Three key physical layer performance indicators for JESD204B transmitters

As the JESD204 interface is increasingly adopted by data converters, there is a pressing need to focus on its performance and optimize the digital interface. The focus should not be solely on the performance of the data converter. The first two versions of the standard, JESD204 in 2006 and JESD204A in 2008, had a nominal data rate of 3.125 Gbps. The latest version, JESD204B, released in 2011, lists three speed grades with a maximum data rate of 12.5 Gbps. These three speed grades follow three different electrical interface specifications defined by the Optical Interconnect Forum (OIF). OIF-Sx5-01.0 defines the electrical interface specifications for data rates up to 3.125 Gbps, while CEI-6G-SR and CEI-11G-SR define the interface specifications for data rates up to 6.375 Gbps and 12.5 Gbps, respectively. High-speed data rates require more careful design and performance considerations for the high-speed CML drivers, receivers, and interconnect networks that make up the physical layer (PHY) of the JESD204B interface.

To evaluate the PHY performance of a JESD204B transmitter, several performance metrics need to be evaluated. These metrics include common-mode voltage, differential peak-to-peak voltage, differential impedance, differential output return loss, common-mode return loss, transmitter short-circuit current, eye diagram mask, and jitter.

This article will discuss three key performance metrics. These metrics are commonly used to evaluate transmitter signal quality, eye diagrams, bathtub plots, and histograms. Since the signal must be correctly decoded at the receiver, these measurements are also completed at the receiver. The eye diagram overlays multiple acquisition paths of the output data transmission to generate a curve that represents the link quality in multiple parameters. Many characteristics of the JESD204B physical interface can be observed through this curve, such as impedance discontinuities and improper terminations. This is only one way to evaluate the physical layer. The bathtub plot and histogram are two other important performance metrics that can be used to evaluate the quality of the JESD204B link. When measuring the unit interval (UI), the bathtub plot can visually represent the bit error rate (BER) for a given eye opening width. The unit interval is the time specified in the JESD204B physical layer specification that represents the time interval for data transmission. The third measurement data is the histogram, which represents the distribution of changes in the measured UI value. This measurement data can also represent the amount of jitter in the measured signal. The histogram, eye diagram, and bathtub plot can be used to represent the overall performance of the physical layer of the JESD204B interface. This example uses a JESD204B transmitter with an output data rate of 5.0 Gbps. The performance of the transmitter at this data rate is defined in detail by the OIF CEI-6G-SR specification.

Eye Diagram
Figure 1 shows the eye diagram of the JESD204B transmitter at a data rate of 5.0Gbps. The ideal waveform is superimposed with the measured waveform. Ideally, the transmission should be instantaneous with no overshoot or undershoot and without any ringing. In addition, there should be no jitter at the crossover point that determines the UI. As shown in Figure 1, it is impossible to obtain an ideal waveform in a real system because the signal is transmitted in a non-ideal medium with loss and imperfectly matched terminations. This eye diagram is measured at the receiver end of the JESD204B system. Before reaching the measurement point, the signal passes through the connector and a differential transmission line with a length of about 20cm. This eye diagram shows that the impedance match between the transmitter and the receiver is reasonable, the transmission medium is good, and there are no large impedance discontinuities. It does have some jitter, but it does not exceed the definition in the JESD204 interface specification. No overshoot is found in this eye diagram, but there is a slight undershoot on the rising edge due to slowing the signal transmission in the transmission medium. This is expected after the signal passes through the connector and 20cm differential transmission line. When there is a small amount of jitter in the signal, the UI average appears to match the expected UI value of approximately 200ps. In summary, this eye diagram indicates a good signal to the receiver, so there should be no problem recovering the embedded data clock and correctly decoding the data.

The eye diagram in Figure 2 is shown for the same transmission medium as used in Figure 1, except that the termination impedance is incorrect. The effect can be seen in the increased amount of signal jitter at the crossing points and in the non-transition regions. There is overall amplitude compression in many acquired data, causing the eye to begin to close. This signal degradation will increase the BER at the receiver and may result in a loss of the JESD204B link at the receiver if the eye closes beyond the receiver’s tolerance.

The eye diagram in Figure 3 shows another case of non-ideal data transmission. In this case, an impedance discontinuity is shown at a point between the transmitter and the receiver (in this case, an oscilloscope). The degradation in performance can be seen in the figure: the eye opening is closing, indicating that the area inside the transition point is getting smaller. The rising and falling edges of the data are severely degraded by the impedance discontinuity in the transmission line. The impedance discontinuity also causes an increase in the amount of jitter at the data transition point. Once the eye closure exceeds the receiver's ability to decode the data stream, the data link is lost. In this case of Figure 3, many receivers will likely be unable to decode the data stream.

In addition to the eye diagram, the bathtub plot
can also provide useful information about the serial data transmission on the JESD204B link. The bathtub plot measures the BER (bit error rate) as a function of the sampling point over time in the eye diagram. The bathtub plot is created by moving the sampling point across the eye diagram and measuring the BER at each point. As shown in Figure 4, the closer the sampling point is to the center of the eye diagram, the lower the BER. As the sampling point moves toward the transition points of the eye diagram, the BER increases. For a given BER, the distance between the two diagonal lines of the bathtub plot is the eye opening area for a specific BER (10-12 in this case).

The bathtub plot also provides information about the jitter (Tj) content of the signal. As shown in Figure 5, when the measurement point is close to or equal to the transition point, the jitter is relatively flat and mainly deterministic jitter. Like the eye diagram measurement, the bathtub plot is measured based on a JESD204B 5.0 Gbps transmitter, with the signal passing through the connector and about 20 cm of transmission line, and then measured at the receiver. As the measurement point moves toward the center of the eye opening, the dominant component of the jitter mechanism becomes random jitter. Random jitter is generated by a large amount of processing and is usually very small. Typical sources are: thermal noise, changes in wiring width, shot noise, etc. The PDF (probability density function) of random noise generally follows a Gaussian distribution. On the other hand, deterministic jitter generated by a small amount of processing can have large dimensions and can be correlated. The PDF of deterministic jitter is limited and has a well-defined peak-to-peak value. Its shape can change and is usually not Gaussian.

The expanded version of the bathtub plot discussed in Figure 4 is shown in Figure 6. For 5.0 Gbps serial data transmission and a BER of 10-12, the figure shows that the eye opening at the receiver is approximately 0.6 UI (unit interval). It is particularly important to note that similar to Figure 6: 5.0 Gbps bathtub plot.

The bathtub plot shown in the figure is an extrapolated measurement. The oscilloscope used to capture the data extrapolates the bathtub plot from a series of measurements. Using a BERT (bit error rate tester) and acquiring enough measurement data to create a bathtub plot can take hours or even days, even with the latest high-speed computing measurement equipment.

As with the eye diagram, improper termination or impedance discontinuities in the system can be detected using the bathtub plot. Compared to Figure 6, the bathtub plots in Figures 7 and 8 have a relatively flat slope on both ends. In this case, the eye opening is only 0.5 UI at a BER of 10-12, which is 10% lower than the good case of 0.6UI. Improper termination and impedance discontinuities cause a lot of random jitter in the system. The relatively flat slopes on both sides of the bathtub plot and the narrow eye opening at a BER of 10-12 indicate a lot of random jitter in the system. There is also a small increase in deterministic jitter. This is again confirmed by the decreasing slope near the edge of the bathtub plot.

Figure 8: 5.0Gbps bathtub plot – impedance discontinuity.

The
third useful measurement is the histogram. This plot shows the distribution of the measured spacing between transition points during data transmission. As with the eye and bathtub plot measurements, the histogram is measured at the receiver after the JESD204B 5.0Gbps transmitter has passed through the connector and approximately 20cm of transmission line. Figure 9 shows a histogram of a relatively good system at 5.0Gbps. The histogram shows that the measured spacing between 185ps and 210ps is roughly Gaussian. The expected spacing for a 5.0Gbps signal is 200ps, which means that the spacing in the plot is roughly distributed within the range of -7.5% to +5% on both sides of the expected value.

As shown in Figure 10, when an improper termination is applied, the distribution becomes wider and will vary between 170ps and 220ps. This will make the distribution percentage change to -15% to +10%, which is twice that of Figure 9. These graphs indicate that there is random jitter in the signal because they have a Gaussian-like shape. However, since these graphs are not truly Gaussian, it means that there is at least a small amount of deterministic jitter.

The histogram in Figure 11 shows the presence of an impedance discontinuity in the transmission line. The graph does not resemble a Gaussian distribution, but has a second, smaller peak. The average value of the measured period is also skewed. Unlike the waveforms in Figures 9 and 10, the average value is no longer 200ps, it is shifted to approximately 204ps. The more bimodal distribution indicates that there is more deterministic jitter in the system. This is due to the presence of an impedance discontinuity in the transmission line and the expected effects of this. The values ​​measured for the interval are not as wide as those for the improper termination, but the range is again widened. In this case the range is 175ps to 215ps, which is approximately -12.5% ​​to +7.5% on either side of the predicted interval. While the range is not as large, again the distribution is more bimodal in nature.

Conclusion
The physical layer performance of the JESD204B transmitter can be evaluated through some performance indicators, including common-mode voltage, differential peak-to-peak voltage, differential impedance, differential output return loss, common-mode return loss, transmitter short-circuit current, eye diagram template, and jitter. This article discusses three key performance indicators that can be used to evaluate the quality of the transmitted signal. Eye diagrams, bathtub plots, and histograms are the three important performance indicators used to evaluate the quality of the JESD204B link. System problems such as improper termination and impedance discontinuities can seriously affect the performance of the physical layer, and these effects can be observed through the graphical degradation shown in the eye diagram, bathtub plot, and histogram. It is very important to maintain good design practices to properly terminate the system and avoid impedance discontinuities in the transmission medium, because these problems can have a significant adverse effect on data transmission, resulting in data link failures between the JESD204B transmitter and receiver. If certain techniques are used to avoid these problems, the normal operation of the system can be ensured.

References:
JEDEC Standard: JESD204B (July 2011). JEDEC Solid State Technology Association. www.jedec.org
Application Note (5989-5718EN): Using Clock Jitter Analysis to Reduce BER in Serial Data Applications. Agilent Technologies, December 2006.
Application Note (5988-9109EN): Measurements in Digital Systems. Agilent Technologies, January 2008.