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

Re: [10GBT] Summary of issues with PAM12


The decrease in clock speed is

(952.381/825 - 1)*100% = 15%

However, it is not correct that this clock rate reduction can be done "with
no significant margin" penalty.

The emissions penalty for this reduction is between 0.62dB and 1.05dB, which
is real margin that we are losing.

The susceptibility penalty for this reduction is between 2.0dB and 3.9dB
over all cable lengths, which is also real margin that we are losing.

Finally, even if you are only interested in the SNR margin, the SNR
requirement for PAM12 is 23.8dB for a 1E-12 BER, vs. 19.9dB for PAM8. This
requires the PAM12 receiver to find an extra 3.9dB somehow, with more
precise analog front ends perhaps.

Sailesh Rao.

>From: Sanjay Kasturia <SKasturia@TERANETICS.COM>
>Reply-To: "IEEE P802.3an" <>
>Subject: Re: [10GBT] Summary of issues with PAM12
>Date: Fri, 20 Aug 2004 09:09:20 -0700
>I will let others speak for themselves but the main reason I prefer
>12PAM over 8PAM is the ~10% to 20% lower symbol rate which will reduce
>the speed requirements for the AFE and the clock speed requirements for
>the digital signal processing.
>I think the lowering of the symbol rate can be done with no significant
>margin penalty.
>cell (650) 704-7686
>office (408) 653-2235
>-----Original Message-----
>From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG] On
>Behalf Of sailesh rao
>Sent: Friday, August 20, 2004 8:33 AM
>Subject: Re: [10GBT] Summary of issues with PAM12
>With regard to issues 1 and 2, I don't believe that there are any
>disputes about my calculations. I've already addressed these so-called
>disputes in my summary e-mail and pointed out why my analysis stands.
>With regard to issue 3, firstly, the main rationale advanced by the
>PAM12 proponents for their 3bits/baud scheme is that it has 0.7dB higher
>Salz SNR margin over the 2.5bits/baud scheme used in PAM8, over a 100m
>However, once you put a hole in the constellation and throw away at
>least 1.1dB, would you now concede that this rationale is technically
>invalid, and that the Salz SNR margin advantage also swings to PAM8 over
>PAM12 at ALL cable lengths, from 0m to 100m?
>Secondly, the hole in the constellation makes the PAM12 transmit PSD a
>function of the THP coefficients. Therefore, every time a system vendor
>finds that the system violates emissions limits, we can expect him/her
>to demand to examine the THP coefficients of the PHY, if this task force
>standardizes PAM12.
>Thirdly, the  hole in the constellation makes the receiver SNR margin a
>function of the THP coefficients of the alien neighbors.
>With regard to issue 5, please check the details on the PAM8 proposal.
>You will see that there are NO fixed patterns used anywhere to delineate
>data frames or anything. Therefore, it is not "necessary" to use fixed
>patterns for this purpose. These fixed patterns waste the precious
>Shannon capacity which the cabling community is working hard to provide
>us, though not on the same scale as the other issues with the PAM12
>proposal, and which is why I listed this issue last.
>Sailesh Rao.
>Is your PC infected? Get a FREE online computer virus scan from
>McAfee(r) Security.

On the road to retirement? Check out MSN Life Events for advice on how to
get there!