| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Anand, to your questions: #1 – when c(0) is lower than 1 and is fixed, changing pre/post (c(-1) or c(+1)) may have an effect on the pulse peak. Since these are negative or zero, the pulse peak can
only be reduced further, but the effect depends on the channel’s pulse response. For simplicity I suggest not to account for this effect, at least at this time and just look at the reduction of the pulse peak. This will be similar to the case of just reducing
the swing (keeping pre/post zero or close to zero) and simple to analyze. #2 – the calculation of COM in past analysis has showed SNDR is often a bottleneck in borderline high-loss channels. It is true that in low-loss channels there is more
margin, and c(0) can be reduced (effectively degrading the SNDR) without failing COM; but for high loss channels c(0) still needs to be reduced for equalization purposes – we can’t assume c(0)=1. The correction of the sigma_tx term assumes this reduction will
be to 0.6, so SNDR is degraded by ~4.4 dB; if it’s 32.5 with c(0)=1 it will become 28.1 dB. This will have a big effect on COM. #3 – I did not propose to use c(0) in the SNDR measurement (although it is possible, if the measurement is done with c(0)!=0, as suggested in the meeting; c(0) is measurable).
I’m proposing to use it in the channel specification (COM) where it is available as the result of the equalizer optimization. If measurement of SNDR is done with c(0) smaller than 1, the measured SNDR should be adjusted by -20*log10(c(0)) (it will be increased). A related point made in the meeting was that reducing the swing will improve transmitter linearity and reduce distortion. This may be true , but again, reducing the swing
is not likely in high-loss channels – instead, the reduction of c(0) is to allow higher absolute values for other coefficients. At the transmitter, the maximum values will still be reached, so we should not assume better linearity. </Adee> From: Adee Ran (aran) <0000147b29386f6c-dmarc-request@xxxxxxxxxxxxxxxxx>
I am getting feedback on several threads in parallel, so I thought it would be best to move this discussion to the reflector. </Adee> From: Anand Ravindran Nair Katchani <anandrk@xxxxxxxxxxx>
No problem and Thank you Beth for starting the thread.
Hi Adee, First of all Thank you for bringing this topic and efforts for real good data.
As discussed in the meeting I would like to understand few things regarding SNDR proposal.
Regards, Anand RK From: Beth Kochuparambil (edonnay) <edonnay@xxxxxxxxx>
EXTERNAL MAIL Hey Anand – Thanks for your questions, sorry we had to push them offline, but we need to move on. Hopefully Adee can answer those questions here instead. ~Beth 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 |