
Table 6.3: Minimum Gap Length for each Decimation value.
For example, consider a preamplifier with a pulse rise-time of 240 ns. For
decimations 2, 4 and 6 SLOWGAP would be set to 3 or greater. For decimation
0 SLOWGAP would be set to 12 or greater. ProSpect will select these values
automatically if you enter a Minimum Gap Time of -240 ns. Therefore, at longer
peaking times (higher Decimation), the Actual Gap Time (displayed in the Edit
Filter Paramters panel) will always be at least 3 decimated clock cycles and may
be longer than the Minimum Gap Time set by the user.
6.6.2
Peak Sampling vs. Peak Finding
The figures below illustrate the two peak capture methods. Under the
'peak finding method' the slow filter output is monitored over a finite interval of
time, and the maximum value within that interval is selected. The interval is set
automatically, solely based on the values of the DXP parameters SLOWLEN
and PEAKINT. SLOWLEN and PEAKINT are both automatically derived from
the peaking time value selected in ProSpect and should normally not be adjusted
by the user. PEAKINT is also a pileup inspection parameter, as will be
discussed in further detail in section 6.8.
8/13/2010
78
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 6.7: Peak finding method: The slow filter output is monitored and the peak value is selected.
In the 'peak sampling' method, the slow filter output is instead sampled
a fixed time after the x-ray is detected. An additional ‘Peak Sampling’ timer is
started when an x-ray step is detected which expires after PEAKSAM decimated
clock cycles. PEAKSAM must be less than PEAKINT, and should typically be
set such that the sample point lies in the ‘flat-top’ region of the slow filter
output:
SLOWLEN ≤ PEAKSAM ≤ SLOWLEN + SLOWGAP
Equation 6-7
The precise PEAKSAM setting has a strong effect on energy resolution
and should be determined empirically for each new detector. More on this
below...
8/13/2010
79
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 6.8: Peak sampling method: The slow filter output is sampled a fixed time after the x-ray is
detected. PEAKSAM must be set properly to achieve optimum performance.
In our experience values at the low end (i.e. PEAKSAM ~ SLOWLEN)
tend to work better. We recommend that you record the initial value of
PEAKSAM and then change it in steps of 1, working out from the initial value.
Certain PEAKSAM values may cause the DXP Mercury to crash. Do not be
alarmed, just restart and be sure to enter a valid PEAKSAM value before
proceeding. Making a plot of energy resolution versus PEAKSAM will indicate
the best value to select.
This determination need only be done for one peaking time per
decimation. The result can then be applied to any value of SLOWLEN and
SLOWGAP using the following recipe:
PEAKSAM = (SLOWLEN + SLOWGAP) – X
Equation 6-8
6.7 Energy Measurement with Resistive Feedback Preamplifiers
In previous sections, the pulse height measurement was shown for the
case of reset-type preamplifiers. The reset-type scheme is most often used for
optimum energy resolution x-ray detectors. Other detectors use an RC-type
preamplifier, as described in section 6.1.2. Resistive feedback is most often
used for gamma-ray detectors which cover a larger dynamic range and where the
electronic noise is not as significant a contribution to energy resolution.
8/13/2010
80
DXP Mercury / Prospect User Manual
Prospect 1.0.x
A
VD
Ve
L∆ t
V0
L∆ t
t1
t2
G∆ t
tD
t
0
Figure 6.9: RC preamplifier output voltage. An x-ray step of amplitude A
occurs at time t=0.
Where analog shaping amplifiers typically have a “pole-zero”
adjustment to cancel out the exponential decay, the DXP uses a patented digital
correction to achieve good energy resolution without a pole-zero stage. Figure
6.9 and Figure 6.10 illustrate the method used. The first shows the output
voltage of a RC feedback preamplifier with a x-ray or γ-ray step of amplitude A
appearing at t=0. Ve is the voltage just before the step pulse arrives and V0 is
the asymptotic value that the signal would decay to in the absence of steps. t1 is
the earliest time used in the slow filter, L and G are the length and gap of the
trapezoidal filter in clock units, and ∆t is the clock period, In addition to the
normal slow filter measurement of the step height, the ADC amplitude, VD is
made at time tD. In the following discussion, it is assumed that the signal risetime is negligible.
8/13/2010
81
DXP Mercury / Prospect User Manual
Prospect 1.0.x
10000
ICR = 41 kcps
8000
Zr Kα
Step
6000
Zr Kβ
4000
2000
Ge escape
peaks
0
Noise
-2000
0
4000
8000
Amplitude
12000
Figure 6.10: Correlation between step size and amplitude for Zr Kα x-ray
events measured with the DXP-4C.
As Figure 6.10 makes clear, there is a linear correlation between the
step height from the trapezoidal filter and the ADC amplitude, for pulses of a
given energy. This is due to the fact that the exponential decay causes a deficit
in the measured step height, which grows linearly with the distance from the
asymptotic ADC offset at zero count rate.
The DSP reads these two values for each event that passes the FiPPI’s
trigger criteria, and makes a correction of the form:
E = k1 ( SX + k2 VX - < SB + k2 VB > )
Equation 6-9
Here the quantities SX and VX are the step height and ADC amplitude
measured for the step, and the corresponding values with the B subscript are
“baseline” values, which are measured frequently at times when there is no
trigger. The brackets <> indicate that the baseline values are averaged over a
large enough number of events to not introduce additional noise in the
measurement. The constant k2 (the DSP parameter called RCFCOR) is inversely
proportional to the exponential decay time; this correction factor is a constant
for a detector channel at a fixed gain and shaping time. The constant k1 is
effectively a gain factor, and is taken into account with a detector gain
calibration.
The parameter RCFCOR is a function of the digital filter parameters
(SLOWLEN, SLOWGAP and DECIMATION) and the preamplifier decay time
(the DSP parameter RCTAU). The decay time defined by RCTAU (and
fractional word RCTAUFRAC) has 50 ns granularity, and is measured and
8/13/2010
82
DXP Mercury / Prospect User Manual
Prospect 1.0.x
entered by the user. At the start of an acquisition run, the DSP calculates
RCFCOR using the following approximate expression:
RCFCOR = 2DEC * (LEN + GAP) / (RCTAU – (LEN + GAP/2 + 3)*2DEC)
Equation 6-10
The above expression is valid for peaking times less than about
RCTAU/2. Alternatively, RCFCOR can be determined empirically in a special
test run from a linear fit of data, as in Figure 6.10.
6.8 Pile-up Inspection:
The captured value Vx (see Figure 6.6) will only be a valid measure of
its associated x-ray’s energy provided that its filtered pulse is sufficiently well
separated in time from its preceding and succeeding neighbor pulses so that its
peak amplitude is not distorted by the action of the trapezoidal filter on those
neighbor pulses. That is, if the pulse is not piled up. The relevant issues may be
understood by reference to Figure 6.11, which shows 5 x-rays arriving separated
by various intervals.
Because the triangular filter is a linear filter, its output for a series of
pulses is the linear sum of its outputs for the individual members in the series.
In Figure 6.11 the pulses are separated by intervals of 3.2, 1.8, 5.7, and 0.7 µs,
respectively. The fast filter has a peaking time of 0.4 µs with no gap. The slow
filter has a peaking time of 2.0 µs with a gap of 0.4 µs.
The first kind of pileup is slow pileup, which refers to pileup in the
slow channel. This occurs when the rising (or falling) edge of one pulse lies
under the peak (specifically the sampling point) of its neighbor. Thus peaks 1
and 2 are sufficiently well separated so that the leading edge (point 2a) of peak 2
falls after the peak of pulse 1. Because the trapezoidal filter function is
symmetrical, this also means that pulse 1’s trailing edge (point 1c) also does not
fall under the peak of pulse 2. For this to be true, the two pulses must be
separated by at least an interval of L + G/2. Peaks 2 and 3, which are separated
by only 1.8 µs, are thus seen to pileup in the present example with a 2.0 µs
peaking time.
This leads to an important first point: whether pulses suffer slow pileup
depends critically on the peaking time of the filter being used. The amount of
pileup which occurs at a given average signal rate will increase with longer
peaking times. We will quantify this in section 0, where we discuss throughput.
Because the fast filter peaking time is only 0.4 µs, these x-ray pulses do
not pileup in the fast filter channel. The DXP can therefore test for slow channel
pileup by measuring for the interval PEAKINT after a pulse arrival time. If no
second pulse occurs in this interval, then there is no trailing edge pileup.
PEAKINT is usually set to a value close to L + G/2 + 1. Pulse 1 passes this test,
as shown in the figure. Pulse 2, however, fails the PEAKINT test because pulse
3 follows in 1.8 µs, which is less than PEAKINT = 2.3 µs. Notice, by the
symmetry of the trapezoidal filter, if pulse 2 is rejected because of pulse 3, then
pulse 3 is similarly rejected because of pulse 2.
Pulses 4 and 5 are so close together that the output of the fast filter does
not fall below the threshold between them and so they are detected by the pulse
detector as only being a single x-ray pulse. Indeed, only a single (though
8/13/2010
83
DXP Mercury / Prospect User Manual
Prospect 1.0.x
somewhat distorted) pulse emerges from the slow filter, but its peak amplitude
corresponds to the energy of neither x-ray 4 nor x-ray 5. In order to reject as
many of these fast channel pileup cases as possible, the DXP implements a fast
channel pileup inspection test as well.
The fast channel pileup test is based on the observation that, to the
extent that the rise-time of the preamplifier pulses is independent of the x-rays’
energies (which is generally the case in x-ray work except for some room
temperature, compound semiconductor detectors) the base-width of the fast
digital filter (i.e. 2Lf + Gf) will also be energy independent and will never
exceed some maximum width MAXWIDTH. Thus, if the width of the fast filter
output pulses is measured at threshold and found to exceed MAXWIDTH, then
fast channel pileup must have occurred. This is shown graphically in the figure
where pulse 3 passes the MAXWIDTH test, while the piled up pair of pulses 4
and 5 fail the MAXWIDTH test.
Thus, in Figure 6.11, only pulse 1 passes both pileup inspection tests
and, indeed, it is the only pulse to have a well defined flattop region at time
PEAKSAMP in the slow filter output.
8/13/2010
84
DXP Mercury / Prospect User Manual
25
Prospect 1.0.x
Digitized MultiPile kfig 960921
Preamp
4 5
20
Passes
PEAKINT
Test
2
1
3
Fails
PEAKINT
Test
Passes
MAXWIDTH
Test
15
Fails
MAXWIDTH
Test
Fast Filter
10
2
1
4
3
5
PEAKSAMP
1b
5
1a
2a 1c
Slow Filter
0
5
10
15
20
25
30
Time ( µs)
Figure 6.11: A sequence of 5 x-ray pulses separated by various intervals to show the origin of both slow
channel and fast channel pileup and demonstrate how the two cases are detected by the DXP.
Note that PEAKINT and MAXWIDTH are both DSP parameters and
are normally set automatically. In particular, there is almost never any benefit to
a longer value of PEAKINT than the standard value as it does not improve
energy resolution and only decreases throughput for a given input rate. Please
see section 4.6.2.1 for details on how to adjust MAXWIDTH.
6.9 Input Count Rate (ICR) and Output Count Rate (OCR):
During data acquisition, x-rays will be absorbed in the detector at some
rate. This is the true input count rate , which we will refer to as ICRt. Because
of fast channel pileup, not all of these will be detected by the DXP’s x-ray pulse
detection circuitry, which will thus report a measured input count rate ICRm,
which will be less than ICRt. This phenomenon, it should be noted, is a
8/13/2010
85
DXP Mercury / Prospect User Manual
Prospect 1.0.x
characteristic of all x-ray detection circuits, whether analog or digital, and is not
specific to the DXP.
Of the detected x-rays, some fraction will also satisfy both fast and
slow channel pileup tests and have their values of Vx captured and placed into
the spectrum. This number is the output count rate, which we refer to as the
OCR. The DXP normally returns, in addition to the collected spectrum, the
REALTIME for which data was collected, the fast channel LIVETIME for
which the fast channel was below threshold (and thus ready to detect a
subsequent x-ray) together with the number FASTPEAKS of fast peaks detected
and the number of Vx captured events EVTSINRUN. From these values, both
the OCR and ICRm can be computed according to Equation 6-11. These values
can then be used to make deadtime corrections as discussed in section 6.11.
ICRm = FASTPEAKS/LIVETIME; OCR = EVTSINRUN/REALTIME
Equation 6-11
Note: The fast channel LIVETIME should only be used to determine
the input count rate according to Equation 6-11. Specifically, it is NOT related
to the energy filter livetime and should not be interpreted as the inverse of the
processor deadtime. The DSP does calculate the energy filter livetime
ELIVETIME, however, it is only an approximation. The most accurate
deadtime measurement is obtained from ICRm and OCR in Equation 6-11, as
discussed in section 6.11.
6.10 Throughput:
Figure 6.12 shows how the values of ICRm and OCR vary with true
input count rate for the DXP and compare these results to those from a common
analog shaping amplifier plus SCA system. The data were taken at a
synchrotron source using a detector looking at a CuO target illuminated by xrays slightly above the Cu K absorption edge. Intensity was varied by adjusting
two pairs of crossed slits in front of the input x-ray beam so that the harmonic
content of the x-ray beam striking the detector remained constant with varying
intensity.
8/13/2010
86
DXP Mercury / Prospect User Manual
Prospect 1.0.x
200
DXP OCR
DXP ICRm
Analog OCR
Analog ICRm
150
Output Count Rate (kcps)
NOTE: The DXP’s peaking
time is twice as long as the
analog system peaking time
in this comparison, and yet
the throughput is nearly the
same.
True ICRt
100
50
ICR/OCR Plot kfig 960922
0
0
50
100
150
200
Input Count Rate (kcps)
Figure 6.12: Curves of ICRm and OCR for the DXP using 2 µs peaking time,
compared to a common analog SCA system using 1 µs peaking
time.
System
OCR Deadtime (µs)
ICR Deadtime (µs)
DXP (2 µs τp, 0.6 µs τg)
4.73
0.83
Analog Triangular Filter Amp (τp = 1 µs)
4.47
0.40
Table 6.4:
Comparing the deadtime per event for the DXP and an analog shaping amplifier. Notice that
that the DXP produces a comparable output count rate even though its peaking time is nearly
twice as long.
Functionally, the OCR in both cases is seen to initially rise with
increasing ICR and then saturate at higher ICR levels. The theoretical form,
from Poisson statistics, for a channel which suffers from paralyzable (extending)
dead time is given by:
OCR = ICRt * exp( - ICRt * τd ),
Equation 6-12
where τd is the dead time. Both the DXP and analog systems’ OCRs are so
describable, with the slow channel dead times - τd - shown in Table 6.4. The
measured ICRm values for both the DXP and analog systems are similarly
describable, with the fast channel dead times - τdf - as shown. The maximum
value of OCR can be found by differentiating Equation 6-12 and setting the
result to zero. This occurs when the value of the exponent is -1, i.e. when ICRt
equals 1/τd. At this point, the maximum OCRmax is 1/e multiplied by the ICR,
or:
8/13/2010
87
DXP Mercury / Prospect User Manual
Prospect 1.0.x
OCRmax = 1/(e τd) = 0.37/τd
Equation 6-13
These are general results and are very useful for estimating experimental data
rates.
Table 6.4 illustrates a very important result for using the DXP: the slow
channel deadtime is nearly the minimum value that is theoretically possible,
namely the pulse base-width. For the shown example, the base-width is 4.6 µs
(2Ls + Gs) while the deadtime is 4.73 µs. The slight increase is because, as
noted above, PEAKINT is always set slightly longer than Ls - Gs/2 to assure
that pileup does not distort collected values of Vx.
The deadtime for the analog system, on the other hand is much larger.
In fact, as shown, the throughput for the digital system is almost twice as high,
since it attains the same throughput for a 2 µs peaking time as the analog system
achieves for a 1 µs peaking time. The slower analog rate arises, as noted earlier
both from the longer tails on the pulses from the analog triangular filter and on
additional deadtime introduced by the operation of the SCA. In spectroscopy
applications where the system can be profitably run at close to maximum
throughput, then, a single DXP channel will then effectively count as rapidly as
two analog channels.
6.11 Dead Time Corrections:
The fact that both OCR and ICRm are describable by Equation 6-12
makes it possible to correct DXP spectra quite accurately for deadtime effects.
Because deadtime losses are energy independent, the measured counts Nmi in
any spectral channel i are related to the true number Nti which would have been
collected in the same channel i in the absence of deadtime effects by:
Nti = Nmi ICRt/OCR
Equation 6-14
Looking at Figure 6.12, it is clear that a first order correction can be
made by using ICRm of Equation 6-11 instead of ICRt, particularly for OCR
values less than about 50% of the maximum OCR value. For a more accurate
correction, the fast channel deadtime τdf should be measured from a fit to the
equation:
ICRm = ICTt * exp( - ICRt τdf )
Equation 6-15
Then, for each recorded spectrum, the associated value of ICRm is
noted and Equation 6-15 inverted (there are simple numerical routines to do this
for transcendental equations) to obtain ICRt. Then the spectrum can be
corrected on a channel by channel basis using Equation 6-12. In experiments
with a DXP prototype, we found that, for a 4 µs peaking time (for which the
maximum ICR is 125 kcps), we could correct the area of a reference peak to
better than 0.5% between 1 and 120 kcps.
8/13/2010
88
DXP Mercury / Prospect User Manual
Prospect 1.0.x
7 DXP Mercury Hardware Description
7.1 DXP Mercury Overview
The DXP Mercury system, shown in Figure 7.1, consists of a Digital
X-ray Processor (DXP) channel (4 channels for the Mercury-4), a Digital Signal
Processor (DSP), a System FPGA, SRAM memory and a PCI interface. Each
DXP channels accepts a preamplified signal input and produces a 16-bit
pipelined output stream of x-ray energies.
Figure 7.1: Block diagram of the Mercury system architecture.
7.1.1
The Digital X-ray Processor (DXP)
The Digital X-ray Processor (DXP) is a proprietary architecture (see
Figure 7.2) designed to rapidly measure the pulse-heights of voltage steps.
Shown below are the three major DXP operating blocks: the Analog Signal
Conditioner (ASC), an analog-to-digital converter (ADC), and the Filter, Peak
Detector, and Pileup Inspector (FiPPI).
The DXP Mercury contains four independent DXP Digital X-ray
Processor (DXP) channels. Each DXP channel accepts a detector preamplifier
signal input and produces a FIFO-buffered output stream of 16-bit pulse-height,
i.e. energy, measurements. Each DXP also keeps track of its run statistics,
including livetime and the number of input counts detected. See sections 7.3 –
7.8 below for a thorough description of the DXP.
8/13/2010
89
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 7.2: Block diagram of the DXP architecture.
7.1.2
Rapid Data Readout
The DXP Mercury was designed for the rapid readout of acquired data
in a multi-element detector system. Some important changes to the architecture,
relative to other DXP products, facilitate this goal:
1) Tasks previously handled by the DSP, such as baseline handling and
ASC control, have been offloaded to programmable logic in the FiPPIs.
This frees the DSP to focus on event handling during data acquisition.
2) Memory management, previously handled by the DSP, is now tasked to
programmable logic in the System FPGA. This reduces the size of the
DSP's event handling loop.
3) The storage of MCA, SCA and statistics data in external SRAM
memory. Previously this data was stored in DSP internal memory.
8/13/2010
90
DXP Mercury / Prospect User Manual
Prospect 1.0.x
7.2 Timing and Synchronization Logic
7.2.1
GATE Function: MCA Mode
The GATE input allows real-time user control over data acquisition
during a normal MCA data acquisition run. Data acquisition is halted when the
GATE signal is asserted, i.e. incident events are not processed, and the real-time
and live-time counters are disabled. The assert-polarity of the GATE signal can
be set via software. The Real-time can be set to increment during GATE
assertion via software. The GATE signal can be ignored entirely via software.
7.2.1.1 GATE Polarity
The interpretation of the user-provided GATE signal can be inverted in
the hardware such that data acquisition is halted when the signal is HI or LO.
The GATE polarity corresponds to the input_logic_polarity setting in the INI
file.
7.2.1.2 Ignore GATE
This is a software-only setting. Checking the “Ignore GATE”
checkbox disables the GATE logic: Data acquisition occurs irrespective of the
GATE signal level.
7.2.1.3 GATE Real Time
This is a hardware-level setting that allows for incrementing only the
real time counter when the GATE is asserted; data acquisition and other
statistics are halted. This setting corresponds to the gate_mode setting in the
INI file. The default setting gate_mode = 0 halts everything. With gate_mode
= 1, the real time increments.
7.2.2
GATE Function: Mapping Mode
In Mapping mode multiple spectra are generated as an x-ray beam is
rastered across the sample. Each spectrum corresponds to a pixel. The 'pixel
advance' controls when the Mercury changes from one pixel/spectrum to the
next. The simplest pixel advance is implemented with the GATE signal.
7.2.2.1 Pixel Advance on GATE Edge
The recommended method for advancing the pixel number is to use the
GATE input, where the pixel number advances on every trailing edge of the
signal (the transition from active data acquisition to the inactive state). By
default the GATE signal halts data acquisition when it is LO and the pixel
advances on every falling edge (see Figure 7.3).
8/13/2010
91
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure 7.3: Mapping mode acquisition using the GATE input with default polarity. The pixel advance
occurs on each falling edge of GATE and data acquisition is halted until the next rising edge.
7.2.2.2 GATE Polarity
The interpretation of the user-provided GATE signal can be inverted
such that data acquisition is halted when the signal is HI and the pixel advance
occurs on rising edges.
Figure 7.4: Mapping mode acquisition using the GATE input with inverted polarity. The pixel advance
occurs on each rising edge of GATE and data acquisition is halted until the next falling edge.
7.2.2.3 GATE Ignore Setting
Normally the GATE signal is used both to advance the pixel and to halt
data taking. In some cases the user may prefer continuous operation through the
pixel advance operations. With the GATE Ignore option selected, the pixel
advance occurs but data acquisition is not halted. Note that the data acquisition
is halted at the beginning of the run until GATE is released the first time, i.e. run
start synchronization is still available in this mode.
Figure 7.5: Mapping mode acquisition using the GATE input with default polarity, and with the GATE
Ignore option selected. The pixel advance occurs on each falling edge of GATE but data
acquisition runs continuously with no pause.
8/13/2010
92
DXP Mercury / Prospect User Manual
7.2.3
Prospect 1.0.x
SYNC Function: Mapping Mode
In Mapping mode multiple spectra are generated as an x-ray beam is
rastered across the sample. Each spectrum corresponds to a pixel. The so-called
'pixel advance' controls when the Mercury changes from one pixel/spectrum to
the next. A more complex pixel advance is implemented using the SYNC input.
7.2.3.1 Pixel Advance using SYNC Clock
The user provides a clock signal to the SYNC input(s). Using this
method, the pixel will advance for every N LO-to-HI transitions on the SYNC
line, where N can be set from 1 to 65535. Note that the pulse widths must be
greater than 40 ns in duration.
7.2.3.2 Synchronous Starts with SYNC
It is especially important to synchronize the beginning of a run when
using the SYNC pixel advance. Two methods are supported: The first edge
detected on SYNC itself can be used (see Figure 7.6); alternatively GATE can
be used as before (see Figure 7.7).
Figure 7.6: Mapping mode acquisition using the SYNC input with N = 8. The pixel advance occurs every
8 SYNC cycles. In this example GATE is not used, and SYNC itself is used to synchronize
the run start: Data acquisition does not begin until the first rising edge is detected.
Figure 7.7: Mapping mode acquisition using the SYNC input with N = 8. The pixel advance occurs every
8 SYNC cycles. Note that GATE is used to achieve a system-wide synchronous start as in
previous figures.
8/13/2010
93
DXP Mercury / Prospect User Manual
Prospect 1.0.x
7.3 The Analog Signal Conditioner (ASC):
The ASC employs a patented ramp-subtraction technique to compress
the dynamic range of the signal without adding significant noise or distortion to
the signal. The typical preamplifier output signal has dynamic range of 100dB,
or 16-bits. The compression produces a signal that can be fully digitized by a
relatively inexpensive 14-bit ADC (with effectively 13-bits of resolution, due to
integral and differential non-linearities). Importantly, the bandwidth of the
signal is unaffected by the compression, limited only by the Nyquist criterion for
the ADC. The digital filters thus operate on an wideband 20MHz signal.
The technique is illustrated in Figure 7.8. Here a resetting preamplifier
output is shown which cycles between about -3.0 and -0.5 volts. We observe
that it is not the large-signal ramp function which is of interest, but rather the
individual steps, such as shown in Figure 6.1, that carry the x-ray amplitude
information. Thus, if we generate a sawtooth function which has the same
average slope and subtract this sawtooth from the preamplifier signal, we can
amplify the difference signal to match the ADC’s input range. The generator
required to produce this sawtooth function is quite simple, comprising a current
integrator with an adjustable offset. The current, which sets the slope, is
controlled by a DAC (with DSP parameter SLOPEDAC). The DAC is
automatically adjusted according the input rate to maintain the ASC output (i.e.
the “Amplified Sawtooth Subtracted Data” of Figure 7.8) within the ADC input
range.
A side-effect of this approach is that fluctuations in data arrival rate
will cause the subtracted signal to pass outside the ADC input range while the
preamplifier output is in its operating range. These signal drifts out-of-range
high or low have been termed DRIFTUPs and DRIFTDOWNs, respectively.
The out-of-range condition is corrected by auto-zeroing the integrator, and thus
bringing the signal to the middle of the ADC input range. Preamplifier resets
are handled similarly, though the auto-zero time is extended to allow the
preamplifier reset transient to settle. The preamplifier settling time (DSP
parameter RESETINT) is thus an important DXP setting.
Any time the ADC is out of range, dead time is incurred. The ASC
compression technique thus incurs additional deadtime, although this is typically
much smaller than the deadtime caused by preamplifier resets.
Note: When viewing the ADC trace in oscilloscope mode, it is easy to mistake
signal drift corrections for actual preamplifier resets.
8/13/2010
94
DXP Mercury / Prospect User Manual
Prospect 1.0.x
ADC Max Input
3.0
2.0
Amplified Sawtooth Subtracted Data
Preamp Output (V)
1.0
ADC Min Input
Preamp
Output
0.0
-1.0
-2.0
Reset
Level
-3.0
0
1
Sawtooth
Function
Preamp-Sawtooth kfig 960923
2
3
4
5
Time (ms)
Figure 7.8: A sawtooth function having the same average slope as the preamp
output is subtracted from it and the difference amplified and offset
to match the input range of the ADC.
7.4 Analog to Digital Converter
Each DXP channel employs a 14-bit, 50MHz ADC to digitize the
conditioned input signal. Unlike competing products, the DXP directly digitizes
the wideband preamplifier signal.
7.5 The Filter, Pulse Detector, & Pile-up Inspector (FiPPI):
The FiPPI performs the various filtering, pulse detection and pileup
inspection tasks discussed in sections6.3 - 6.8. As described therein the FiPPI
contains up to three digital trapezoidal filters: A fast filter for pulse detection
and pileup inspection and event rejection; an intermediate filter for low energy,
i.e. soft x-ray, event detection and baseline acquisition; a slow filter for pulseheight (energy) measurement. An output stream of accepted baseline-subtracted
16-bit x-ray event energies is stored in a FIFO pipeline that is periodically read
out by the DSP. The FiPPI also measures live time, real time, and the number of
detected (accepted and rejected) input events. In the DXP Mercury, the FiPPI
also takes on the tasks of ASC monitoring and control and baseline processing,
reducing the work load of the DSP and thus improving the system throughput.
8/13/2010
95
DXP Mercury / Prospect User Manual
7.5.1
Prospect 1.0.x
FiPPI Configuration
The FiPPI is implemented in a field-programmable-gate-array (FPGA).
Configuration code for the DSP and FPGAs is contained in the XIA proprietary
FDD firmware file. FiPPI code is downloaded to the Mercury hardware when:
7.5.2
•
The hardware is initialized during startup.
•
A new initialization (INI) file and/or FDD file is selected.
•
The peaking time is adjusted such that a Decimation boundary is
crossed.
FiPPI Version and Variants
Not all FiPPI configuration files are the same. FiPPIs are first
distinguished by variant. Two standard FiPPI variants are available to all users:
A variant for pulse-reset type preamplifiers and a variant for RC-decay type
preamplifiers. . From time to time improved standard FiPPI versions are
released. In some cases custom FiPPI variant designs are used by certain
customers. It is important to make sure that you are using the latest version of
the appropriate FiPPI variant.
Note: RC-decay variant FiPPIs only have fast and slow filters, and the baseline
is acquired from the slow filter.
7.5.3
FiPPI Decimation
FiPPI’s are distinguished also by decimation. Decimation refers to preaveraging of the ADC signal prior to the FPGA processing pipeline. Each
decimation accommodates a specific range of peaking times, i.e. the shaping or
integration time of the slow (energy) filter. Typically four (4) FiPPI
configuration files are included in an FDD file. When the peaking time is
changed such that a range boundary is crossed, the host software downloads the
appropriate FiPPI configuration to the DXP Mercury.
Note: Pulsed-reset variant FiPPIs with decimation 0 only have fast and slow
filters, and the baseline is acquired from the slow filter.
7.5.4
Digital Trapezoidal Filtering
The digital trapezoidal filter produces a trapezoid response to a step
input. A filter is defined by its length (ascending or descending time of the
trapezoid), and its gap (flat-top time of the trapezoid). A constant input slope
input produces a constant DC offset; the bigger the slope, the bigger the offset.
This offset is referred to as the baseline.
The peak value of the value of the trapezoid response to an input step,
minus the baseline offset, is directly proportional to the height of the step, i.e.
the x-ray energy. The goal of the FiPPI is thus to measure the baseline, to
determine when to sample the peak of each trapezoid, and to subtract the
baseline from the sampled peak.
8/13/2010
96
DXP Mercury / Prospect User Manual
Prospect 1.0.x
7.5.4.1 Noise and Pileup
Detection occurs when the filter output crosses a constant threshold.
Generally speaking, a longer filter length produces better noise reduction, and
thus allows for a lower threshold. Pileup occurs when two or more successive
pulses result in an output trapezoid with a peak height that is dependent on a
combination of the input step heights. Piled-up events must be discarded. The
minimum detectable pileup time, or pulse-pair resolution, is proportional to the
filter length plus the filter gap. A tradeoff must therefore to be made between
pulse pair resolution and the minimum x-ray energy that can be detected.
7.5.4.2 Fast (Trigger) Filter
The fast filter is used solely to detect incoming x-ray events and to
determine whether a given event can be processed in the slow filter, or whether
it should be rejected. The fast filter’s peaking-time τp (τpf) can be adjusted
from 40 ns to 2.56 µs.
When τpf is 40 ns, the pulse pair resolution is typically less than 100
ns, however, x-rays in the 0eV - 2keV range may not be detected. When τpf is 1
µs, x-rays with energies below 500 eV can be detected, however, the pulse pair
resolution is greater 2 µs. We typically recommend running at τpf = 100ns, with
the fast gap time set to zero.
7.5.4.3 Slow (Energy) Filter
The slow filter trapezoid peak value is sampled for the energy
measurement. The slow filter’s peaking-time τp (τps, or simply referred to as
the peaking time τp) can be adjusted from 100 ns to 164 µs. At low input rates
τps should be set simply to optimize energy resolution. The optimal value will
depend on the detector and preamplifier. At higher input rates, τps should be
chosen such that maximum throughput is achieved. The resulting improvement
in statistics will compensate for any degradation of energy resolution.
The slow filter gap should always be longer than the response time of
the preamplifier, i.e. the settling time of input x-ray steps. If the input signal
displays a range of rise-times (as in the “ballistic deficit” phenomenon) the slow
filter gap time should be extended to accommodate that range.
7.5.4.4 Intermediate (Baseline) Filter
Note: The intermediate filter is not included for decimation 0, pulsed-reset
variant FiPPIs, and all RC-decay variant FiPPIs; the baseline is acquired from
the slow filter.
7.5.5
Statistics
The FiPPI also includes a livetime counter which counts the 20 MHz
system clock, divided by 16, so that one “tick” is 800 ns. This counter is
activated any time the DSP is enabled to collect x-ray pulse values from the
FiPPI and therefore provides an extremely accurate measure of the system
livetime. In particular, as described in section 0, the DSP is not live either
during preamplifier resets or during ASC out-of-ranges, both because it is
adjusting the ASC and because the ADC inputs to the FiPPI are invalid. Thus
the DXP measures livetime more accurately than an external clock, which is
8/13/2010
97
DXP Mercury / Prospect User Manual
Prospect 1.0.x
insensitive to resets and includes them as part of the total livetime. While the
average number of resets/sec scales linearly with the count-rate, in any given
measurement period there will be fluctuations in the number of resets which
may affect counting statistics in the most precise measurements.
All FiPPI parameters, including the filter peaking and gap times,
threshold, and pileup inspection parameters are externally supplied and may be
adjusted by the user to optimize performance. Because the FiPPI is
implemented in a Xilinx field programmable gate array (FPGA), it may also be
reprogrammed for special purposes, although this process is non-trivial and
would definitely require XIA contract support.
7.6 The Digital Signal Processor (DSP):
The DXP Mercury architecture was designed for speed. Because there
is only one DSP for four DXP channels, the processor load has therefore been
minimized. During a run, the DSP only processes events and compiles run
statistics. Prior to a run, the DSP initializes the FiPPIs according to user
settings. The DSP also performs various diagnostic tasks such as oscilloscope
mode trace capture and baseline histogramming. Such diagnostic tasks are
performed on one DXP channel at a time.
The processor is an Analog Devices ADSP-2183 16 bit Fixed-Point
DSP optimized for fixed point arithmetic and high I/O rates. Different DSP
program or code variants are used for different types of data acquisition.
7.6.1
Event Processing
Event processing consists of reading the event FIFOs of the DXP
channels, scaling the raw value of each measured pulse-height to the appropriate
spectrum bin number, and writing the scaled events to the System FPGA's event
post-processor.
The ADSP-2183 has 16K words of 16-bit wide data memory and 16K
words of 24-bit wide program memory, part of which is used as data memory to
hold the MCA spectrum. (If more memory is required for special purposes, up
to 4 Mbytes of extended memory can be added by specifying option M).
Transferring data to/from these memory spaces is done through the DSP’s builtin IDMA port, which does not interfere with the DSP program operation.
7.6.2
Statistics
Run statistics are updated periodically and written to the SRAM
memory.
7.7 System FPGA
The System FPGA primarily serves as a bus interface and arbitrator for
the various busses on the DXP Mercury: PCI Local Bus, Memory Bus, the DSP
Host Port and the DSP Data Bus.
During initialization the System FPGA routes the DSP program code
from the Local Bus to the DSP Host Port. The System FPGA supports
8/13/2010
98
DXP Mercury / Prospect User Manual
Prospect 1.0.x
diagnostic tasks such as reading oscilloscope trace data and DSP parameters via
the DSP Data Bus.
During a run, the System FPGA post-processes events and negotiates
host read operations from, and DSP write operations to, the SRAM memory.
The event post-processor first compares DSP-scaled energies to minimum
(underflow) and maximum (overflow) user-defined values. If the measured
energy is within the allowed range, the post-processor performs a readincrement-write operation to the appropriate address in memory, i.e. it
increments the corresponding spectrum bin.
The DXP Mercury currently supports basic spectrum acquisition as
well as time-resolved multi-spectrum acquisition. The two data acquisition
modes use different memory architectures and thus require different firmware
code to be downloaded.
In MCA mode on-board memory is configured as a single 1MEG by
32-bit device simultaneously accessible to both the host and the on-board DSP.
Each spectral bin is thus a 32-bit value, allowing for up to 4,294,967,295 events
per bin per run. The memory is normally cleared at the beginning of a run, but
can be preserved, allowing for 'pause and resume' functionality.
In 16-bit mode a data acquisition run produces multiple spectra for each
DXP processing channel. Spectrum memory is configured as two 1MEG by 16bit devices, memory A and memory B, each accessible to either the host or the
on-board DSP. Each spectral bin is thus a 16-bit value, allowing for up to
65,535 events per bin. Continuous operation is achieved by reading memory A
while writing memory B, and vice-versa.
7.7.1
Basic 32-bit MCA Data Acquisition
Figure 7.9: Data flow diagram for single spectrum mode.
A basic 32-bit MCA data acquisition run produces a single MCA
energy spectrum, ranging from 256 bins to 16384 bins, for each DXP processing
channel.
The external logic (BNC) input can be configured to halt data
acquisition, i.e. implemented as a GATE function. Data acquisition runs can be
started and stopped manually, or can be stopped automatically according to a
preset real time, live time or number of input or output events.
8/13/2010
99
DXP Mercury / Prospect User Manual
7.7.2
Prospect 1.0.x
Full Spectrum 16-bit MCA Mapping/Scanning Mode
This mode produces multiple spectra, where an external logic signal
typically is used. In the mapping mode, an x-ray beam is scanned across a
sample, and each spectrum corresponds to a scan point, or pixel. The external
logic (LEMO) input can be configured to control the pixel advance function,
which creates a new spectrum corresponding to a new pixel. Data acquisition
runs can be started and stopped manually, or can be stopped automatically
according to a preset number of pixels.
Pseudo-normal spectrum mode operation is supported for diagnostic
purposes, with the following limitations: Data cannot be read out during a run;
The 16-bit bins can easily overflow, depending on the run length and input count
rate.
Figure 7.10: Data flow diagram for multiple spectrum mode.
7.7.3
Other Data Acquisition Modes
The Mercury hardware supports additional data acquisition modes for
which firmware and software have yet to be developed. Please contact XIA if
you have are interested in using these modes, as their development schedules are
contingent on user interest.
8/13/2010
100
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Appendices
Appendix A. Accessing the Circuit Board in Bench-Top Models
The circuit board in the Mercury and Mercury-4 bench-top models can be
accessed by first removing the front and rear bezels from the box by gently
flexing the bezel along its shorter side. Next unscrew the two upper screws on
the front and rear panels. Next remove the top section of the box with firm
upward pressure along the side slots of the box; see figure A.1. It may be
necessary to flex the cover slightly to release it from the side channel. After the
circuit board is exposed, jumper settings can be changed as required. Replace
the top cover by performing the procedure in the reverse order.
Screws located
underneath
bezel
Front and rear
bezels
Side slot
Fig A.1: Mercury Bench-Top model housing,
8/13/2010
101
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Appendix B. Mercury Revision C Circuit BoardDescription
Figure B.1: Jumper, device and connector locations for the DXP Mercury printed circuit card.
8/13/2010
102
DXP Mercury / Prospect User Manual
Prospect 1.0.x
B.1. Jumper Settings
Reference
JP100
JP102
JP103
Name
Analog input
attenuation jumper
(0.100” shunt is
placed in one of two
positions on a 3-pin
header, i.e. forms a
SPDT switch)
Analog input
configuration
(2 adjacent 0.100”
shunts form a DPDT
switch)
Position Labels
0dB
(towards board edge)
-6dB
(towards board
center)
SINGLE
(towards board edge)
DIFF
(towards board
center)
Description
Default 0dB attenuation
setting: 10.0KΩ input
impedance; +/-4V input
range
-6dB attenuation setting:
500 Ω input impedance;
+/-8V input range.
Single-ended input
configuration (default)
Differential input
configuration
(for special custom
assemblies)
B.2. LED Indicators
Note: The default Mercury assembly includes on-board LEDs and
omits the LED output connector J25. The alternate assembly omits on-board
LED and includes the LED output connector J25 (see Connectors section
below).
Reference
U130
Name (Part #)
Rate Indicator
Red/Green bi-color
T-1
(Gilway P/N:
EA202)
Front Panel Labels
RATE
D3*
Status Indicators
Red
T-1
(Lumex P/N:
SSF-LXH103ID)
Status Indicators
Yellow
T-1
(Lumex P/N:
SSF-LXH103YD)
ERROR
D2*
8/13/2010
I/O
Description
This bi-color LED is a
rate indicator. Each flash
represents a preamplifier
reset.
- Green indicates a low
rate (<50% dead time)
- Yellow (green+red)
indicates a moderate rate
(>50% dead time, but less
than maximum
throughput)
- Red indicates a high rate
(at or above the point of
maximum throughput).
Illuminated red when
hardware is in an error
state.
Illuminated yellow while
USB I/O is busy.
103
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Status Indicators
RUN
Illuminated green when a
Green
run is in progress.
T-1
(Lumex P/N:
SSF-LXH103GD)
* LEDs also flash in a test pattern when firmware is being downloaded to the
board.
D4*
Figure B.2: Gilway EA202 dimensions.
Figure B.3: Lumex SSF-LXH103*D dimensions.
8/13/2010
104
DXP Mercury / Prospect User Manual
Prospect 1.0.x
B.3. Connectors
J100 – Signal Input: BNC (low-profile: center height = 0.515”), connects preamplifier
output to the Mercury.
BOMAR P/N: 364A595BL
J5 – GATE Input: BNC (low-profile: center height = 0.515”) CMOS/TTL input.
BOMAR P/N: 364A595BL
J6 – SYNC Input: BNC (low-profile: center height = 0.515”) CMOS/TTL input.
BOMAR P/N: 364A595BL
J103 – USB 2.0 Port: High speed parallel communications port; standard pinout. 4 pin
USB type B connector; right angle
Waldom/Molex P/N: 67068-0000
J25 – LED output port (connector omitted in standard assembly; LED D4 must be
removed to install shrouded header):
Low-profile 10-pin shrouded 0.100” DIP header.
3M P/N: N2510-6002RB (mating connector e.g. D89110-0131HK; pre-assembled
IDC ribbon-cable harnesses available, e.g. Digi-Key P/N M3AKK-1006J-ND)
Pin(s)
Net
Description
5V power
8,10
V+5
Ground
5,7,9
GND
TTL/CMOS output (0 – off; 1 – on)
1
RateGreen
TTL/CMOS output (0 – off; 1 – on)
2
RunLED
TTL/CMOS output (0 – off; 1 – on)
3
RateRed
TTL/CMOS output (0 – off; 1 – on)
4
ErrLED
TTL/CMOS output (0 – off; 1 – on)
6
I/OLED
J105 – Power Input: 12-pin 0.156” ramp-lock header
Waldom/Molex P/N: 22-05-3121 (mating connector 22-01-3127)
Pin(s)
Net
Description
1, 3, 5, 7
GND
0
2
V+12Raw
+12V to +15V
50mA
4
V-12Raw
-12V to -15V
50mA
6
V+6Raw
+5.5V to 6.0V
1.0A
8
PWR_DIG4
Configurable TTL/CMOS
input/output. Unused for now.
9
PWR_DIG3
Configurable TTL/CMOS
input/output. Unused for now.
10
PWR_DIG2
Configurable TTL/CMOS
input/output. Unused for now.
11
PWR_DIG1
Configurable TTL/CMOS
input/output. Unused for now.
12
PWR_DIG0
Configurable TTL/CMOS
input/output. Unused for now.
8/13/2010
105
DXP Mercury / Prospect User Manual
Prospect 1.0.x
J108 – Auxiliary Port: 24 TTL/CMOS configurable lines, plus VCC and GND
connections.
Low-profile 34-pin shrouded 0.100” DIP header.
3M P/N: N2534-6002RB (mating connector e.g. D89134-0131HK; pre-assembled
IDC ribbon-cable harnesses available, e.g. Digi-Key P/N M3AKK-3406J-ND)
Pin Net (Description)
Pin Net (Description)
1
GND (Ground)
2
AUX0 (TTL/CMOS
input/output)
3
AUX1 (TTL/CMOS input/output)
4
AUX2 (TTL/CMOS
input/output)
5
AUX3 (TTL/CMOS input/output)
6
AUX4 (TTL/CMOS
input/output)
7
AUX5 (TTL/CMOS input/output)
8
GND (Ground)
9
GND (Ground)
10
AUX6 (TTL/CMOS
input/output)
11
AUX7 (TTL/CMOS input/output)
12
AUX8 (TTL/CMOS
input/output)
13
AUX9 (TTL/CMOS input/output)
14
AUX10 (TTL/CMOS
input/output)
15
AUX11 (TTL/CMOS input/output)
16
GND (Ground)
17
AUX12 (TTL/CMOS input/output)
18
AUX13 (TTL/CMOS
input/output)
19
GND (Ground)
20
AUX14 (TTL/CMOS
input/output)
21
AUX15 (TTL/CMOS input/output)
22
AUX16 (TTL/CMOS
input/output)
23
AUX17 (TTL/CMOS input/output)
24
AUX18 (TTL/CMOS
input/output)
25
AUX19 (TTL/CMOS input/output)
26
AUX20 (TTL/CMOS
input/output)
27
GND (Ground)
28
GND (Ground)
29
AUX21 (TTL/CMOS input/output)
31
AUX22 (TTL/CMOS
input/output)
31
AUX23 (TTL/CMOS input/output)
32
GND (Ground)
33
VCC (+3.3V)
34
GND (Ground)
See also Appendix D for Auxiliary Port functions.
8/13/2010
106
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Appendix C. Mercury-4 Revision A Circuit Board Description
Figure C.1: Jumper, device and connector locations for the DXP Mercury4 printed circuit card.
8/13/2010
107
DXP Mercury / Prospect User Manual
Prospect 1.0.x
C.1 Jumper Settings
Reference
JP100
JP200
JP300
JP400
Name
Analog input
attenuation jumpers
(standard 0.100”
shunts are placed in
one of two positions
on a 3-pin header)
Position Labels
0dB
(towards board edge)
-20dB (erroneous)
(towards board center)
Description
Default 0dB
attenuation setting:
10.0KΩ input
impedance; +/-4V
input range
-6dB attenuation
setting: 500 Ω input
impedance; +/-8V
input range.
C.2. LED Indicators
Note: The default Mercury4 assembly includes on-board LED and omits the
LED output connector J19. The alternate assembly omits on-board LED and
includes the LED output connector J19 (see Connectors section below).
Reference Name (Part #)
Front Panel Labels
Description
Rate Indicators
RATE
These bi-color LEDs are
U130
Red/Green bi-color
rate indicators for each
U230
T-1
channel. Each flash
U330
(Gilway P/N:
represents a preamplifier
U430
EA202)
reset.
- Green indicates a low
rate (<50% dead time)
- Yellow (green+red)
indicates a moderate rate
(>50% dead time, but less
than maximum
throughput)
- Red indicates a high rate
(at or above the point of
maximum throughput).
Status Indicators
RUN
Illuminated green when a
U51*
Red/Yellow/Green
run is in progress.
T-1
I/O
Illuminated yellow when
(Dialight P/N:
a USB transfer to or from
570-0100-132)
the module is in progress.
ERR
Illuminated when the
module is in an error state.
*This LED flashes in a test pattern when firmware is being downloaded to the
board.
8/13/2010
108
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Figure C.2: Gilway EA202 dimensions.
Figure C.3: Dialight 570-0100-132 dimensions.
8/13/2010
109
DXP Mercury / Prospect User Manual
Prospect 1.0.x
C.3 Connectors
J100, J200, J300, J400 – Signal Inputs (4): BNC (low-profile: center height =
0.515”), connects preamplifier output to the Mercury4.
BOMAR P/N: 364A595BL
J5 – GATE Input: BNC (low-profile: center height = 0.515”) CMOS/TTL input.
BOMAR P/N: 364A595BL
J6 – SYNC Input: BNC (low-profile: center height = 0.515”) CMOS/TTL input.
BOMAR P/N: 364A595BL
J103 – USB 2.0 Port: High speed parallel communications port; standard pinout. 4 pin
USB type B connector; right angle
Waldom/Molex P/N: 67068-0000
J19 – LED output port (connector omitted in standard assembly):
Low-profile 16-pin shrouded 0.100” DIP header.
3M P/N: N2516-6002RB (mating connector e.g. D89116-0131HK; pre-assembled
IDC ribbon-cable harnesses available, e.g. Digi-Key P/N M3AKK-1606J-ND)
Pin(s)
Net
Description
5V power
1, 5
V+5
Ground
8, 11, 14
GND
TTL/CMOS output (0 – off; 1 – on)
2
RunLED
TTL/CMOS output (0 – off; 1 – on)
3
I/OLED
TTL/CMOS output (0 – off; 1 – on)
4
ErrLED
TTL/CMOS output (0 – off; 1 – on)
6
RateGreen0
TTL/CMOS output (0 – off; 1 – on)
7
RateRed0
TTL/CMOS output (0 – off; 1 – on)
9
RateGreen1
TTL/CMOS output (0 – off; 1 – on)
10
RateRed1
TTL/CMOS output (0 – off; 1 – on)
12
RateGreen2
TTL/CMOS output (0 – off; 1 – on)
13
RateRed2
TTL/CMOS output (0 – off; 1 – on)
15
RateGreen3
TTL/CMOS output (0 – off; 1 – on)
16
RateRed3
J105 – Power Input: 12-pin 0.156” ramp-lock header
Waldom/Molex P/N: 22-05-3121 (mating connector 22-01-3127)
Pin(s)
Net
Description
1, 3, 5, 7
GND
0
2
V+12Raw
+12V to +15V
100mA
4
V-12Raw
-12V to -15V
100mA
6, 8
V+6Raw
+5.5V to 6.0V
3.0A
9
PWR_DIG3
Configurable TTL/CMOS
input/output. Unused for now.
10
PWR_DIG2
Configurable TTL/CMOS
input/output. Unused for now.
11
PWR_DIG1
Configurable TTL/CMOS
input/output. Unused for now.
12
PWR_DIG0
Configurable TTL/CMOS
input/output. Unused for now.
8/13/2010
110
DXP Mercury / Prospect User Manual
Prospect 1.0.x
J108 – Auxiliary Port: 40 TTL/CMOS configurable lines, plus VCC and GND
connections.
Low-profile 50-pin shrouded 0.100” DIP header.
3M P/N: N2550-6002RB (mating connector e.g. D89150-0131HK; pre-assembled
IDC ribbon-cable harnesses available, e.g. Digi-Key P/N M3AKK-5006J-ND)
Pin Net (Description)
Pin Net (Description)
1
AUX0 (TTL/CMOS input/output)
2
AUX1 (TTL/CMOS
input/output)
3
AUX2 (TTL/CMOS input/output)
4
AUX3 (TTL/CMOS
input/output)
5
AUX4 (TTL/CMOS input/output)
6
AUX5 (TTL/CMOS
input/output)
7
AUX6 (TTL/CMOS input/output)
8
AUX7 (TTL/CMOS
input/output)
9
GND (Ground)
10
GND (Ground)
11
AUX8 (TTL/CMOS input/output)
12
AUX9 (TTL/CMOS
input/output)
13
AUX10 (TTL/CMOS input/output)
14
AUX11 (TTL/CMOS
input/output)
15
AUX12 (TTL/CMOS input/output)
16
AUX13 (TTL/CMOS
input/output)
17
AUX14 (TTL/CMOS input/output)
18
AUX15 (TTL/CMOS
input/output)
19
VCC (3.3V)
20
VCC (3.3V)
21
AUX16 (TTL/CMOS input/output)
22
AUX17 (TTL/CMOS
input/output)
23
AUX18 (TTL/CMOS input/output)
24
AUX19 (TTL/CMOS
input/output)
25
AUX20 (TTL/CMOS input/output)
26
AUX21 (TTL/CMOS
input/output)
27
AUX22 (TTL/CMOS input/output)
28
AUX23 (TTL/CMOS
input/output)
29
GND (Ground)
31
GND (Ground)
31
AUX24 (TTL/CMOS input/output)
32
AUX25 (TTL/CMOS
input/output)
33
AUX26 (TTL/CMOS input/output)
34
AUX27 (TTL/CMOS
input/output)
35
AUX28 (TTL/CMOS input/output)
36
AUX29 (TTL/CMOS
input/output)
37
AUX30 (TTL/CMOS input/output)
38
AUX31 (TTL/CMOS
input/output)
39
VCC (+3.3V)
40
VCC (+3.3V)
41
AUX32 (TTL/CMOS input/output)
42
AUX33 (TTL/CMOS
input/output)
43
AUX34 (TTL/CMOS input/output)
44
AUX35 (TTL/CMOS
input/output)
45
AUX36 (TTL/CMOS input/output)
46
AUX37 (TTL/CMOS
input/output)
47
AUX38 (TTL/CMOS input/output)
48
AUX39 (TTL/CMOS
input/output)
49
GND (Ground)
50
GND (Ground)
See also Appendix D for Auxiliary Port functions.
8/13/2010
111
DXP Mercury / Prospect User Manual
8/13/2010
Prospect 1.0.x
112
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Appendix D. Specification for ROI outputs on the Mercury and
Mercury4 Auxiliary Port
The standard firmware for direct to customer sales of the Mercury and
Mercury4 will include support for real-time TTL outputs for regions of interest
(ROI’s), as well as an output of a real time signal indicating triggers (detected
events) and another signal indicating live time.
The Mercury Auxiliary port is a 34-pin dual-row header that contains
24 fully programmable digital I/O lines, connected to the System FPGA
(SysFPGA) through a series 22 ohm resistor. The Mercury4 increases the total
number of I/O lines to 40 in a 50-pin dual-row header. The Mercury will support
TTL outputs for up to 16 ROI’s, while the Mercury4 will support up to 8 ROI’s
per channel.
D.1 Signal Assignment
For the Mercury, the auxiliary lines are numbered Aux0 through
Aux23; similarly, for the Mercury4, the signals are labeled Aux0 through
Aux39.
8/13/2010
Mercury
Signal
Aux0 through Aux15
Aux16
Aux17
Aux18 through Aux 23
Assignment
ROI0 through ROI15
Trigger output
Live time output
Unused; drive low
Mercury-4
Signal
Aux0 through Aux7
Aux8
Aux9
Aux10 through Aux17
Aux18
Aux19
Aux20 through Aux27
Aux28
Aux29
Aux30 through Aux37
Aux38
Aux39
Assignment
ROI0 through ROI7, Channel 0
Trigger output, Channel 0
Live time output, Channel 0
ROI0 through ROI7, Channel 1
Trigger output, Channel 1
Live time output, Channel 1
ROI0 through ROI7, Channel 2
Trigger output, Channel 2
Live time output, Channel 2
ROI0 through ROI7, Channel 3
Trigger output, Channel 3
Live time output, Channel 3
113
DXP Mercury / Prospect User Manual
Prospect 1.0.x
D.1.1. Front panel Connector assignment for Mercury
Bench-Top models.
For the bench-top models, not all digital outputs are available on the
external connector – only 16 connections are available for the Mercury and 32
for the Mercury-4. As a result, some signals are dropped, and some assignments
changed from the general case above.
For the single channel Mercury bench-top model the auxiliary lines are
brought out to a 25-way connector on the front panel. For the Mercury-4 two
25-way connectors are utilized. Break-out cables with BNC connectors are
provided for easy access to the front panel signals. The default pin and cable
assignments are listed below.
Mercury 25-way connector: Default pin and break-out cable assignments
Signal
25-way pin
Break-out
assignment
cable
ROI 0
14
Black 1
ROI 1
15
Black 2
ROI 2
16
Black 3
8/13/2010
ROI 3
ROI 4
17
5
Black 4
Black 5
ROI 5
ROI 6
ROI 7
18
6
19
Black 6
Black 7
Black 8
ROI 8
ROI 9
ROI 10
7
20
8
Gray 1
Gray 2
Gray 3
ROI 11
ROI 12
ROI 13
21
22
23
Gray 4
Gray 5
Gray 6
Trigger
Live Time
24
25
Gray 7
Gray 8
114
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Mercury-4 two 25-way connector pin assignments.
Connector #1
Signal
Connector #2
Break-out
cable
Black 1
Black 2
Black 3
Signal
ROI 0 Ch 0
ROI 1 Ch 0
ROI 2 Ch 0
25-way pin
assignment
14
15
16
ROI 3 Ch 0
ROI 4 Ch 0
17
5
ROI 5 Ch 0
Trig Ch 0
LT Ch 0
ROI 0 Ch 2
ROI 1 Ch 2
ROI 2 Ch 2
25-way pin
assignment
14
15
16
Break-out
cable
Black 1
Black 2
Black 3
Black 4
Black 5
ROI 3 Ch 2
ROI 4 Ch 2
17
5
Black 4
Black 5
18
6
19
Black 6
Black 7
Black 8
ROI 5 Ch 2
Trig Ch 2
LT Ch 2
18
6
19
Black 6
Black 7
Black 8
ROI 0 Ch 1
ROI 1 Ch 1
ROI 2 Ch 1
7
20
8
Gray 1
Gray 2
Gray 3
ROI 0 Ch 3
ROI 1 Ch 3
ROI 2 Ch 3
7
20
8
Gray 1
Gray 2
Gray 3
ROI 3 Ch 1
ROI 4 Ch 1
ROI 5 Ch 1
21
22
23
Gray 4
Gray 5
Gray 6
ROI 3 Ch 3
ROI 4 Ch 3
ROI 5 Ch 3
21
22
23
Gray 4
Gray 5
Gray 6
Trig Ch 1
LT Ch 1
24
25
Gray 7
Gray 8
Trig Ch 3
LT Ch 3
24
25
Gray 7
Gray 8
D.2 Signal Descriptions
For this application, all auxiliary signals are outputs, driven at 3.3V CMOS
levels.
D.2.1. ROI Outputs
The Mercury will support logic outputs for up to 16 ROI’s, and the
Mercury4 will support up to 8 ROI’s per channel. When the DSP processes a
detected event, it will write a word to a register in the SysFPGA that will have a
bit set for every ROI that contains the processed event (note that for this
application, the ROI’s are allowed to overlap). A write to the ROI register in the
SysFPGA will generate a 500 ns pulse (high true) on any ROI line
corresponding to a set bit in the register. After a pulse, the ROI line must be low
for at least 500 ns before processing the next event; this will likely require some
buffering of the values written to the ROI register, as the DSP can process
events faster than 1 microsecond per event.
To support the ROI outputs, the Mercury will need a single ROI
register, while the Mercury4 will require one register per channel (a total of
four). These registers will be described in more detail later.
8/13/2010
115
DXP Mercury / Prospect User Manual
Prospect 1.0.x
D.2.2. Trigger and Live Time Outputs
Both the Mercury and Mercury4 will provide trigger and live time
information on the auxiliary port along with the ROI pulses. These signals are
real-time signals, and will come directly from the Filter FPGA (FiPPI) of the
appropriate channel using the SysFip bus. The SysFip bus assignments are given
below separately for the Mercury then the Mercury4.
Mercury
Signal
SysFip0
SysFip4
Assignment
Triggers
Live time
Mercury-4
Signal
SysFip0
SysFip1
SysFip2
SysFip3
SysFip4
SysFip5
SysFip6
SysFip7
Assignment
Triggers, Channel 0
Triggers, Channel 1
Triggers, Channel 2
Triggers, Channel 3
Live time, Channel 0
Live time, Channel 1
Live time, Channel 2
Live time, Channel 3
There are several choices for the trigger output for each channel; this
choice is written to a register in the appropriate FiPPI, which will drive the
selected signal onto the appropriate SysFip line. The signal will normally be
driven low; when an event is detected, the signal is driven high. The choices are
described below:
Trigger Choice
0
1
2
3
Description
Trigger output disabled – drive low
Fast filter threshold crossing
Baseline filter threshold crossing (decimation 0:
defaults back to fast filter)
Energy filter threshold crossing
Similarly, there are several settings for the live time output, described
below. Note that the live signal is driven high when the system is live; the
system is not live whenever the signal is out of range (due to a reset or drift out),
or when the selected trigger filter is above threshold.
Live Time Choice
0
1
2
3
4
8/13/2010
Description
Live Time output disabled – drive low
Not live when fast filter over threshold
Not live when baseline filter over threshold
(defaults to same as choice 1 for decimation 0)
Not live when energy filter over threshold
Not live when PKBUSY=0 (triggered by any
active threshold)
116
DXP Mercury / Prospect User Manual
Prospect 1.0.x
D.3. Register Definitions
This section describes the registers used to control the ROI outputs on
the Mercury and Mercury4.
D.3.1. FiPPI Registers
Registers in the FiPPI will be used to select the source of the Trigger
and Live Time outputs. For the Mercury, one register will be used to select the
Trigger output and another will be used to select the Live Time output. For the
Mercury4, there will be one register in each FiPPI to select the outputs; each
register will control the two channels contained in that FiPPI.
D.3.2. SysFPGA Registers
For the Mercury, only a single target register is needed. The 16 bits in
that register are mapped directly to the 16 auxiliary outputs corresponding to the
ROI’s. Since the output rate of the auxiliary register is limited by the 500 ns
pulse widths (and minimum 500 ns period between ROI pulses), some buffering
will be necessary between the register input and the auxiliary port output. A 16deep FIFO should be sufficient.
For the Mercury4, four target registers are needed; one for each
channel. Only 8 bits are required for each channel. For the Mercury4, the DSP
creates an ROI map, where one 16-bit word contains the ROI definitions for 2
channels; channels 0 (lower byte) and 1 (upper byte) are combined into one
array, while channels 2 and 3 (upper and lower respectively) are combined into
another. For channels 0 and 2, the lower byte will be used to transfer the ROI
pulse, while for channels 1 and 3, the upper byte will be used. Similarly to the
Mercury, buffering is required between the ROI register and the auxiliary
register outputs. The ROI register connections are summarized below.
ROI Register
Channel 0 ROI Register
Channel 1 ROI Register
Channel 2 ROI Register
Channel 3 ROI Register
Bits
0 – 7 (8 – 15 are ignored)
8 – 15 (0 – 7 are ignored)
0 – 7 (8 – 15 are ignored)
8 – 15 (0 – 7 are ignored)
Auxiliary Register Output
Aux0 – Aux7
Aux10 – Aux17
Aux20 – Aux27
Aux30 – Aux37
Addresses: Channel offset + 0xE
8/13/2010
117
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Appendix E. Mapping Buffer Specification
In mapping mode, the DXP-Mercury uses two completely separate
memory buffers, enabling the system to take data into one buffer while the other
buffer can be read out y the host. The size of each buffer is 2MB, organized as
1Mword by 16 bits. Several mapping modes will be supported, including
mapping with full spectra, mapping with multiple regions of interest (ROI’s),
and list mode readout.
For all timing modes, the buffer starts out with a buffer header,
containing general information about the data contained in the memory block.
For all timing modes involving sequential pixels, there is also a pixel header
block, typically containing statistics information (used to make pileup
corrections on a pixel by pixel basis).
The format of the data contained in the buffers is described in detail in
the following sections.
E.1. Buffer Header
For all timing applications that use the dual buffers, the buffer header
will have a fixed, 256-word length (the word size is 16 bits in this mode). The
contents of the header are defined below:
Word Number
Contents
0
Tag Word 0: 0x55AA
1
Tag Word 1: 0xAA55
2
Buffer Header Size (=256)
Mapping Mode:
1: Full Spectrum
3
2: Multiple ROI
3: List Mode
8/13/2010
4
Run Number
5 to 6
Sequential Buffer Number (low word first)
7
BufferID (0:A, 1:B)
8
Number of Pixels in buffer
9 to 10
Starting Pixel Number (low word first)
118
DXP Mercury / Prospect User Manual
Prospect 1.0.x
11
Module Serial Number?/Module #
Word Number
Contents
12
Detector Channel 0 (set by host in DSP)
13
Det. Element, Ch0
14
Reserved (set to 0)
15
Reserved (set to 0)
16
Reserved (set to 0)
17
Reserved (set to 0)
18
Reserved (set to 0)
19
Reserved (set to 0)
20
Channel 0 Size (number of words)
21
Reserved (set to 0)
22
Reserved (set to 0)
23
Reserved (set to 0)
Buffer errors: Buffer overrun
24
0: No error
>0: Number of extra pixels combined with last
pixel in buffer
25-31
Reserved (set to 0)
32-63
32 User words (set in USER DSP array)
64-255
Reserved (set to 0)
E.2. Pixel Data Block
For all mapping modes based upon pixels, the data block for each pixel
will start with a pixel header, followed by the data collected for the pixel. The
header can differ in size for different mapping applications; in general, the
header contains the statistics data required to make pileup corrections ona pixel
by pixel basis (livetime, realtime, input triggers, and output events). The full
data blocks are described below for the various mapping modes.
8/13/2010
119
DXP Mercury / Prospect User Manual
Prospect 1.0.x
E.2.1. Mapping Mode 1: Full Spectrum Mapping
The pixel header for full spectrum mapping mode is described below;
due to the constraint that the spectra sizes are a multiple of 256 and must start on
an even multiple of 256, the size of the pixel header is 256 words in this mode.
The data block for full spectrum mapping mode contains four sections;
each section holds the spectrum from one of the four detector channels in the
module. The length of the spectra are constrained to be a multiple of 256, and
must start on a memory location that is a multiple of 256. For (at least) the first
version of the mapping firmware, the spectra must be the same size for all
channels in a system (and will in general equal 256 * 2^n, ie 256, 512, 1024,
2048, etc). The format for the entire pixel block is described in the table below;
please note that the pixel header definition is designed to be consistent with the
xMAP, which requires room for three additional channels.
Word Number
Contents
0
Tag Word 0: 0x33CC
1
Tag Word 1: 0xCC33
2
Pixel Header Size (=256)
3
Mapping Mode (=1)
Pixel Number
(low word first)
4 to 5
In the case of a mapping error where one pixel
record combines data from several pixels, this is
the number of the last pixel recorded.
Total Pixel Block size in words
6 to 7
(including header)
(low word first)
8/13/2010
8
Channel 0 Size (K words)
9
Reserved (set to 0)
10
Reserved (set to 0)
11
Reserved (set to 0)
12 to 31
Reserved (set to 0)
Word Number
Contents
120
DXP Mercury / Prospect User Manual
Prospect 1.0.x
Channel 0 Statistics:
Realtime (2 words, low word first)
32 to 39
Livetime (2 words)
Triggers (2 words)
Output events (2 words)
40 to 47
Reserved (set to 0)
48 to 55
Reserved (set to 0)
56 to 63
Reserved (set to 0)
64 to 255
Reserved (set to 0)
256 to (256 + K – 1)
Channel 0 Spectrum
E.2.2. Mapping Mode 2: Multiple SCA Mapping
The pixel header for multiple SCA (or ROI) mapping mode is
described below; there is no constraint on the data alignment in the buffer, so the
header length is shorter than the 256 words required for full spectrum mapping.
There are four sections in the pixel data block for this mode, containing
the ROI totals for each of the four detector channels. Up to 64 ROI’s can be
defined for each channel. A 16K word array is used to hold the mapping
between MCA bins and SCA regions; the user can either select to have all
channels use the same SCA definitions for all channels (which supports the full
maximum MCA length of 16K channels), or use separate definitions of the SCA
regions for each channel (where the maximum supported MCA length is 4K
channels). This format specification does allow differences in the number of
SCA’s between channels. Two words (32 bits total) are used to store the total
number of events in each ROI; the low word is stored first in memory.
The format for the entire pixel data block is described in the table
below.
8/13/2010
Word Number
Contents
0
Tag Word 0: 0x33CC
1
Tag Word 1: 0xCC33
2
Pixel Header Size (=64)
121
DXP Mercury / Prospect User Manual
Prospect 1.0.x
3
Mapping Mode (=2)
Pixel Number (low word first)
4 to 5
In the case of a mapping error where one pixel
record combines data from several pixels, this is the
number of the last pixel recorded.
6 to 7
Total Pixel block size in words (including header)
(low word first)
8
Number of ROI, Channel 0 (K ROI’s)
9
Reserved (set to 0)
10
Reserved (set to 0)
11
Reserved (set to 0)
12
ROI Size in words (=2)
13 to 31
Reserved (set to 0)
Channel 0 Statistics:
Realtime (2 words, low word first)
32 to 39
Livetime (2 words)
Triggers (2 words)
Output events (2 words)
40 to 47
Reserved (set to 0)
48 to 55
Reserved (set to 0)
56 to 63
Reserved (set to 0)
64 to (64 + 2*K – 1)
Channel 0 ROI data
E.2.3. Mapping Mode 3: List Mode Mapping
Note: List Mode Mapping is not yet supported.
For the data block in list mode, each word contains data for a single
event. The events are stored in the order they are processed, and so the data are
not separated according to channel. The channel number information is
embedded into the upper two bits of the data word itself; the lower 14 bits are
sued to store energy information (typically MCA channel number); for the case
of the Mercury, the upper two bits will always be zero.
8/13/2010
122
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement