Joe,
Having spent the better part looking at
that channel in relation to the previous channels as well as the newer channels
that Charles provided, and this channel has some unique return loss
characteristics. And when I say that, you need to realize that this
channel is not symmetrical and SDD11 is significantly different than
SDD22. This channel is the worst in terms of return loss of anything I have
really seen, especially interesting is the low frequency return loss. Also,
I noticed that the behavior of the reflections on this channel is higher in
amplitude based on the SDD11 and if you look at the pulse response of the
forward channel response is different in overall behavior.
So, I would say this channel is proof that
we do need to add the return loss specification and that our theory that we
were capturing all of this in the forward response mask set was probably not
100% correct.
John
-----Original Message-----
From: Joe M Abler
[mailto:abler@US.IBM.COM]
Sent: Tuesday, February 28, 2006
7:05 PM
To:
STDS-802-3-BLADE@listserv.ieee.org
Subject: Re: [BP] channels to test
our spec.
Charles,
I
ran these channels with the basic sig adhoc setup but with TJ=0.28UI and
DCD=3.125%. I used a PRBS15 data pattern (which is what we defined for
the sig adhoc setup) and simulated for 1M bit times. These were
through-channel sims only, no crosstalk applied. The performance of all
channels were considerably better than the thru_worstcase channel. The
lowest performing channel was m82ripple90 with 18% opening at E-12.
I
was surprised at the difference compared to thru_worstcase, which had an eye
which was just closed at E-12. One difference in the previous sim is that
I used PRBS23 (since that's what is defined for EIT) and simulated for 10M
bits. I reran the channel using PRBS15 & 1M bits, which resulted in
about an 8% eye at E-12. So there is a fair impact from PRBS23, but this
channel is still significantly worse than the others.
Bottom
line is that I'm not finding anything with these new channels that would
provide insight on how to redraw the line.
Thanks, Joe
Joe Abler
abler@us.ibm.com
IBM Systems & Technology Group 919-254-0573
High Speed Serial Link Solutions
919-254-9616 (fax)
3039 Cornwallis Road
Research Triangle Park, NC 27709
|
Charles Moore
<charles.moore@AVAGOTECH.COM>
02/22/2006 07:41 PM
Please
respond to Charles Moore
|
To: STDS-802-3-BLADE@listserv.ieee.org
cc:
Subject: [BP] channels to
test our spec.
|
joe,
et al,
Here are
some channels to look at the ripple vs attenuation
problem.
charles
--
|--------------------------------------------------------------------|
| Charles Moore
| Avago Technologies
| ISD
| charles.moore@avagotech.com
| (970) 288-4561
|--------------------------------------------------------------------|
####
constructedChannels1.zip has been removed from this note on February 28, 2006
by Joe M Abler