Texas Instruments | Understanding Serial LVDS Capture in High-Speed ADCs | Application notes | Texas Instruments Understanding Serial LVDS Capture in High-Speed ADCs Application notes

Texas Instruments Understanding Serial LVDS Capture in High-Speed ADCs Application notes
Understanding Serial LVDS Capture in
High-Speed ADCs
Application Report
Literature Number: SBAA205
July 2013
Contents
7
....................................................................................................................................... 4
Introduction ........................................................................................................................ 5
Capturing Data from a Serial LVDS Interface: Ideal Case ......................................................... 7
Receiver Capture Schemes ................................................................................................. 10
3.1
Latching Serialized ADC Data Bits into the S2P Shift Register ..................................................... 10
3.1.1 Using Delay Elements ........................................................................................... 10
3.1.2 Using PLLs ........................................................................................................ 11
3.1.3 PLL Clock Edge Selection Logic ............................................................................... 12
3.1.4 Determination of Correct Data .................................................................................. 12
3.2
Aligning Parallel Output Data from the S2P Shift Register .......................................................... 13
3.2.1 Determining if Captured Data are Aligned Correctly ......................................................... 16
3.2.2 Frame Alignment Logic Using SYNC Pattern ................................................................. 18
Summary of Capture Schemes ............................................................................................ 19
Understanding ADC Interface Timing Specifications .............................................................. 22
Achieving Timing Closure in the System .............................................................................. 25
6.1
Actual Setup Time ......................................................................................................... 26
6.2
Actual Hold Time .......................................................................................................... 27
6.3
PCB Skew .................................................................................................................. 28
Understanding Source Synchronous Interface ...................................................................... 29
2
Contents
Preface
1
2
3
4
5
6
SBAA205 – July 2013
Submit Documentation Feedback
Copyright © 2013, Texas Instruments Incorporated
www.ti.com
List of Figures
1-1.
LVDS Output Timing Diagram ............................................................................................. 5
1-2.
Typical Multichannel ADC with a Serial LVDS Interface ............................................................... 6
2-1.
Capturing Serial LVDS Data: Simplest Scheme ........................................................................ 7
2-2.
Double Data Rate IO ....................................................................................................... 8
2-3.
DDR IO Timing .............................................................................................................. 8
3-1.
Using Variable Delays in an FPGA
3-2.
PLLs in an FPGA with Multi-Phase Outputs ........................................................................... 11
3-3.
Multiple Phase Clocks at the PLL Output .............................................................................. 11
3-4.
Data Captured with Various Clock Edges .............................................................................. 12
3-5.
Data Alignment with Frame Clock: Ideal Case ........................................................................ 13
3-6.
Data Misalignment Resulting from a Delayed Frame Clock ......................................................... 14
3-7.
Data Misalignment Resulting from an Advanced Frame Clock...................................................... 15
3-8.
Using the SYNC Test Pattern to Determine Data Misalignment: Delayed Frame Clock......................... 16
3-9.
Using the SYNC Test Pattern to Determine Data Misalignment: Advanced Frame Clock
3-10.
Overall Frame Alignment Scheme ...................................................................................... 18
4-1.
Capture Scheme Using PLL: One-Wire Interface ..................................................................... 19
4-2.
Capture Scheme Using Delays: One-Wire Interface
4-3.
4-4.
4-5.
5-1.
5-2.
5-3.
5-4.
6-1.
6-2.
6-3.
7-1.
7-2.
7-3.
7-4.
.....................................................................................
......................
.................................................................
Capture Scheme for Two-Wire Interface ...............................................................................
Capture Scheme for Multiple ADC Devices using PLLs .............................................................
Capture Scheme for Multiple ADC Devices using Delays ...........................................................
Setup Time Definition .....................................................................................................
Hold Time Definition.......................................................................................................
Data Valid Time Definition ................................................................................................
Skew Among LVDS Outputs of the ADC ...............................................................................
Timing Analysis ............................................................................................................
Actual Setup Time .........................................................................................................
Actual Hold Time ..........................................................................................................
Jitter Source ................................................................................................................
Jitter in Data and Clock Paths ...........................................................................................
Eye Using an Internal Bit Clock (Case 1) and External Bit Clock (Case 2)........................................
Jitter in Data and Clock Paths Compared with an External or Ideal Clock ........................................
SBAA205 – July 2013
Submit Documentation Feedback
List of Figures
Copyright © 2013, Texas Instruments Incorporated
10
17
19
20
21
21
22
23
23
24
25
26
27
29
30
31
32
3
Preface
SBAA205 – July 2013
Abstract
Vinod Paliakara, Shantanu Prabhudesai ...................................... High-Performance Analog Products
This application note describes various schemes of interfacing serialized low-voltage differential signaling
(LVDS) data outputs from high-speed analog-to-digital converters (ADCs) to a field-programmable gate
arrays (FPGAs) or other application-specific integrated circuit (ASIC)-based receivers. This note provides
an introduction to standard one-wire interfaces and other interface variants (such as two-wire). This note
describes in detail the two key points required for reliable data capture: bit clock edge selection and frame
alignment. Schemes for capturing data from multiple ADC devices are also detailed. LVDS timing
parameters, as found in TI data sheets, are explained along with a detailed introduction to jitter in a
source-synchronous interface.
All trademarks are the property of their respective owners.
4
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
Chapter 1
SBAA205 – July 2013
Introduction
Among high-speed ADCs operating at tens of megahertz to hundreds of megahertz, the preferred output
data interface is serialized LVDS. This preference is especially true in multichannel ADCs such as TI’s
ADS528x, ADS529x, and ADS5263 families. The serialized LVDS interface is also seen in analog frontends (AFEs) that include a multichannel ADC, such as the AFE580x families. The differential signaling
protocol of an LVDS pair offers common-mode noise rejection, thus enabling higher data transmission
speeds (typically up to 1 Gbps per LVDS pair).
Figure 1-1 shows a typical serialized LVDS interface for one ADC channel operating on a sampling clock
frequency (fS). The characteristics of this interface are:
1. ADC data are output serially from D0 to DN-1 (for an N-bit resolution ADC) at every sampling clock
cycle. This output results in a serial data rate of (fS × N) bits per second.
2. An associated 50% duty cycle bit clock is output with a frequency of (fS × N / 2). The bit clock is
typically center-aligned and both clock edges can be used to latch serial ADC data. Therefore, the bit
clock is referred to as a double data rate (DDR) bit clock.
3. An associated 50% duty cycle frame clock is output with a frequency of fS. As the name suggests, the
frame clock rising edge transitions are aligned with the framing ADC data bits (D0 and DN-1). This
alignment helps the receiver to correctly load parallel data after de-serialization. Note that the frame
clock rising and falling edges are aligned with the transitions of data.
This interface is described as a one-wire, LSB-first interface with N-x serialization.
Input Clock
(CLK Frequency = fS)
Frame Clock
(ADCLK Frequency = 1x fS)
Bit Clock
(LCLK Frequency = 6x fS)
Output Data(1)
(OUTA Data rate = 12x fS)
D11
(D0)
D10
(D1)
D9
(D2)
D8
(D3)
D7
(D4)
D6
(D5)
D5
(D6)
D4
(D7)
D3
(D8)
D2
(D9)
D1
(D10)
D0
(D11)
D11
(D0)
D10
(D1)
Sample N
Sample N+1
(1)
The upper data bit is the MSB-first mode data bit and the lower data bit is the LSB-first mode data bit.
Figure 1-1. LVDS Output Timing Diagram
Some variations of this interface include:
1. Two-wire interface: At high sampling clock frequencies, the serial data rate becomes very high. In
such cases, the ADC data are output serially over two LVDS pairs. For example, bits D0 to D(N/2)-1
are output over one LVDS pair and the other bits D(N/2) to DN-1 over the second pair. With half the
number of bits transmitted per LVDS pair on each sampling clock cycle, the serial data rate is reduced
to (fS × N / 2) bits per second.
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
5
www.ti.com
2. MSB-first: the ADC data bits are serially transmitted from the MSB bit (DN-1) to the LSB bit, D0.
An advantage of the serial LVDS interface is the savings in pin-count resulting from the use of only two
pins (one LVDS pair) per ADC channel. Therefore, this interface is most commonly found in multichannel
ADC devices. A typical multichannel device has one (or two) LVDS pairs per ADC channel, one common
bit clock output, and one frame clock output. Refer to Figure 1-2 for the LVDS output interface of an 8channel ADC device.
OUT1A_P
IN1P
IN1N
Sampling
Circuit
12-Bit ADC
Digital Processing
Block
OUT1A_N
Serializer
OUT1B_P
OUT1B_N
OUT2A_P
IN2P
IN2N
Sampling
Circuit
12-Bit ADC
Digital Processing
Block
OUT2A_N
Serializer
OUT2B_P
OUT2B_N
OUT3A_P
IN3P
IN3N
Sampling
Circuit
12-Bit ADC
Digital Processing
Block
OUT3A_N
Serializer
OUT3B_P
OUT3B_N
OUT4A_P
IN4P
IN4N
Sampling
Circuit
12-Bit ADC
Digital Processing
Block
OUT4A_N
Serializer
OUT4B_P
OUT4B_N
LCLKP
LCLKN
CLKP
CLOCKGEN
CLKN
PLL
ADCLKP
ADCLKN
SYNC
OUT5A_P
IN5P
IN5N
Sampling
Circuit
12-Bit ADC
Digital Processing
Block
OUT5A_N
Serializer
OUT5B_P
OUT5B_N
OUT6A_P
IN6P
IN6N
Sampling
Circuit
12-Bit ADC
Digital Processing
Block
OUT6A_N
Serializer
OUT6B_P
OUT6B_N
OUT7A_P
IN7P
IN7N
Sampling
Circuit
12-Bit ADC
Digital Processing
Block
OUT7A_N
Serializer
OUT7B_P
OUT7B_N
OUT8A_P
IN8P
IN8N
Sampling
Circuit
12-Bit ADC
Digital Processing
Block
OUT8A_N
Serializer
OUT8B_P
OUT8B_N
Control
Interface
Reference
SDOUT
Device
Figure 1-2. Typical Multichannel ADC with a Serial LVDS Interface
6
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
Chapter 2
SBAA205 – July 2013
Capturing Data from a Serial LVDS Interface: Ideal Case
The simplest capture scheme to receive data from a serial interface consists of a double data rate (DDR)
logic block followed by a serial-to-parallel shift register, as shown in Figure 2-1. The length of the shift
register chain must be at least (N / 2) flip-flops. At every frame clock rising edge, the parallel data output
from both shift register chains are latched. The output is a frame-aligned parallel data word of N-bits (D0
to DN-1).
D10
Serial Data Input
OUTn
D
Q Rising
D
Q Falling
Bit Clock
LCLK
Q
D8
D
CLK
Q
D6
D
CLK
D4
Q
D
CLK
Q
D2
D
CLK
Q
D0
D
CLK
Q
CLK
CLKOUT
D11
D
CLK
Q
D9
D
CLK
Q
D7
D
D5
Q
D
CLK
Q
CLK
D2
D
CLK
Q
D1
D
Q
CLK
D0
D[11:0]
DATA[11:0]
D
ADCLK
D3
Q
CLK
Figure 2-1. Capturing Serial LVDS Data: Simplest Scheme
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
7
www.ti.com
The DDR block latches input serial data at both edges of the bit clock and outputs two data streams: one
corresponding to the rising edge of the bit clock (Q rise) and the other corresponding to the falling edge of
the bit clock (Q fall, as shown in Figure 2-2). The rising and falling data streams are then applied to the
shift register.
Serial Data Input
OUTn
D
Q
D
Q
Data Out Rising
Bit Clock
LCLK
D
Q
Data Out Falling
Data Out Falling
Relatched
DDR IO
Output Clock
Figure 2-2. Double Data Rate IO
Note that the DDR block re-latches the data stream falling edge at the next bit clock rising edge and
outputs the rising and falling edge data streams at the rising bit clock edge, which is used by all the flipflops in the shift register chain. As a result, data are valid for roughly half a bit clock period at the DDR IO
input, whereas data are valid for almost one bit clock period for the remaining flops in the shift register
chain. Figure 2-3 shows a DDR IO timing diagram.
Data are valid for approximately
a half bit clock period.
Data are valid for approximately
one bit clock period.
Figure 2-3. DDR IO Timing
This difference implies that the timing constraint for the flip-flops in the DDR block alone is critical
compared to the rest of the flip-flops.
8
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
www.ti.com
Most FPGAs have a DDR flip-flop and register as part of the logic library. The DDR IO element accepts a
single clock and registers data at the input data pin at both clock edges. The DDR IO element is also
physically located close to the FPGA receiver pins, thus helping minimize any delays caused by routing in
the FPGA. Therefore, TI recommends using the DDR IO element to interface to the ADC serial LVDS
interface.
Ideally, there is no additional skew between the serial LVDS data, LVDS bit clock, and LVDS frame clock
signals caused by delays in the PCB and FPGA internal routing. In a real situation, the effects of trace
delays in the PCB and FPGA routing cannot be ignored.
Chapter 3 and Chapter 4 describe various implementations for capturing data from a serialized ADC
interface in a real-world situation.
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
9
Chapter 3
SBAA205 – July 2013
Receiver Capture Schemes
Any capture solution must implement two major functions:
1. Latching the serialized ADC data bits correctly into the serial-to-parallel (S2P) shift register using the
ADC bit clock (serialized clock domain), and
2. Aligning the parallel output data from the S2P registers correctly (parallel clock domain).
Various methods of implementing these functions are discussed in the Latching Serialized ADS Data Bits
into the S2P Shift Register, Aligning Parallel Output Data from the S2P Shift Register, Determining if
Captured Data are Aligned Correctly, and Frame Alignment Logic Using SYNC Pattern sections.
3.1
Latching Serialized ADC Data Bits into the S2P Shift Register
As explained previously, a real-world application has delays resulting from PCB traces and FPGA routing.
These delays must be taken into account during timing analysis. In many cases (especially for data rates
> 600 Mbps), meeting the timing requirements with these effects can be difficult. For such cases, one
common solution is using a delayed version of the ADC bit clock for latching data into the shift register.
The delay can be accomplished by:
• Using delay elements (external to the ADC and receiver OR built into the receiver itself), and by
• Using phase-locked loops (PLLs)
3.1.1 Using Delay Elements
An important consideration when using this scheme is the delay variation across process, supply voltage,
and temperature. Also, ensure that timing is met at both minimum and maximum delay values. With these
considerations in mind, using circuits such as delay-locked loops (DLLs) is preferable to generate delays
that do not vary with process, supply voltage, and temperature. Many FPGAs have built-in DLL elements
that can be used to provide delays within certain ranges (up to a few hundred picoseconds). Figure 3-1
shows a diagram of variable delays used in an FPGA.
Odd Sample Bit
Serial Data Input
OUTn
IDELAY
LVDS
IDDR
Even Sample Bit
Bit Clock
LCLK
LVDS
IDELAY
BUFIO
Sample Clock
Figure 3-1. Using Variable Delays in an FPGA
10
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
Latching Serialized ADC Data Bits into the S2P Shift Register
www.ti.com
3.1.2 Using PLLs
An alternate scheme for deriving a delayed bit clock makes use of a PLL with multiple delayed edges in
each clock cycle. Figure 3-2 shows a case where eight equally-spaced clock edges are available within
each bit clock cycle. In the receiver, additional logic is required for selecting the correct clock edge. The
edge selection logic must run when the ADC interface initializes after power-up and when the bit clock
output is stable. Figure 3-3 shows the timing of multiple phase clocks at the PLL output.
2p
PLL
LCLK
0
p/4
p/2
3p / 4
p
5p / 4
3p / 2
7p / 4
Figure 3-2. PLLs in an FPGA with Multi-Phase Outputs
Figure 3-3. Multiple Phase Clocks at the PLL Output
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
11
Latching Serialized ADC Data Bits into the S2P Shift Register
www.ti.com
3.1.3 PLL Clock Edge Selection Logic
A typical implementation of the edge selection logic functions by capturing the ADC data for all PLL output
clock edges. For data captured at each edge, the logic determines if data are correct or incorrect. If data
are correct, the corresponding clock edge can be used for latching serial data.
3.1.4 Determination of Correct Data
How are captured data determined to indeed be correct? All TI high-speed ADCs have a set of test
patterns that can be enabled to implement receiver-capture logic. For clock edge selection, a suitable test
pattern is termed deskew. When enabled, the ADC serial LVDS interface outputs a data bit sequence of
alternating '1's and '0's. When captured with the correct clock edge, the expected data are either
101010101010 or 010101010101 for a 12-bit LVDS interface. Figure 3-4 shows an example of captured
data for various clock edges. Note how the captured data are correct for a range of clock edges from 3 to
6. The robustness of the scheme can be improved by selecting the clock edge at the center of the good
clock edge range, in this case that would be clock edges 4 or 5. This selection provides a margin for any
timing changes caused by temperature and supply voltage variations.
Figure 3-4. Data Captured with Various Clock Edges
Note that whenever the ADC input clock is interrupted, the PLL in the receiver is no longer locked. When
the ADC input clock is restarted, the receiver PLL locks again and the output edges stabilize after the PLL
lock time. Most PLLs have a reset signal that ensures that clock edge 0 has a fixed relationship with the
input clock (either aligned, 90° phase shifted, or so forth). Executing the clock selection logic in the
receiver is not required every time the PLL reacquires lock. The same clock edge identified earlier by the
logic (clock edge 5 from the example in Figure 3-4) can be reused. TI recommends executing the edge
selection logic whenever there is a significant change in temperature, supply voltage, or bit clock
frequency.
Other schemes for determining the correct clock edge from captured data also exist. Many FPGAs include
a dynamic clock edge selection scheme that runs continuously in the background and does not require
any test patterns from the ADC. By running in the background, such a scheme can make dynamic clock
edge changes based on temperature and supply variations.
12
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
Aligning Parallel Output Data from the S2P Shift Register
www.ti.com
3.2
Aligning Parallel Output Data from the S2P Shift Register
After the serialized ADC data bits are latched correctly by the shift register using one of the previous
schemes, the next point to consider is the alignment of the parallel data output of the shift register. As
explained in the Latching Serialized ADS Data Bits into the S2P Shift Register section, ideally the ADC
frame clock output can be directly used to latch the parallel data output of the shift registers. This data
output is at the frame clock rising edge because the frame clock rising edge is aligned with the start of the
parallel word boundary.
In an actual case, the frame clock can be delayed (or advanced) with respect to the serialized output data
outputs. For example, this delay can occur as a result from skews in the PCB traces of the frame and
output data lines. However, such skews are reasonably easy to address and minimize during PCB layout.
Skews are more likely to occur inside receivers. This occurrence is especially true when using FPGAs as
receivers and also when the serial data outputs from multiple ADC devices are routed to one FPGA
receiver.
The effect of a skewed frame clock on the final captured data using the ideal receiver scheme is described
in Figure 3-5 and Figure 3-6. Note that for skews less than (1/2) a bit clock period, the captured data are
correct and include all bits from the same sample (see Figure 3-7). For larger skews, the captured data
consist of a mix of bits from two successive samples.
Figure 3-5. Data Alignment with Frame Clock: Ideal Case
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
13
Aligning Parallel Output Data from the S2P Shift Register
www.ti.com
Figure 3-6. Data Misalignment Resulting from a Delayed Frame Clock
14
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
Aligning Parallel Output Data from the S2P Shift Register
www.ti.com
Figure 3-7. Data Misalignment Resulting from an Advanced Frame Clock
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
15
Aligning Parallel Output Data from the S2P Shift Register
www.ti.com
3.2.1 Determining if Captured Data are Aligned Correctly
Certain test patterns can be advantageously used to determine if captured data are aligned properly. TI
ADCs have a SYNC pattern that consists of a sequence of six '1's followed by six '0's (for a 12-bit
serialized ADC). When enabled, each serial data output consists of this pattern. Note that when displayed
on an oscilloscope, this pattern resembles a frame clock. Figure 3-8 and Figure 3-9 show the effect of a
skewed frame clock when the SYNC test pattern is enabled.
Figure 3-8. Using the SYNC Test Pattern to Determine Data Misalignment: Delayed Frame Clock
16
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
www.ti.com
Aligning Parallel Output Data from the S2P Shift Register
Figure 3-9. Using the SYNC Test Pattern to Determine Data Misalignment: Advanced Frame Clock
Using the SYNC pattern, the Frame Alignment Logic Using SYNC Pattern section describes a scheme to
correctly align the parallel output of the S2P shift register.
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
17
Aligning Parallel Output Data from the S2P Shift Register
www.ti.com
3.2.2 Frame Alignment Logic Using SYNC Pattern
A common logic for aligning to the correct frame boundary is described in this section. Here, the parallel
output of the S2P shift registers is first latched using the (skewed) frame clock. This note refers to the
latched data as S2P_PAR_OUT. The alignment logic requires the SYNC test pattern in the ADC to be
enabled. With the SYNC test pattern enabled in the ADC, the S2P_PAR_OUT data are compared with the
ideal (or expected) data, which is 111111000000 for a 12-bit serialized ADC. To correct for the effect of
the frame clock skew, two successive samples of S2P_PAR_OUT are required. The correction and
determination of the final aligned output (termed FINAL_OUT) is described by Figure 3-10.
To summarize Figure 3-10:
• For no skew, FINAL_OUT[11:0] = S2P_PAR_OUT[23:12];
• For one bit clock period positive skew (shown in Figure 3-8),
– FINAL_OUT[11:0] = S2P_PAR_OUT[22:11]; and
• For two bits clock period positive skew,
– FINAL_OUT[11:0] = S2P_PAR_OUT[21:10], and so on.
• Similarly, for one bit clock period negative skew (shown in Figure 3-9),
– FINAL_OUT[11:0] = S2P_PAR_OUT[12:1]: and
• For two bits clock period negative skew,
– FINAL_OUT[11:0] = S2P_PAR_OUT[13:2], and so on.
Figure 3-10. Overall Frame Alignment Scheme
Note that the skew between the frame clock and serial ADC data output can be different across channels.
The logic shown in Figure 3-10 determines the skew for every channel and corrects for it.
18
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
Chapter 4
SBAA205 – July 2013
Summary of Capture Schemes
Figure 4-1 and Figure 4-2 show block diagrams of the two capture schemes described in the Receiver
Capture Schemes section. The block diagram is shown for one channel and is applicable for the one-wire
interface.
Figure 4-1. Capture Scheme Using PLL: One-Wire Interface
Figure 4-2. Capture Scheme Using Delays: One-Wire Interface
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
19
www.ti.com
Figure 4-3 shows the block diagram for a two wire interface. Here, the capture scheme for each wire is
treated identical to that of a one wire interface. After the data from each wire is frame aligned, the data
from both wires are combined and output. Note that while a common PLL can be used for both wires,
each wire includes an independent bit align block to account for different skews in each wire.
Figure 4-3. Capture Scheme for Two-Wire Interface
The capture schemes in Figure 4-1 to Figure 4-3 are valid even multiple multichannel ADCs are used in
the system. Figure 4-4 and Figure 4-5 describe two recommended ways of capturing data from multiple
ADC devices. Specifically, note that the individual bit clock output from each ADC is used by the
respective capture module. This architecture provides better setup and hold timing compared to using a
common bit clock across all devices. (See the Understanding Source Synchronous Interface section).
Therefore, the FPGA code consists of a dedicated capture module per ADC device (supporting all eight
channels within one device). The output of the capture module is a parallel data bus corresponding to the
eight channels of one ADC device, running from a 1x clock derived within the module. For example, this
bus is 96-bits wide (12x8 bits) for a 12-bit, 8-channel ADC device.
Note that the 1x clock used within the capture modules are not aligned together. This misalignment is
primarily because of the various input clock to output clock delays across the ADC devices.
The parallel data output from all capture modules are then latched using a common system clock that runs
at 1x frequency (or at the same frequency as the ADC sample rate). Note that for multiple devices, TI
recommends using a common system clock to combine the parallel data from all devices, rather than
using a common high-speed clock (bit clock) to combine the serial data from all devices. When multiple
devices are used, using the individual bit clock output from each device (as explained in the
Understanding Source Synchronous Interface section) is advantageous.
While combining the parallel data from all devices, the system clock frequency is at the same frequency
(or 1x) as the ADC sample rate. At this low frequency, a sufficient timing margin is available even if the
parallel data outputs from multiple devices are misaligned.
20
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
www.ti.com
Figure 4-4. Capture Scheme for Multiple ADC Devices using PLLs
Figure 4-5. Capture Scheme for Multiple ADC Devices using Delays
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
21
Chapter 5
SBAA205 – July 2013
Understanding ADC Interface Timing Specifications
TI high-speed ADC data sheets specify the LVDS interface setup and hold times measured with respect to
the output bit clock. See the Understanding Source Synchronous Interface section for further information
regarding the advantages of using the output bit clock from each device.
Setup time is the time delay from the instant when data becomes valid to the rising or falling bit clock
edge. The minimum value of this time delay is specified as the minimum setup time in the data sheet.
Refer to Figure 5-1 for the typical and minimum setup times.
(1)
Refer to the tSU minimum specification in the Timing Characteristics table of the device data sheet.
(2)
Refer to the tSU typical specification in the Timing Characteristics table of the device data sheet.
Figure 5-1. Setup Time Definition
22
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
www.ti.com
Hold time is the time delay from the rising or falling bit clock edge to the instant when data becomes
invalid. The minimum value of this time delay is specified as the minimum hold time in the data sheet.
Refer to Figure 5-2 for the typical and minimum hold times.
(1)
Refer to the tHO minimum specification in the Timing Characteristics table of the device data sheet.
(2)
Refer to the tHO typical specification in the Timing Characteristics table of the device data sheet.
Figure 5-2. Hold Time Definition
Data eye is the solid blue box in Figure 5-3 and represents the amount of time for which the data logic
levels are valid. Having as wide and tall an eye as possible is preferable.
Figure 5-3. Data Valid Time Definition
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
23
www.ti.com
Referring to Figure 5-1 to Figure 5-3, data are considered valid for logic levels greater than or equal to
±100 mV and are invalid otherwise. The waveforms in Figure 5-1 and Figure 5-2 can be displayed on an
oscilloscope by probing the serial output data and bit clock in infinite persistence mode and triggering on
the output bit clock. The resulting serial data waveform is also referred to as an eye. Note that the serial
data waveform crosses the ±100-mV thresholds at multiple points, which implies that the setup time (or
hold time) is sometimes large or small at certain times. This effect is a result of jitter in the output data and
bit clock. The minimum value of the setup time (or hold time) is considered as the worst-case value and is
the specified value in the TI data sheet.
Ideally, the setup time for all LVDS pairs should be identical. Though care is taken to match delays across
LVDS outputs within the die, there is skew among LVDS channels caused by mismatches within the die
and within the package because of asymmetric pin locations. As a result, the setup and hold times are
different across LVDS channels. Figure 5-4 shows an example of an 8-channel ADC with four LVDS
outputs. Note that the channel 3 output (OUT3) has the worst setup time among all channels. The
minimum setup time on OUT3 is specified in the data sheet. Similarly, note that the channel 4 output
(OUT4) has the worst hold time among all channels. The minimum hold time on OUT4 is likewise
specified in the data sheet.
Figure 5-4. Skew Among LVDS Outputs of the ADC
To summarize, the minimum setup and hold specifications in the data sheet already take into account
effects of jitter in output data and clock and skew across LVDS outputs.
24
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
Chapter 6
SBAA205 – July 2013
Achieving Timing Closure in the System
A typical system has multiple ADCs connected to the receiver using one of the schemes described in the
Summary of Capture Schemes section. Timing analysis methods are used to verify if all data outputs
(from the ADCs) satisfy the setup and hold timing specifications of the receiver. Timing analysis takes into
account:
• The setup and hold specifications of the transmitter (in this case, the ADC).
• Delays in the data path, which are composed of PCB trace delays and internal delays in the receiver
itself.
• Delays in the clock path, which are also composed of PCB trace delays and internal delays in the
receiver itself.
• Minimum setup and hold specification of the receiver flip-flop (tSU_RX, min and tHO_RX, min).
Figure 6-1 shows a model for the timing analysis that includes the additional delays resulting from PCB
traces and receiver internal routing. For the receiver flip-flop to latch data correctly, the setup and hold
times of the input data at the receiver must satisfy Equation 1 and Equation 2.
Actual Setup Time (Equal to Clock Arrival Time – Data Valid Time) > tSU_RX, min
Actual Hold Time (Equal to Data Invalid Time – Clock Arrival Time) > tHO_RX, min
(1)
(2)
Figure 6-1. Timing Analysis
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
25
Actual Setup Time
6.1
www.ti.com
Actual Setup Time
The actual setup time can be derived using Equation 3 to Equation 5 and by referring to Figure 6-2.
Clock Arrival Time = tSU_ADC + tPCB_CLK + tINT_CLK = tSU_ADC + tDEL_CLK
Data Valid Time = tPCB_DATA + tINT_DATA = tDEL_DATA
Actual Receiver Setup Time = tSU_ADC + tDEL_CLK – tDEL_DATA = tSU_ADC – tSKEW
(3)
(4)
(5)
The worst-case setup time occurs with minimum ADC setup time, minimum clock delay, and maximum
data delay, as described in Equation 6.
Worst-Case Setup Time at Receiver = tSU_ADC (min) – tSKEW (max)
where:
•
tSKEW (max) = tDEL_DATA (max) – tDEL_CLK (min)
(6)
Usually, data path delays are larger than clock path delays. In such a case, tSKEW is a positive number and
Equation 6 shows that the setup time at the receiver is less than the setup time provided by the ADC.
Therefore, TI recommends minimizing data path delays (or matching the data and clock path delays) to
minimize tSKEW.
Figure 6-2. Actual Setup Time
26
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
Actual Hold Time
www.ti.com
6.2
Actual Hold Time
The actual hold time can be derived using Equation 7 to Equation 9 and by referring to Figure 6-3.
Data Invalid Time = tHO_ADC + tPCB_DATA + tINT_DATA = tHO_ADC + tDEL_DATA
Clock Arrival Time = tPCB_CLK + tINT_CLK = tDEL_CLK
Actual Hold Time at Receiver = tHO_ADC + tDEL_DATA – tDEL_CLK = tHOD_ADC + tSKEW
(7)
(8)
(9)
The worst-case hold time occurs with minimum ADC hold time, minimum data delay, and maximum clock
delay, as described in Equation 10.
Worst-Case Hold Time at Receiver = tHOD_ADC (min) + tSKEW (min)
where:
•
tSKEW (min) = tDEL_DATA (min) – tDEL_CLK (max)
(10)
Again, Equation 10 shows that minimizing data path delays (or matching the data and clock path delays)
in order to minimize tSKEW is preferable.
Figure 6-3. Actual Hold Time
The analysis for actual setup and hold times must be done for all the data lines. Timing analysis can be
referred to as closed when the actual setup (and hold) time at the receiver is greater than the minimum
setup (and hold) time specification of the receiver for every data input.
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
27
PCB Skew
6.3
www.ti.com
PCB Skew
As discussed previously, minimizing the skew between the output data and output clock signals is
advantageous. One major component of the skew results from PCB trace delays, see Figure 6-1.
Minimizing PCB skew is especially important for higher data rates. PCB traces result in 150 ps to 200 ps
per inch of trace length depending on the type of traces used (micro-strip versus strip line). For data rates
greater than 500 Mbps, TI recommends matching the PCB trace lengths for data and clock signals to
within 100 mils of each other. This match ensures that the skew is low (approximately 15 ps to 20 ps) and
maximizes the setup and hold times at the receiver.
28
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
Chapter 7
SBAA205 – July 2013
Understanding Source Synchronous Interface
The serial LVDS interface consists of serial data outputs (one or two outputs per ADC channel)
accompanied by a bit clock output (refer to Figure 1-1). This interface, in which the clock is transmitted
along with data, is referred to as a source-synchronous interface.
Source-synchronous interfaces are especially beneficial at high data rates, whereby matching output data
and output clock signal routing, any skews between them can be minimized. Another important benefit of
this interface is with respect to the effect of jitter. Jitter is the instantaneous deviation of the rising and
falling edges of data and clock signals. Jitter is primarily caused by supply noise, where vertical noise is
translated to horizontal or time-instant deviations, depending on the signal rise and fall times, as shown in
Figure 7-1.
In a serial LVDS transmitter, both output data and clock signals have jitter. Jitter has two components:
correlated and uncorrelated. By carefully matching clock and data paths, the data and clock jitter can be
ensured to be largely correlated.
Figure 7-1. Jitter Source
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
29
www.ti.com
Figure 7-2 shows the effect of correlated jitter where the instantaneous time deviations of the output data
and clock signals are identical. Thus, when the data transition is delayed (the red line), the clock edge is
also delayed similarly and when the data transition is advanced (the green line), the clock edge also
arrives earlier compared to the nominal or ideal time instants (the black lines).
Figure 7-2. Jitter in Data and Clock Paths
When the output data timing is measured relative to the output clock, the correlated (or common) jitter
component in the output data and bit clock cancels, resulting in a larger eye (or available valid data time).
Effectively, this cancellation results in better setup and hold timing specifications when measured with
respect to the output clock.
30
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
www.ti.com
Figure 7-3 shows the data eye for two cases.
Figure 7-3. Eye Using an Internal Bit Clock (Case 1) and External Bit Clock (Case 2)
Case 1: In this case, the eye is the result when output data are measured by triggering the oscilloscope on
the output bit clock of the ADC (see Figure 7-2). As explained before, the jitter is largely correlated and
explains the resultant wide eye. This case is the condition under which TI specifies the output LVDS timing
parameters of setup and hold times.
SBAA205 – July 2013
Submit Documentation Feedback
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
31
www.ti.com
Case 2: In this case, the eye is the result when output data are measured by triggering the oscilloscope
using an external bit clock that may be generated, for example, using an external PLL from the input
clock, as shown in Figure 7-4. In this case, the correlated component of jitter is very small. The eye is
noticeably smaller compared to the first case because most of the jitter is uncorrelated. This case
effectively results in worse setup and hold timings when output data are measured with respect to the
external bit clock.
The main conclusion from this discussion is that using the output bit clock from the ADC to latch the
serialized output data is recommended, especially at high data rates, because of the wide eye available
with this scheme.
Figure 7-4. Jitter in Data and Clock Paths Compared with an External or Ideal Clock
32
Understanding Serial LVDS Capture in High-Speed ADCs
Copyright © 2013, Texas Instruments Incorporated
SBAA205 – July 2013
Submit Documentation Feedback
IMPORTANT NOTICE
Texas Instruments Incorporated and its subsidiaries (TI) reserve the right to make corrections, enhancements, improvements and other
changes to its semiconductor products and services per JESD46, latest issue, and to discontinue any product or service per JESD48, latest
issue. Buyers should obtain the latest relevant information before placing orders and should verify that such information is current and
complete. All semiconductor products (also referred to herein as “components”) are sold subject to TI’s terms and conditions of sale
supplied at the time of order acknowledgment.
TI warrants performance of its components to the specifications applicable at the time of sale, in accordance with the warranty in TI’s terms
and conditions of sale of semiconductor products. Testing and other quality control techniques are used to the extent TI deems necessary
to support this warranty. Except where mandated by applicable law, testing of all parameters of each component is not necessarily
performed.
TI assumes no liability for applications assistance or the design of Buyers’ products. Buyers are responsible for their products and
applications using TI components. To minimize the risks associated with Buyers’ products and applications, Buyers should provide
adequate design and operating safeguards.
TI does not warrant or represent that any license, either express or implied, is granted under any patent right, copyright, mask work right, or
other intellectual property right relating to any combination, machine, or process in which TI components or services are used. Information
published by TI regarding third-party products or services does not constitute a license to use such products or services or a warranty or
endorsement thereof. Use of such information may require a license from a third party under the patents or other intellectual property of the
third party, or a license from TI under the patents or other intellectual property of TI.
Reproduction of significant portions of TI information in TI data books or data sheets is permissible only if reproduction is without alteration
and is accompanied by all associated warranties, conditions, limitations, and notices. TI is not responsible or liable for such altered
documentation. Information of third parties may be subject to additional restrictions.
Resale of TI components or services with statements different from or beyond the parameters stated by TI for that component or service
voids all express and any implied warranties for the associated TI component or service and is an unfair and deceptive business practice.
TI is not responsible or liable for any such statements.
Buyer acknowledges and agrees that it is solely responsible for compliance with all legal, regulatory and safety-related requirements
concerning its products, and any use of TI components in its applications, notwithstanding any applications-related information or support
that may be provided by TI. Buyer represents and agrees that it has all the necessary expertise to create and implement safeguards which
anticipate dangerous consequences of failures, monitor failures and their consequences, lessen the likelihood of failures that might cause
harm and take appropriate remedial actions. Buyer will fully indemnify TI and its representatives against any damages arising out of the use
of any TI components in safety-critical applications.
In some cases, TI components may be promoted specifically to facilitate safety-related applications. With such components, TI’s goal is to
help enable customers to design and create their own end-product solutions that meet applicable functional safety standards and
requirements. Nonetheless, such components are subject to these terms.
No TI components are authorized for use in FDA Class III (or similar life-critical medical equipment) unless authorized officers of the parties
have executed a special agreement specifically governing such use.
Only those TI components which TI has specifically designated as military grade or “enhanced plastic” are designed and intended for use in
military/aerospace applications or environments. Buyer acknowledges and agrees that any military or aerospace use of TI components
which have not been so designated is solely at the Buyer's risk, and that Buyer is solely responsible for compliance with all legal and
regulatory requirements in connection with such use.
TI has specifically designated certain components as meeting ISO/TS16949 requirements, mainly for automotive use. In any case of use of
non-designated products, TI will not be responsible for any failure to meet ISO/TS16949.
Products
Applications
Audio
www.ti.com/audio
Automotive and Transportation
www.ti.com/automotive
Amplifiers
amplifier.ti.com
Communications and Telecom
www.ti.com/communications
Data Converters
dataconverter.ti.com
Computers and Peripherals
www.ti.com/computers
DLP® Products
www.dlp.com
Consumer Electronics
www.ti.com/consumer-apps
DSP
dsp.ti.com
Energy and Lighting
www.ti.com/energy
Clocks and Timers
www.ti.com/clocks
Industrial
www.ti.com/industrial
Interface
interface.ti.com
Medical
www.ti.com/medical
Logic
logic.ti.com
Security
www.ti.com/security
Power Mgmt
power.ti.com
Space, Avionics and Defense
www.ti.com/space-avionics-defense
Microcontrollers
microcontroller.ti.com
Video and Imaging
www.ti.com/video
RFID
www.ti-rfid.com
OMAP Applications Processors
www.ti.com/omap
TI E2E Community
e2e.ti.com
Wireless Connectivity
www.ti.com/wirelessconnectivity
Mailing Address: Texas Instruments, Post Office Box 655303, Dallas, Texas 75265
Copyright © 2013, Texas Instruments Incorporated
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

advertising