| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Hi Geoff, The measurement of EOJ as defined in 120D.2.1.8.2 is specifically intended to capture the timing difference between even and odd transitions. This impairment is characteristic
of the driving clock, and is therefore independent of the channel. Having a channel just makes it hard to measure. Any dependence of the EOJ on the channel would be a measurement artifact. I’m glad to know that this concern is removed. @Rich: the pulse fitting (which is the first step in SNDR measurement) should be performed with averaged signal, which I thought should cancel the jitter effect. If the
SJ frequency is coherent with the PRBS frequency it would not cancel out, but I doubt that. I would suspect insufficient averaging as a likely cause, since with SJ one would need to average a lot more than without it. Looking at sigma_e of the SNDR, with and without
SJ, should give a good hint. I still don’t see how this affects VEC/EH in real life – since these measurements do not use the fitted pulse? </Adee> From: Richard Mellitz <richard.mellitz@xxxxxxxxx>
I believe is an effect of measuring only timing differences for 12 specific edges (table 120D-4) in the PRBS13Q pattern. See 120d.2.1.8.2. I do not think this is what may be called pulse shrinkage or duty cycle distortion in other specifications. … Rich From: Geoff Zhang <geoffz@xxxxxxxxxx>
Hi Rich, Where is it stated that EoJ does not depend on channel (loss)? Thanks, Geoff From: Richard Mellitz <richard.mellitz@xxxxxxxxx>
CAUTION: This message has originated from an External Source. Please use
proper judgment and caution when opening attachments, clicking links, or responding to this email. Follow up on data presented on EoJ John Calvin performed a targeted experiment and proved EoJ does not depend on channel (loss). John Calvin also raised concern over Sj in VEC/EH is approximately the same (for the same channel) when computed from the COM Matlab code using s-parameters and the channel fitted pulse response (from the pattern generator/BERT) using a special version of COM
which imports a pulse response instead of s-parameters. The BERT VEC/EH software produces approximately the same results too, So what is going on with Sj? Why does it egregiously close the eye? If Sj is at 40 MHz and Fb is 50GHz. The ratio of Sj frequency to Fb is around 3 orders of magnitude or approximately 1000 UI. The pulse fit is done over Nv=200 UI, The data pattern is thousands of symbols long. I suspect SNDR going be smaller (larger sigma_e) if Sj is turned on. Q: Could this account for the eye reduction when Sj is applied? Thank you all, Rich Mellitz Sent from my iPad To unsubscribe from the STDS-802-3-100GEL list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-100GEL&A=1
To unsubscribe from the STDS-802-3-100GEL list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-100GEL&A=1 To unsubscribe from the STDS-802-3-100GEL list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-100GEL&A=1 |