Re: [10GBT] Proposed PAM8 vs. PAM12 resolution process
Alon,
A friendly amendment to your third criterion (power). Since power will be a tricky thing to define, much less predict, I propose that we use implementation parameters, such as ADC ENOB, canceller lengths, equalizer lengths, and of course signaling rate as the means to evaluate it. Since PAM-8 and PAM-12 are close enough architecturally, we can assume that any implementation innovation that applies to one will also apply to the other.
Regards,
Vivek
-----Original Message-----
From: stds-802-3-10gbt@ieee.org [mailto:stds-802-3-10gbt@ieee.org]On
Behalf Of Meisler, Alon
Sent: Sunday, July 18, 2004 2:35 AM
To: STDS-802-3-10GBT@listserv.ieee.org
Subject: Re: [10GBT] Proposed PAM8 vs. PAM12 resolution process
Hiroshi,
I think you probably put your finger right at the essence of the
difference between the 2 (out of 3) analysis methods we can adopt - but
then you dismissed it almost immediately without analyzing it
thoroughly.
Like you, I do believe 1 gap between the PAM 8 and PAM 12 will be
implementation loss immunity. (Maybe I'm wrong - but if I am, I'm sure
someone will correct me :-) ). I don't think that since it's too
complicated, we just have to let it go.
I also believe that another gap between the PAM 8 and PAM 12 is External
noise immunity (say, immunity for ANEX)
I also believe that the 3rd gap between the 2 is power along the years
(technology development).
However - we should be able to reflect all those gaps in a quantitative
method and to see in one glance all the differences together. I'm sure
we'll be able as a team to take the right decision when seeing the
quantitative data for all these 3 items together.
So, basically, what I'm suggesting is not that one method will be
preferable upon the 2 others - but all the 3 will be done (impl. Losses
immunity, external noise immunity and power) and if there's more - then
the 4th or the 5th should also be done and presented.
Is it fine with you?
Does someone see more than these 3 gaps?
Alon
-----Original Message-----
From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG] On
Behalf Of hiroshi takatori
Sent: Sunday, July 18, 2004 8:16 AM
To: STDS-802-3-10GBT@listserv.ieee.org
Subject: Re: [10GBT] Proposed PAM8 vs. PAM12 resolution process
Sailesh,
Sailesh,
It only matters how robust the receiver is against "EXTERNAL NOISE".
Do not confuse how robust the receiver against "INTERNAL NOISE". When I
said "INTERNAL NOISE" it includes all IC implementation losses due to
any non-ideal devices such as ADC, DAC, jitter, analog circuit linearity
and others.
When we discuss PAM8 and 12 comparisons, it should be focused on which
solution performs better with ideal conditions, meaning no
implementation loss. This tells which solution is fundamentally better
in theory and this has to be done first.
You showed in your presentation PAM12 performs better in models 1, 3,
and 4 when BGN is -150dBm. This means that with very small
implementation losses, PAM12 performs better than PAM8 over the long
cable. I agree that and also I agree that PAM8 is better than or as good
as PAM12 for model 2, which is 55m case.
It is true that all IEEE folks want to know if it can be practically
implemented or not. When we argue the comparison with the implementation
losses, the argument will be very muddy because it changes many things
depending on which criteria we use, power and complexity with the
function of technology assumption in time. Therefore, KeyEye's approach
has been the comparison with the define BGN -140dBm/Hz as the
implementation loss.
In summary,
1. We all agree PAM12 is better in theory or with very small
implementation loss for 100m cable.
2. We all agree PAM8 and 12 will be performing about same for
55m cable.
3. If we need to argue more on the comparison with the
implementation loss, I propose to define BGN such as -140dBm/Hz as the
implementation loss and then compare PAM8 and PAM12.
Hiroshi Takatori
Keyeye Communications, Inc.
-----Original Message-----
From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG] On
Behalf Of sailesh rao
Sent: Saturday, July 17, 2004 7:01 PM
To: STDS-802-3-10GBT@listserv.ieee.org
Subject: Re: [10GBT] Proposed PAM8 vs. PAM12 resolution process
Hiroshi,
The residual noise power isn't just the background noise, but also
includes
the alien NEXT power, the residual echo/NEXT/FEXT power etc. This is the
input-referred residual noise power that each system is withstanding at
the
slicer.
Sailesh.
>From: hiroshi takatori <hiroshi.takatori@KEYEYE.NET>
>Reply-To: "IEEE P802.3an" <STDS-802-3-10GBT@listserv.ieee.org>
>To: STDS-802-3-10GBT@listserv.ieee.org
>Subject: Re: [10GBT] Proposed PAM8 vs. PAM12 resolution process
>Date: Sat, 17 Jul 2004 18:14:06 -0700
>
>Sailesh,
>I think all you are saying is that
> i) $B!! (JBoth systems, PAM8 and PAM12 are showing roughly
same
>noise margin with -150dBm/Hz BGN,
> ii) -150dBm/Hz BGN is defined as spectral density, and
> iii) PAM8 uses wider bandwidth than PAM12 by 17.5%,
>
>therefore, PAM8 is suffereing bigger BGN at input by 0.8dB (not 2dB as
you
>presented) than PAM12.
>
>
>THIS IS NOT CORRECT!
>
>The tolerable external noise is identical for both systems. That is
>-150dBm/Hz! You can not convert this into rms value unless bandwidth is
>defined. The bandwidth is determined by the receiver itself. PAM8 needs
>more bandwidth than that of PAM12 which means that PAM8 is more
sensitive
>to the higher frequency noise. What I am pointing out is that PAM12
rejects
>more noise than that can be rejected by PAM8.
>
>Plus, as we discussed back room in Portland hotel, PAM8 pollute more
than
>PAM12 does. I will show you more in detail soon.
>
>I will take a look at default values in the matlab code and do more
>details. Do you prefer to have 0.75+0.25/Z in your PAM8 and compare
with
>PAM12 without such filtering or just simple PAM8 and PAM12?
>
>Hiroshi Takatori
>KeyEye Communication
>
>
>
>-----Original Message-----
>From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG] On
>Behalf Of Sailesh Rao
>Sent: Saturday, July 17, 2004 3:30 PM
>To: STDS-802-3-10GBT@listserv.ieee.org
>Subject: Re: [10GBT] Proposed PAM8 vs. PAM12 resolution process
>
>Hiroshi,
>
>The default cancellation parameters are in the matlab source code on
the
>task force web site:
>
>http://grouper.ieee.org/groups/802/3/an/public/material/index.html
>
>As far as the transmit PSD is concerned, please feel free to send me
>what you think is optimum for the PAM12 system and I would be happy to
>use that. Please note that the matlab code currently uses a flat
>transmit PSD. However, I believe that the proposition before us, viz.,
>
>"When two PAM systems have roughly the same SNR margin
>(Margin(PAM8)=5.6dB, Margin(PAM12)=5.9dB), the PAM system with the
>smaller number of levels (PAM8) will be fundamentally more robust
>towards external noise,"
>
>holds regardless of the transmit PSD. For instance, whatever transmit
>PSD is proposed for the PAM12 system, we can always scale that in
>amplitude and frequency and reuse it for the PAM8 system.
>
>Regards,
>Sailesh Rao
>srao@phyten.com
>
> >From: hiroshi takatori <hiroshi.takatori@KEYEYE.NET>
> >Reply-To: "IEEE P802.3an" <STDS-802-3-10GBT@listserv.ieee.org>
> >To: STDS-802-3-10GBT@listserv.ieee.org
> >Subject: Re: [10GBT] Proposed PAM8 vs. PAM12 resolution process
> >Date: Fri, 16 Jul 2004 23:09:07 -0700
> >
> >Sailesh,
> >
> >Please, define default cancellation parameters and necessary
parameters
> >to create transmit PSDs for both PAM8 and 12.
> >
> >Hiroshi
> >
> >KeyEye
> >
> >________________________________
> >
> >From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG] On
> >Behalf Of Sailesh Rao
> >Sent: Friday, July 16, 2004 9:01 PM
> >To: STDS-802-3-10GBT@listserv.ieee.org
> >Subject: [10GBT] Proposed PAM8 vs. PAM12 resolution process
> >
> >
> >
> >All,
> >
> >
> >
> >I would like to propose the following process for resolving the
> >robustness of PAM8 vs. PAM12 towards external noise.
> >
> >
> >
> >1. Compute the Optimum DFE SNR Margin for PAM8 and PAM12 using
> >solarsep_varlen7a.m for Models 1 and 3 using default cancellation
> >parameters and -150dBm/Hz background noise.
> >
> >
> >
> >2. Compute the input-referred RMS noise power at the slicer by
> >integrating the residual noise in the Optimum DFE solution. I
volunteer
> >to add this code to solarsep_varlen7a.m unless someone else wants to
do
> >so.
> >
> >
> >
> >3. Compute the input-referred external noise power that can be
>tolerated
> >for a BER of 1E-12 for both systems using the results from (1) and
(2)
> >above. I volunteer to add this code to solarsep_varlen7a.m unless
> >someone else wants to do so.
> >
> >
> >
> >Regards,
> >
> >Sailesh Rao.
> >
>
>_________________________________________________________________
>Overwhelmed by debt? Find out how to 'Dig Yourself Out of Debt' from
MSN
>
>Money. http://special.msn.com/money/0407debt.armx
_________________________________________________________________
Planning a family vacation? Check out the MSN Family Travel guide!
http://dollar.msn.com