What is the channel “flight time”?
For a straight trace
For a trace with a bend
For a trace with a serpentine
For a measurement or EM model with a leakage path through air?
This is a trick question.
Scott McMorrow, CTO Signal Integrity Group
Samtec
Office 401-284-1827 | +1-800-726-8329
www.samtec.com<http://www.samtec.com/>
Scott McMorrow,
Office: 401-284-1827
www.samtec.com
This email and any appended documents are only for the intended person/entity
and may contain information of Samtec, Inc., that is PRIVILEGED, PROPRIETARY,
CONFIDENTIAL, AND/OR PROTECTED BY LAW. If you are not the intended recipient
you are hereby notified that any dissemination, disclosure, use or copying of
this email or its contents is prohibited. If you this message in error, please
notify Samtec immediately and delete the email, attachments and all copies. The
intended recipient should not disclose the content to third parties or
reproduce the content without Samtec's written consent.
From: michael.huang@xxxxxxxxxx <michael.huang@xxxxxxxxxx>
Sent: Tuesday, February 25, 2020 1:43 PM
To: Scott McMorrow <Scott@xxxxxxxxxxxxx>; ibis-macro@xxxxxxxxxxxxx
Subject: RE: [EXTERNAL][ibis-macro] Re: Reminder to a strawmen vote next
Tuesday (Feb 25)
Hi Scott,
“When does the rising edge start” is exactly my question… a tolerance setting
is usually needed plus some assumptions.
Given an impulse response h(t), say I convolve it with a bit whose PWL points
are (0, 0), (t1, 0), (t1+dt, 1), (t2, 1), (t2+dt, 0) (100, 0)
Please correct me if I am wrong… I will assume the convolved pulse response
should be flat at least before t1+dt+(channel flight time)
Thanks and regards,
Wei-hsing
From: Scott McMorrow <Scott@xxxxxxxxxxxxx<mailto:Scott@xxxxxxxxxxxxx>>
Sent: Tuesday, February 25, 2020 10:25 AM
To: michael.huang@xxxxxxxxxx<mailto:michael.huang@xxxxxxxxxx>;
ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: RE: [EXTERNAL][ibis-macro] Re: Reminder to a strawmen vote next
Tuesday (Feb 25)
Define non-causal.
When does the rising edge start?
Scott McMorrow, CTO Signal Integrity Group
Samtec
Office 401-284-1827 | +1-800-726-8329
www.samtec.com<http://www.samtec.com/>
Scott McMorrow,
Office: 401-284-1827
www.samtec.com<http://www.samtec.com>
This email and any appended documents are only for the intended person/entity
and may contain information of Samtec, Inc., that is PRIVILEGED, PROPRIETARY,
CONFIDENTIAL, AND/OR PROTECTED BY LAW. If you are not the intended recipient
you are hereby notified that any dissemination, disclosure, use or copying of
this email or its contents is prohibited. If you this message in error, please
notify Samtec immediately and delete the email, attachments and all copies. The
intended recipient should not disclose the content to third parties or
reproduce the content without Samtec's written consent.
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of michael.huang@xxxxxxxxxx<mailto:michael.huang@xxxxxxxxxx>
Sent: Tuesday, February 25, 2020 1:21 PM
To: ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: [EXTERNAL][ibis-macro] Re: Reminder to a strawmen vote next Tuesday
(Feb 25)
Hi,
Ideally, the convolved pulse response is flat right before rising edge like
what Walter has drawn. In Hansel’s plot, it has a dip which may or may not
exceed some tolerance. If channel impulse response is not clean (i.e.
non-causal etc), then data prior to the first rising edge may be noisy. In that
case, a tolerance setting is needed in order to determine the starting point
pointed by the red below(This is plot from COM taskforce, COM has a setting for
that)
[cid:image001.png@01D5EBE2.9C607100]
If model returns the sampling points, does it mean only the model can
control this “tolerance” (i.e. maybe hard coded within the model)? Can EDA tool
instruct model what the tolerance should be? I am thinking EDA tool may or may
not preprocess impulse response so the tolerance may be different.
Thanks.
Wei-hsing
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of Walter Katz
Sent: Thursday, February 20, 2020 6:16 AM
To: 'Hansel Dsilva'
<hanseldsilva@xxxxxxxxxxxx<mailto:hanseldsilva@xxxxxxxxxxxx>>;
Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>;
ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: [ibis-macro] Re: Reminder to a strawmen vote next Tuesday (Feb 25)
Hansel,
You captured the points correctly. Thanks for the great picture. The black
circles are 1 UI apart, and one UI from the “Actual Cursor Point” returned by
Statistical_Sample_Time. I also think that this example can be used to support
Discussion 1 for the following reasons.
1. Discussion 2 requires a list of optional methods to determine the sample
point (and therefore the series of 1 UI spaced sample points which determine
the eye height PDF at the sample point.
2. Generating this list of methods in the committee, and documenting each
one will be challenging, and I will leave it at that.
* To paraphrase something a back surgeon told me: “The only thing that
two signal integrity engineers can agree to is that the third one is all wrong.”
1. This list will not include the “Modified Mueller-Muller” phase detector
implemented in this model. The “Modified” is likely IP that the developer of
this Rx will want to protect.
Arpad and I are clear supporters of Discussion 1. I think we can assure that
the straw vote will overwhelmingly approve Discussion 1 by stating up front
that if Discussion 2 passes, those voting for Discussion 2 will be responsible
for generating the list of methods the EDA tool must support to calculate the
sample location in the pulse response and document each one thoroughly.
[Smiling face with no fill]
Walter
Walter Katz
wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx>
Office 978.461-0449 x 133
Mobile 720.417-3762
[cid:image003.jpg@01D5EBE2.9C607100]
From: Hansel Dsilva
<hanseldsilva@xxxxxxxxxxxx<mailto:hanseldsilva@xxxxxxxxxxxx>>
Sent: Thursday, February 20, 2020 1:57 AM
To: wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx>;
Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>;
ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: RE: Reminder to a strawmen vote next Tuesday (Feb 25)
Hi Walter, Arpad,
Resending this email in correcting for typing mistakes (green).
I recognized the need for a better illustration for this as part of this
discussion given the EDA tools will need to work on a post Rx waveform (i.e.
TXLE, CTLE, DFE applied).
I see two discussions part of this email thread. Did I capture your points
correctly? Feel free to make edits through the attached .pptx if I am going in
the wrong direction.
A thank you to you two (Arpad and Walter) for this valuable discussion which
hopefully will lead to a more robust modelling environment through IBIS.
[cid:image004.png@01D5EBE2.9C607100]
Thanks,
Hansel Dsilva
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of [EXTERNAL] Hansel Dsilva
Sent: 20 February 2020 12:06 PM
To: wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx>;
Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>;
ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: [ibis-macro] Re: Reminder to a strawmen vote next Tuesday (Feb 25)
Hi Walter, Arpad,
I recognized the need for a better illustration for this discussion given the
need to work on a post Rx waveform (i.e. TXLE, CTLE, DFE applied).
I see two discussion part of this email thread. Did I capture your points
correctly? Feel free to make edits through the attached .pptx if I am going in
the wrong direction.
A thank you to you two for this valuable discussion which hopefully will lead
to a robust modelling environment through IBIS.
[cid:image004.png@01D5EBE2.9C607100]
Thanks,
Hansel Dsilva
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of Walter Katz
Sent: 20 February 2020 11:32 AM
To: wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx>;
Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>;
ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: [ibis-macro] Re: Reminder to a strawmen vote next Tuesday (Feb 25)
Arpad,
It is a poor graphic, that might not translate to other e-mail viewers. Here is
a screen shot of what I intended. The Blue dot is at the time that the model is
telling the EDA tool to use to sample the waveform. In this example the Blue
dot is located at the time of the largest pulse width height to maximize eye
height margin, while the Hula Hoop method might pick a time when the eye width
margin is maximized.
[cid:image005.png@01D5EBE2.9C607100]
Walter
Walter Katz
wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx>
Office 978.461-0449 x 133
Mobile 720.417-3762
[cid:image003.jpg@01D5EBE2.9C607100]
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of Walter Katz
Sent: Thursday, February 20, 2020 12:30 AM
To: Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>;
ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: [ibis-macro] Re: Reminder to a strawmen vote next Tuesday (Feb 25)
Arpad,
The blue dot is at the time chosen by the model to sample the waveform. It is
at he time returned by the model in Statistical_Sample_Time.
Walter
Walter Katz
wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx>
Office 978.461-0449 x 133
Mobile 720.417-3762
[cid:image003.jpg@01D5EBE2.9C607100]
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of Muranyi, Arpad
Sent: Wednesday, February 19, 2020 9:21 PM
To: ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: [ibis-macro] Re: Reminder to a strawmen vote next Tuesday (Feb 25)
Walter,
Thanks for your comments. I am trying to figure out what those blue
dots with the “x” inside them represent. Could you please explain it?
Thanks,
Arpad
======================================================
From: Walter Katz [mailto:wkatz@xxxxxxxxxx]
Sent: Wednesday, February 19, 2020 6:16 PM
To: Muranyi, Arpad <Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>>;
ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: RE: [ibis-macro] Reminder to a strawmen vote next Tuesday (Feb 25)
All,
One well defined way to determine the location of the clock in the statistical
eye is to first convert the impulse response to a pulse response by convolving
it with a unit pulse response (one for 1 UI, followed by all zeros), and then
dropping a 1UI “Hula Hoop” over the pulse response. This is to find the time in
the pulse response that the width of the pulse response is 1 UI.
“One option is to introduce a new parameter for the Rx AMI executable model to
return a value to the EDA tool and tell it where to sample the waveform.”
The value returned would be an alternative to the center of the Hula Hoop
described above.
[cid:image006.png@01D5EBE2.9C607100]This begs for a carefully draw picture,
here is the best that I can do. The following shows the Hula Hoop algorithm.
The curved line is the pulse response generate by convolving a 1 UI unit pulse
response with the impulse response. The horizontal line (with arrows) is 1 UI
wide. The center of this line is the location of the clock using the Hula Hoop
algorithm. The new Reserved Parameter (Statistical_Sample_Time (Usage Out)(Type
Float)(Value <NA>)(Description “Pulse Response Sample Time”)) might contain the
location indicated by the
[cid:image007.png@01D5EBE2.9C607100]
Walter
Walter Katz
wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx>
Office 978.461-0449 x 133
Mobile 720.417-3762
[cid:image003.jpg@01D5EBE2.9C607100]
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of Muranyi, Arpad
Sent: Tuesday, February 18, 2020 4:12 PM
To: ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: [ibis-macro] Reminder to a strawmen vote next Tuesday (Feb 25)
Hello Everyone,
I was reminded in today’s IBIS-ATM teleconference that we were planning to
do a strawmen vote on the subject of “Gap in IBIS for sampling with
statistical mode AMI models”. There are (at least) two possible solutions
to this “gap” and we would like to find out how people interested in this topic
would go about solving the problem.
One option is to introduce a new parameter for the Rx AMI executable model
to return a value to the EDA tool and tell it where to sample the waveform.
The other option is to write some rules in the IBIS-AMI specification that
describes to the EDA tool how to find the proper sampling time(s) along the
waveform.
Not to bias anyone with the vote, but it seems that it would make more sense
to have the Rx model return this information for the EDA tool, since it usually
contains a DFE and/or CDR through which the UI boundaries and/or sampling
times are already known. If the spec would describe rules for the EDA tool to
find that information, a discrepancy might arise between the way the EDA
tool and the Rx model finds the sampling time, and the simulation results might
become garbage…
Questions, comments are welcome, but at least please give this some thought
for next week, so we could take a strawmen vote… If you can’t attend next
week, you may send me your vote in an email too.
Thanks,
Arpad
=============================================================
delete the email, attachments and all copies. The intended recipient should
not disclose the content to third parties or reproduce the content without
Samtec's written consent.