19.6 NEXRAD OPEN RADAR DATA ACQUISITION (ORDA

19.6
NEXRAD OPEN RADAR DATA ACQUISITION (ORDA) RECEIVER CHARACTERISTICS
Nita K. Patel* and Gordon W. Jim
RS Information Systems, Inc., Norman, Oklahoma
Alan D. Free
SI International, Norman, Oklahoma
1.
ABSTRACT
The Open Radar Data Acquisition (ORDA)
system receiver path is significantly different
than the legacy NEXRAD Radar Data
Acquisition (RDA) receiver path. The ORDA
system realizes improved performance and
reliability through the integration of the SIGMET
72Mhz IF digital receiver that greatly reduces
system component count. This paper will
discuss
system
performance
parameters
including minimum detectable signal, dynamic
range, matched filtering and other intrinsic
receiver characteristics. The inter-relationship
between these parameters and procedures for
optimization will be discussed. A comparison
against legacy performance will also be provided
where appropriate.
2.
INTRODUCTION
The receiver channel detects and converts
the returned RF energy into a complex, phasecoherent analog (video) signal.
The RF
generator creates transmitter RF, test RF,
STALO & COHO signals. The in-phase (I) and
0
quadrature (Q-90 shifted) video contains the
echo return amplitude and phase information
needed for base data generation. Log video is
used to provide a wide dynamic range automatic
gain control (AGC) function and also supports
the interference detection function. The I/Q and
log video are sent to the A/D converter assembly
where it is digitized, purged of ground clutter,
and converted to radial base data.
The RF test signal generation sub-function
sends samples of signals (in digital format) from
various internal monitor points to the signal
processor (HSP and/or PSP).
The signal
processor takes this data and calculates
parameters needed for receiver calibration.
From UD3
1DC 1
Directional
Coupler
RF s ign al to UD3 RF Dr ive r
Test Pat h to 4A2 2 4 Po s. D iod e Switch
AT4
High Power
3 dB
attenuator
Tr ansmitter RF signal
The NEXRAD ORDA enhancement program
replaces proprietary hardware with openplatform commercial hardware, a digital IF
receiver and Graphical User Interface based on
open architecture software principles (Patel,
2004). This paper presents an analysis of the
ORDA receiver path. A comparison of system
performance parameters such as sensitivity and
dynamic range will be presented with an
explanation
of
differences
from
and
enhancements to the legacy system.
R 49
AT33
6 dB
Attenuator
UD4 Legacy Signal Path
Path Loss
- 39 dB
Received
RADAR Signal
or CW Test
from UD4
cabinet
R72 , R 73
J2
R5 0, R51 , R5 2
J3
Test Path 1 to AT34
10 dB Attenuator
4 A20
J4
4 Way Splitter
RF Sample
Jack
4 A1
RF Generator
STALO J2
R83
J1
4A5
J3
Mixer
J 2 Preamplifier
R 74
2A4
Low Noise
Amplifier
28 dB gain
2A1 A3FL1
Bandpass
Filter
R 77
UD2
UD4
4A2 6
J1Power SensorJ 2
J1
R7 8
4A3 6
Optional
Attenuator
J5
XMTR J1
2 A3
Receiver
Protector
R6 9
R 81
4 A4
Pre -Select
Bandpass
Filter
4DC2
Directional
Coupler
R8 8
R 91
R9 2
R 94
Matched
Filter
Delay
Line
IF (AGC )
Attenuator
IF
Amp
20 dB gain
Test RF J3
3.
LEGACY SIGNAL PATH
J4
R9 7
I /Q Phase
Detector
COHO J 4
(Optional )
Interference
Suppression
Unit
The legacy receiver signal path (Figure 1)
consists of the receiver channel with an RF
generator and an RF test signal generation and
processing path.
R F/ IF Test
Monitor
RF /IF Test
Monitor
R9 8
A/D
Converter
Receiver control data
Log Video signal
from receiver
Figure 1, Legacy Receiver Signal Path
* Corresponding author address: Nita K. Patel,
RS Information Systems, Inc., 2227 W. Lindsey
Ave., Suite 1500, Norman, OK 73069; e-mail:
Nita.K.Patel@noaa.gov
4.
Raw I and Q data
ORDA SIGNAL PATH
The ORDA design is functionally similar to
the legacy receiver signal path; however, the
ORDA receiver signal path has replaced legacy
matched filtering, A/D converting and signal
processing with a SIGMET Intermediate
Frequency Digitizer (IFD) and signal processor
(RVP), as shown in Figure 2. The ORDA design
digitizes IF signal while maintaining all minimum
WSR-88D receiver characteristics (such as
dynamic range and sensitivity). This reduces
noise while improving receiver performance and
reliability.
Transmitter RF s ignal
From UD3
1DC1
Directional
Coupler
20.66 d Bm
R F signal t o UD 3 RF D riv er
Tes t P ath t o 4A 22 4 Pos. Diode Swit ch
AT4
High Power
3 dB
attenuator
R49
AT33
6 dB
Attenuator
14.81 dBm
UD4 ORDA Signal Path
Path Loss
-39 d B
Received
RADAR Signal
or CW Test
from UD4
cabinet
R72, R73
(CW Test
Signal)
2.62 dBm
2A3
Receiver
Protector
R69
J1
4.3 Signal Levels
4A26
J1
J2
Power Sensor
RF Sample
Jack
4A20 J4
4 Way Splitter
J5
8.46 dBm
22.54 dBm
Site variable attenuator
50 dB
may be 3 or 6 dB
Attenuator
13.00 dBm
4A1
RF Generator
XMTR J1
STALO J2
15.35 dBm
6 dB
Directional
Coupler
LO In
STALO 1
STALO 2
R257, R258
Test RF J3
R81
4A4
Pre-Select
Bandpass
Filter
- 53.54 dBm
24-34 dB
Amplifier
R259
26 dB 0.77 dBm
Attenuator
R255
-24.54 dBm
R83
R249
J3
J2
4A5
20.06 dBm Site
J4
Mixer
J1 IFD
J3
Variable
Preamplifier
R250
J5
Attenuator
J2 20 dB gain
Site variable attenuator
may be 1,2,3,4,5,6 dB
J1
17.0 6 dBm
4DC2
Directional
Coupler
13 .55 dBm
R78
4A36
Optional
Attenuator
UD4
R252
Site
Var. - 44.54 dBm
Atten.
RF In
IF Out
Burst Mixer
9.35 dBm STALO In
R254
26.77 dBm
COHO J4
R77
UD2
Test Path 1 to AT34
10 dB Attenuator
J2
R50, R51, R52
J3
R74
2A4
Low Noise
Amplifier
28 dB gain
2A1A3FL1
Bandpass
Filter
The WSR-88D uses 2 pulse widths, and
therefore requires 2 different matched filters for
optimal performance. Short pulse is filtered at
600KHz, and Long pulse is filtered at 200KHz.
The matched filter used is a Finite Impulse
Response (FIR) filter.
The filter selected has wide implications on
the receiver’s performance. As noted above, the
bandwidth affects the noise floor, and therefore
the receiver sensitivity.
However, a filter
narrower than the pulse width causes signal
loss. In the WSR-88D, the 600kHz short pulse
matched filter causes approximately 0.6dB
signal loss. The 200kHz long pulse matched
filter results in approximately 0.2dB signal loss.
To
RVP 8
(UD90)
A transmitted pulse sample is used to
generate a burst pulse, which is used for phase
reference in the IFD. Typical field values for
signal level through the receiver path are given
in Figure 2.
The maximum input signal for the IF and
burst input is +6.5dBm. An attenuator is used at
the input of the IFD to adjust our signal levels to
meet the IFD specifications.
2.59 dBm
All measurements in
average power
Figure 2, ORDA Receiver Signal Path
4.1 Digital Receiver
To maintain system calibration, the WSR88D system uses four built-in test signals as
shown in Figure 3. Typical values through the
receiver path are provided in this figure for these
test signals. The NOISE test source is used to
measure system noise temperature. CW signal
is used to measure system linearity and dynamic
range. The KD test signal is used for clutter
suppression measurements and provides a
general measure of system phase stability.
KD Test Path
From 4 A20
4 Way Power
Splitter
36.60 dBm
R53
R54
AT 34
10 dB
Attenuator
4DC1
Directional
Coupler
KD ) -4.96 dBm
RFD) 19.76 d B m
R5 7, R58, R59 , R60 CW ) 20.57 dBm
-18.47 d B m
RFD Test Path
CW Test Path
21.62 d B m
22.54 d B m
NOISE ) 60.64 ENR
KD
4A22
4 Position
J5
Diode
CW
Switch
RFD
NOISE
62.69 EN R
4 A25
J1 Noise J2
Source
KD ) -22.92 dBm
RFD ) 1.80 dBm
CW ) 2.62 dBm
NOISE ) 42.68 ENR
R63
J1
J2
4A23
7 Bit RF Test
Attenuator
0 dB
attenuator
selected
4A24 J2
2 Position
J1
Diode J3
Switch
R265
4A T9
10 dB
Attenuator
R66 , R67
KD ) -12.52 dBm
RFD ) 12.20 dBm
CW ) 13.01 dBm
NOISE ) 5308.68 ENR
Cabinet
injection
to UD4
4DC2
Directional
Coupler
4.2 Matched Filter
The digital Matched filter has many
advantages: low filter power loss, adjustable
bandwidth and filter length, and DC Bias control.
Front
End
injection
to UD2
2A3
Receiver
Protector
UD4
R55
4A21
Microwave
Delay Line
Noise Te st Pa th
A 14-bit Analog to Digital (A/D) converter in
the IFD captures and digitizes the IF energy and
the transmit pulse burst sample at 72MHz.
Using a digital receiver eliminates the need for
AGC circuitry, DC Bias adjustment and I/Q
phase alignment. Digitized data is transferred
from the IFD to Receiver Circuit Card Assembly
(Rx CCA) in the RVP for matched filter
processing and conversion to I/Q data.
Configurable matched filtering provides
significant improvement over the fixed legacy
hardware matched filter, especially in long pulse
(4.57 µs) operation. The matched filter is
configured to duplicate the bandwidth of the
legacy, matched filter.
4.4 Test Path
Figure 3, ORDA Test Path
5.
RECEIVER PERFORMANCE
The ORDA receiver signal path meets or
exceeds legacy receiver path performance as
measured by the following parameters:
v
v
v
v
System Noise Level
System Dynamic Range
System Sensitivity
System Linearity
PARAMETERS
ORDA
LEGACY
System Noise
short pulse
long pulse
-80 dBm
-85 dBm
-57dBm
Dynamic range
93 dB
92 dB
Matched filter loss
-0.6dB
-1.5dB
-114dBm
-113dBm
Sensitivity
5.1 Noise Level
Many types of noise contribute to a
receiver, but for the WSR-88D the predominant
noise is thermal. Active components in the
receiver, namely, the Low Noise Amplifier and
the Mixer/Preamp, add phase noise, shot noise
and non-linearities. The IFD adds quantization
noise and sampling noise from the A/D
converter and the input clock. These noise
quantities are all orders of magnitude less than
the thermal noise contribution through the entire
transfer range of the receiver. Quantization
noise affects remain constant for the IFD as
opposed to the variations induced by the AGC
circuitry in the legacy receiver signal path.
Therefore, the thermal noise is the dominant
contributor to any variations in the measured
system noise floor.
The actual noise level measured by the
signal processor is based on the thermal noise
temperature at the front end, the system
bandwidth, receiver gain and the thermal noise
temperature contributed by the receiver
components.
NoisedBm = 10 log(kB(T ant + T Rx )) + g + 30
(1)
Where k is Boltzman’s Constant, B is the
receiver bandwidth, Tant is the thermal
temperature at the front end, TRx is the noise
contribution of the receiver and g is the receiver
gain from the receiver protector to the IFD.
With the 14-bit, 72MHz IFD, the noise level
with the input terminated at 50O is nominally
given as –85dBm/Mhz. With a 600KHz
bandwidth, this translates into a noise level of
approximately –87dBm (NIFD ) at the IFD input.
Terminated at the antenna, the system noise
floor in short pulse is approximately –81dBm
(NFE).
The Long pulse noise floor is
approximately –85dBm. The primary contributor
to the difference in the noise floor between short
pulse and long pulse is the differences in
bandwidth, 600KHz for short pulse and 200KHz
for long pulse.
5.2 Sensitivity
Sensitivity for the WSR-88D receiver is
referenced to the receiver front end input. It is
directly related to the noise floor so that a
quieter receiver will have a better sensitivity.
Sensitivity is therefore directly related to
thermal noise, matched filter bandwidth, and
receiver noise figure.
With the removal of analog IF and video
components, ORDA has a slightly improved
noise figure, 0.1 to 0.2dB. The digital matched
filter is better than the legacy’s analog filter,
gaining as much as 1dB. This means ORDA is
approximately 1dB more sensitive than legacy.
Sensitivity is measured for both systems at
the 0dB S/N ratio, where signal input power
equals the noise power, i.e., the point 3dB
above the noise floor.
This is defined as
Minimum Discernible Signal (MDS), but is not
actually the smallest coherent signal detectable.
As Figure 4 shows, the signal is easily
discernible to approximately 5-6dB below MDS.
5.3 Dynamic Range
The dynamic range is defined as the
difference between the MDS, where the S/N
ratio is 0dB, and the IFD’s 1dB compression
point, where the signal deviates 1dB from linear.
The legacy dynamic range was typically
measured to be 91dB to 92dB (Sirmans, 2000).
The IFD’s normal compression point is
+6dBm, giving a dynamic range of 87dB.
However, SIGMET uses a statistical linearization
technique for signals above compression,
thereby recovering another 6dB of signal. This
gives a dynamic range of 93dB to 94dB for the
ORDA receiver signal path.
Figure 4 shows the ORDA off-line linearity
and reflectivity test measurement display. This
calibration test computes the system noise floor,
compression point, minimum detectable signal,
linearity and dynamic range and shows the
results in a graphical window. This test was
done in short pulse at the KCRI channel 2 test
bed system in Norman, OK. As shown here, the
dynamic range from 0dB S/N to 1dB
compression is given as 95 dB.
6dB above the IFD noise floor, reducing the
dynamic range by 7dB and the sensitivity by
1dB.
Figure 5, Sensitivity-Dynamic Range Trade Off
Figure 4, Receiver Transfer Curve
5.4 Linear Transfer Curve
The linear slope of the receiver over the
IFD’s dynamic range is 1.00 (measured power
equals input power). Computation of the linear
slope is a performance check that confirms the
IF’s expected and measured values are within
tolerance.
Linear system response is vital to ensure
system accuracy. The excellent linearity of the
WSR-88D receiver is seen in Figure 4, where
the slope is 1.0003, the variance is a negligible
0.0053, and all the data points in the linear
region conform to the curve. The receiver
displays excellent linearity to within 1-2dB of
1dB compression. The low end, where the
noise floor affects the signal, shows expected
behavior with no anomalies.
5.5 Sensitivity vs. Dynamic Range
In a digital receiver, sensitivity and dynamic
range form a parametric relationship so that
optimizing for one reduces the other.
The tradeoff between sensitivity versus
dynamic range is based upon the noise ratio of
the receiver to the IFD (NFE-NIFD ). The WSR88D receiver has an excellent noise figure of
around 3dB. To maintain the WSR-88D
sensitivity, the IFD operates at approximately
5.6 Reliability
ORDA has replaced several legacy
components with COTS equipment. The new
components have a high mean time between
failure (MTBF), low noise figures, and costs
much
less
than
the
previous
legacy
components. The component reduction realized
with the ORDA architecture minimizes downtime
and provides higher reliability.
Removal of the analog components for
AGC and I/Q detection eliminates the receiver
channel alignment. All the receiver channel
setup is now done in software and does not
need to be done periodically, only upon
component replacement. This setup consists of
matched filter configuration for short pulse and
long pulse and burst pulse configuration.
A minor consideration for reliability is the
reduced power needed for the new receiver.
The new ORDA receiver components require
approximately 10 watts, compared to several
hundred for the removed legacy components.
5.7 Reflectivity Equation
The performance parameters determine
system calibration and its ability to accurately
measure reflectivity, velocity and spectrum
width. Reflectivity is computed with equation 3.
P −N 
dBZ = 10 log R
 + 20 log(R ) − A × R + dBZ 0 (3)
 N 
Where PR is the return signal power, N is
the Noise value corrected for elevation, R is
range, A is the two-way atmospheric loss and
dBZ0 is the system calibration constant,
computed using equation 4. dBZ0 represents the
reflectivity of a 0dB Signal-to-Noise target at a
range of 1km, and includes all the constants in
the radar equation (Rinehart, 1997).
The receiver has no effect on Range or
Atmospheric Loss, but its accuracy is critical for
the other values. The Return Signal Power
depends on the receiver’s linear response,
especially at low power levels where noise has a
large affect.
Accurate measurement of the Noise
depends upon the receiver’s ability to accurately
model the fluctuations in noise, so the proper
noise value is used in the equation. The dBZ0
calibration constant requires an accurate
measurement of the system’s MDS to determine
the conversion factor from power to reflectivity.
6.
CONCLUSION
ORDA implementation of Sigmet’s IFD and
RVP has improved the overall performance of
the NEXRAD radar. The replacement of several
legacy components with ORDA components has
reduced the system noise while increasing the
linearity and dynamic range. The new Sigmet
equipment has software adjustable filters, which
can closely match both pulse widths. This
provides for a better filter for eliminating
spurious signals and background noise from the
I and Q data.
7.
ACKNOWLEGEMENTS
The authors would like to thank Dale Sirmans
and William Urell for their support in writing this
paper. We would like to thank the Radar
Operations Center for their continued support
and for their assistance in evaluating the ORDA
design.
Note: The views expressed are those of the
author(s) and do not necessarily represent those
of the National Weather Service.
8.
REFERENCES
Free, A., Patel, N., and Heck, A., 2004: ORDA
Internal Report – ORDA System Calibration
Jim, G., and Free, A., 2004: ORDA Internal
Report – ORDA Receiver Path
Sirmans, D. and Urell, W., 2000: ROC Internal
Report – Digital Receiver Test Results
Operational Support Facility, 1992: Internal
Report – Calibration of the WSR-88D
Nathanson, F., 1969: Radar Design Principles:
Signal Processing and the Environment,
McGraw-Hill Book Company, New York
Rinehart, Ronald E., 1997: Radar for
rd
Meteorologists,
3
edition,
Rinehart
Publications, Columbia, MO.
Ice, R., McGehee, T., Rhoton, R., Saxion, D.,
Warde, D., Guenther, R., Sirmans, D., and
Rachel, D., 2005: Radar Operations Center
(ROC) Evaluation of New Signal Processing
st
Techniques
for
the
WSR-88D,
21
International AMS Conference on Interactive
Information and Processing Systems for
Meteorology, Oceanography, and Hydrology
Free, A., Heck, A., and Patel, N., 2005:
NEXRAD Open Radar Data Acquisition
st
(ORDA)
Receiver
Calibration,
21
International AMS Conference on Interactive
Information and Processing Systems for
Meteorology, Oceanography, and Hydrology
Patel, N. and Macemon, B., 2004: NEXRAD
Open Radar Data Acquisition (ORDA) Signal
th
Processing
and
Signal
Path,
20
International AMS Conference on Interactive
Information and Processing Systems for
Meteorology, Oceanography, and Hydrology
SIGMET, 2004: RVP8 User’s Manual