RE: [IBIS-Users] RE: Role of C_comp

From: Ling Li Ong <llong@altera.com>
Date: Thu Sep 05 2013 - 18:32:08 PDT
Hi Todd,

Thanks for your email.  I think I've gotten the answer from your email.  Good IBIS model is really important.

As for my experiment after trying the smaller time steps as suggested by Arpad, the waveform has become smoother but yet, they are still not identical.

[cid:image001.jpg@01CEAAE3.F489E920]
Regards,
Ling Li

From: owner-ibis-users@eda.org [mailto:owner-ibis-users@eda.org] On Behalf Of Todd Westerhoff
Sent: Thursday, September 05, 2013 8:58 PM
To: ibis-users@eda.org
Subject: RE: [IBIS-Users] RE: Role of C_comp

Ling Li,

The V-T curves are generated with c_comp present, so, in theory, the output waveforms should remain the same even if c_comp is changed.

Reality is not always the same as theory, however.  In some cases, you may see differences that can be overcome by changing the simulator setup, as Arpad suggests.

But - and this is a big but - there are, and will always be, limits. The V-T waveforms tell the simulator what the output behavior should be into a specified loading condition, with c_comp included.  The output driver has finite resistance, and a capacitance of c_comp, which forms a circuit with an RC time constant. If you crank c_comp way up, say to 50pf, you will describe a condition that is physically impossible.  The RC time constant of the output driver and c_comp will be large, while the switching time described by the V-T curves will be small.  There is no way for a simulator to resolve the conflict - one number or the other will win out.

Here's my point - an IBIS [model] contains matched sets of data that describe a buffer in multiple ways, with overlap and redundancy. If all of the data tells the same story [a well-constructed IBIS model], the correct simulation result is well-defined. If, however, different parts of the model tell different stories, you get what you get. This is usually the result of a model having been modified or created by hand ... when automated tools are used to generate models, all the data is consistent. When models are edited manually, the result tends to reflect the user's understanding of the spec, with that can produce different results in different tools.

... a long way of saying, people should edit IBIS files with caution backed by understanding of the spec. Just because a model works and produces a good result in one tool doesn't mean it will work in another.

Todd.


Todd Westerhoff
VP, Software Products
Signal Integrity Software Inc. * www.sisoft.com<http://www.sisoft.com>
6 Clock Tower Place * Suite 250 * Maynard, MA 01754
(978) 461-0449 x24  *  twesterh@sisoft.com<mailto:twesterh@sisoft.com>

"I want to live like that"
                                             -Sidewalk Prophets

From: owner-ibis-users@eda.org<mailto:owner-ibis-users@eda.org> [mailto:owner-ibis-users@eda.org] On Behalf Of Ling Li Ong
Sent: Thursday, September 05, 2013 1:09 AM
To: Muranyi, Arpad; ibis-users@eda.org<mailto:ibis-users@eda.org>
Cc: Siong Hee Lim
Subject: [IBIS-Users] RE: Role of C_comp

Hi Muranyi,

Do you mean for an output buffer, if purely changing the c_comp value between 0.1pF and 50pF and maintain other parameters as identical, it should not generate any different waveforms?

I tried out the a simple output buffer simulation ran on HSPICE 2011.09-SP1.  No transmission line or termination impedance but mainly the ibis model.   C_comp swept at 3.5pF (purple) and 5pf (green).
I observed the waveforms are not overlapped.  Is this the correct result from HSPICE?  If it is not, what will be the expected result?


[cid:image004.jpg@01CEAAE3.6C42FA70]

Regards,
Ling Li


From: Muranyi, Arpad [mailto:Arpad_Muranyi@mentor.com]
Sent: Thursday, September 05, 2013 11:48 AM
To: Ling Li Ong; ibis-users@eda.org<mailto:ibis-users@eda.org>
Cc: Siong Hee Lim
Subject: RE: Role of C_comp

Ling Li,

Regarding: "it has no difference between the simulation on C_comp 0.1pF and 50pF".
That's correct (for the case when the buffer model is driving), and
that's the way HyperLynx 9.0 should also behave.

Thanks,

Arpad
======================================================================

From: Ling Li Ong [mailto:llong@altera.com]
Sent: Wednesday, September 04, 2013 2:19 AM
To: Muranyi, Arpad; ibis-users@eda.org<mailto:ibis-users@eda.org>
Cc: Siong Hee Lim
Subject: RE: Role of C_comp

Hi Arpad,

Thanks for your fast response.  I was referring to the mentor's website which commented with using the version 8.2.1, it has no difference between the simulation on C_comp 0.1pF and 50pF.
This version may be outdated compare to latest as I noticed the posting date was on last year.  However, I assume it should perform the same simulation result after this version.
http://communities.mentor.com/mgcx/thread/6935

May I know is this comment align to your understanding?

Thanks.

Regards,
Ling Li

From: Muranyi, Arpad [mailto:Arpad_Muranyi@mentor.com]
Sent: Wednesday, September 04, 2013 2:46 PM
To: Ling Li Ong; ibis-users@eda.org<mailto:ibis-users@eda.org>
Cc: Siong Hee Lim
Subject: RE: Role of C_comp

Ling Li,

It seems that your conclusion from reading Todd Westerhoff's email
in the link you provided is not quite correct.  Todd seems to say
(correctly) that the output waveform of a driver will not change
when C_comp is changed in the IBIS [Model].  You should only see
differences in the waveforms when a reflection hits the driver
and that reflection bounces back off of the driver.  This secondary
reflection from the driver should depend on the C_comp value.

Similarly, if an I/O buffer model is in the receive mode (i.e. input
state), you should see different waveforms when you change the C_comp
value.

As far as I can tell from my days when I worked at Intel and used
HSPICE in my work, HSPICE does what I described above correctly.

Now that I work for Mentor, I can also tell you that the results
of HyperLynx are correct the same way.  I am actually surprised to
hear that you see different results.  Are you sure your test case
is designed well to show and observe these effects correctly?

By the way, the latest HyperLynx version is now 9.0 (not 8.2.1).

I hope this helps answering your question.

Thanks,

Arpad
=======================================================================

From: owner-ibis-users@eda.org<mailto:owner-ibis-users@eda.org> [mailto:owner-ibis-users@eda.org] On Behalf Of Ling Li Ong
Sent: Wednesday, September 04, 2013 12:47 AM
To: ibis-users@eda.org<mailto:ibis-users@eda.org>
Cc: Siong Hee Lim
Subject: [IBIS-Users] Role of C_comp

Hi,

I saw some discussions on c_comp issue at user archive at link:
http://www.eda.org/ibis/users_archive/0487.html

I am confused by the c_comp whether it is double-counted and the different result obtained when using different tools.
According to the comment in the link above, varying the c_comp may see the changes in rise/fall curve.  This is definitely true when using the HSPICE simulator.  But, if you using the latest Hyperlynx  version 8.2.1, changing the c_comp will not reflect the changes in rise/fall curve.

Can you please advise which tools are showing the correct response on c_comp when sweeping at different value?
Thanks.

Regards,
Ling Li






________________________________
Confidentiality Notice.
This message may contain information that is confidential or otherwise protected from disclosure. If you are not the intended recipient, you are hereby notified that any use, disclosure, dissemination, distribution, or copying of this message, or any attachments, is strictly prohibited. If you have received this message in error, please advise the sender by reply e-mail, and delete the message and any attachments. Thank you.

--
This message has been scanned for viruses and
dangerous content by MailScanner<http://www.mailscanner.info/>, and is
believed to be clean.

________________________________
Confidentiality Notice.
This message may contain information that is confidential or otherwise protected from disclosure. If you are not the intended recipient, you are hereby notified that any use, disclosure, dissemination, distribution, or copying of this message, or any attachments, is strictly prohibited. If you have received this message in error, please advise the sender by reply e-mail, and delete the message and any attachments. Thank you.

________________________________
Confidentiality Notice.
This message may contain information that is confidential or otherwise protected from disclosure. If you are not the intended recipient, you are hereby notified that any use, disclosure, dissemination, distribution, or copying of this message, or any attachments, is strictly prohibited. If you have received this message in error, please advise the sender by reply e-mail, and delete the message and any attachments. Thank you.

--
This message has been scanned for viruses and
dangerous content by MailScanner<http://www.mailscanner.info/>, and is
believed to be clean.

--
This message has been scanned for viruses and
dangerous content by MailScanner<http://www.mailscanner.info/>, and is
believed to be clean.

________________________________
Confidentiality Notice.
This message may contain information that is confidential or otherwise protected from disclosure. If you are not the intended recipient, you are hereby notified that any use, disclosure, dissemination, distribution, or copying of this message, or any attachments, is strictly prohibited. If you have received this message in error, please advise the sender by reply e-mail, and delete the message and any attachments. Thank you.

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.



--------------------------------------------------------------------
|For help or to subscribe/unsubscribe, e-mail mikelabonte@eda-stds.org
|or ibis-request@eda-stds.org
|
|IBIS reflector archives exist under:
|
|  http://www.eda-stds.org/ibis/email_archive/ Recent
|  http://www.eda-stds.org/ibis/users_archive/ Recent
|  http://www.eda-stds.org/ibis/email/         E-mail since 1993
image004.jpg
image001.jpg
Received on Thu Sep 5 18:32:56 2013

This archive was generated by hypermail 2.1.8 : Thu Sep 05 2013 - 18:33:41 PDT