Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [802.3_100GCU] Recommended channels to use



Hi Vivek,

Yes, you interpreted the data correctly. The channel set includes both the NEXT and FEXT contribution from each aggressor. Including both crosstalk per aggressor is not a realistic case. The user will have to determine which to use based on their configuration as you have illustrated in your example.

Thanks,

Megha 

-----Original Message-----
From: Vivek Telang [mailto:vtelang@xxxxxxxxxxxx] 
Sent: Wednesday, June 13, 2012 5:17 PM
To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100GCU] Recommended channels to use

Charles,

Based on the TEC's pdf that came with the data, specifically page 5, my understanding is that the victim pins are C9D9, surrounded by 8 aggressors. The crosstalk s-parameters were measured from each of the aggressor ports to the victim in both directions (NEXT and FEXT), but in a real application, only one of the directions would be valid.

So for example, one configuration might be:
FEXT: D8D9_C8C9, C5C6_C8C9, C11C12_C8C9, B8B9_C8C9
NEXT: D5D6_C8C9, D11D12_C8C9, B5B6_C8C9, B11B12_C8C9

Can someone from TEC (Megha?) comment on this?

Vivek


-----Original Message-----
From: Charles Moore [mailto:charles.moore@xxxxxxxxxxxxx]
Sent: Wednesday, June 13, 2012 6:29 PM
To: Vivek Telang
Cc: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100GCU] Recommended channels to use

vivek,

I do not understand TEC's channel naming well enough to comment. Could it be that the FEXT aggressors and the NEXT aggressors come from pins with the same names but on different boards?

charles

|--------------------------------------------------------------------|
|       Charles Moore 
|       Avago Technologies
|       APD
|       charles.moore@xxxxxxxxxxxxx
|       (970) 288-4561
|--------------------------------------------------------------------|



Vivek Telang wrote:
> Charles,
>
> Thanks for volunteering and taking the time to do this. This is very useful!
> I am parsing the data as we speak, and will have some meaningful feedback soon.
>
> BTW, I noticed that for the TEC channels, you are using many of the same pins for NEXT as well as FEXT. I understand that this channel configuration is somewhat academic, but should we try to make it more realistic by making the NEXT and FEXT ports mutually exclusive?
>
> Regards
> Vivek
>
>
>
>
> -----Original Message-----
> From: Charles Moore [mailto:charles.moore@xxxxxxxxxxxxx]
> Sent: Tuesday, June 12, 2012 4:42 PM
> To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
> Subject: [802.3_100GCU] Recommended channels to use
>
> folks,
>
>     At the June 7 channel model consensus group discussion i agreed to 
> provide recommended channels to use for testing potential channel 
> evaluation methods.  Enclosed is a csv file with my recommendations 
> for NRZ channels intended to work without FEC. A recommended set for 
> PAM4 will be following at a later date.
>
>    In the file i list 10 channels,
>
>    2 which qikSN says are definitely bad, even with FEC,
>      one a long channel, the other short;
>
>    6 which qikSN says are near the limit without FEC, some
>      slightly above the limit, some slightly below.
>
>    2 which qikSN say are definitely good both moderately long.
>
> Two of the channels, which i call FCI_Long4, and FCI_short_7 have 
> questionable data at 50MHz, the lowest available frequency.  I 
> recommend that anyone trying to use them delete or correct the 50MHz data.
>
>                        charles
>
>