| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Tingting Yes, you are correct in general the difference in absolute SNR between PAM-3 and PAM-4 should be about the same as the difference is the signal power. And as PAM-4 should operate at a lower symbol rate it has some advantage here which should
also translate into improved SNR. At lower noise levels in my sims I see between 0.5 and 1.5dB between PAM-3 at 75 MSym/s and PAM-4 at 62.5 MSym/s.
However, the most important difference between PAM-3 and PAM-4 is that the PAM-4 levels are closer together and you require a higher SNR, by 3.5 dB to get the same bit error rate. The 3.5 dB, is compensated by the 0.7 dB difference in lower
PAM-4 signal power so this translates to about 2.8 dB. So we can’t compare absolute SNR levels when comparing PAM-3 and PAM-4. The next level of difference is probably down to implementation details. In my presentations I have outlined the basic parameters of the data path of the time domain simulation. I would need to see some more details your data path parameters
to comment. And you have only given a single cable length and noise level, which makes comparison difficult. For PAM-4 at 62.5 MSym/s for 500m at noise levels of -113 dBm/Hz over 100 MHz ( 7 mV rms) I am seeing error propagation in my time domain simulations,
which is degrading the SNR. But this is a real issue at high noise levels, once the SNR get a few dB below the 10-10 BER rate we will start to see this effect, which is a few dB below 23 dB for PAM-4.
Thanks Brian To unsubscribe from the STDS-802-3-SPEP2P list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-SPEP2P&A=1 |