October 4, 2016
As high-speed serial data rates advance beyond 50 Gb/s, NRZ signaling is being replaced by PAM4. While the modulation scheme of PAM4 takes half the bandwidth to transmit the same payload as an equivalent NRZ signal, it does pose design challenges, particularly when it comes to test. Using the proper vector network analyzer (VNA) can solve these issues.
The bottom line is that PAM4 requires more accurate measurements. Designs have to be verified before they’re produced. Simulations and measurements have to agree on a point-by-point basis. Since SNR is about 10 dB worse for PAM4 than NRZ, qualitative agreement is no longer acceptable. Simulation accuracy and channel evaluation begin and end with accurate S-parameter measurements.
A Closer Look at PAM4
PAM4 brings loss levels down to acceptable specifications but the trade-off is an incremental leap in signal complexity. PAM4 also introduces potential nonlinearities. Variations in the eye heights of the three eye openings is called “eye compression.” Similarly, “timing skew” occurs when the centers of the three eyes are misaligned. Figure 1 shows an eye diagram that suffers both eye compression and timing skew.
Of course, signal integrity engineers must still contend with jitter, noise, and crosstalk when measuring PAM4 signals. Very clean clocks, robust clock recovery circuits, and sensitive symbol decoders/voltage slicers are all necessary. In fact, three slicers are required to decode the four possible symbols. Better equalization schemes at both the transmitter and receiver to address inter-symbol interference (ISI) caused by the channel are necessary with PAM4 designs.
While equalization can compensate for the ISI caused by channel frequency and phase response, it causes trouble in the presence of crosstalk. The parameters that comprise ISI, crosstalk, and equalization present a multi-variable optimization problem. Signal integrity engineers can think of it as a multi-dimensional design space where they search for that point where BER is minimized. S-parameters provide the tools to predict the combined effects of ISI, equalization, and crosstalk.
The introduction of PAM4 signaling complicates design and test of high-speed components and systems at every step, from design to development to manufacture. The key measurements demand accurate S-parameters that can only be measured on VNAs. Since VNAs are high-quality precision instruments that can be major investments, it’s important to use the appropriate equipment for the product stage. For SERDES and interconnect design, a full featured VNA, such as the Anritsu VectorStar™, satisfies the performance requirements. For system design and development, a 4-port ShockLine™ Performance VNA is probably more than adequate, and for verification and manufacture, a ShockLine Economy VNA might be all that is necessary. Table 1 shows important features for PAM4 applications for two such VNA models.
VNA measurements are more accurate in both the time and frequency domains than TDT/TDR measurements for three reasons:
- Noise floor: The noise floor of any measurement is proportional to the bandwidth over which the quantity is measured. By converting a step response to an impulse response, TDT/TDR systems make measurements over 35 GHz. A VNA can make many narrowband measurements each over tens of hertz, making TDT/TDR measurement noise floors inherently higher.
- Dynamic range: The typical dynamic range of a TDT/TDR test set is around 40 dB. The dynamic range of the VNAs shown in Table 1 are typically over 100 dB—a million times that of TDT/TDR!
- Reciprocity: With their short rise-time voltage steps and intrinsic timebase uncertainty, TDT/TDR measurements suffer synchronization problems that can lead to inconsistent S-parameters and make it impossible to model PAM4 problems like timing skew.
While TDT/TDR can be used to measure crosstalk S-parameters, at least in principle, their limited dynamic range and high noise floors make it difficult for them to access the weak coupling between victims and aggressors with any accuracy. Precise NEXT and FEXT S-parameters are necessary to perform accurate IBIS-AMI models or COM measurements.
Since PAM4 signaling is used at high data rates in high loss environments, the dynamic range and accuracy of a VNA are necessary. Figure 2 shows a measurement of the VectorStar ME7838D dynamic range from 70 kHz to 145 GHz, as an example.
For a more thorough explanation on the challenges associated with PAM4 signals and how to measure them using a VNA, download a new white paper from Anritsu.