From owner-ibis-users@eda.org Fri Oct 19 16:21:14 2001 Received: by server.eda.org (8.12.0.Beta7/8.12.0.Beta7) id f9JNJo8k020866; Fri, 19 Oct 2001 16:19:50 -0700 (PDT) From: "Raj Raghuram" To: "Chao Su" Cc: Subject: RE: High frequency performance of IBIS model Date: Fri, 19 Oct 2001 16:19:33 -0700 Message-ID: MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0055_01C158B9.D6DC4B40" X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0) In-Reply-To: <710197BD5AF9D4119E4400508BCFA13601A7B24F@zcard04u.ca.nortel.com> X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Importance: Normal Sender: owner-ibis-users@eda.org Precedence: bulk This is a multi-part message in MIME format. ------=_NextPart_000_0055_01C158B9.D6DC4B40 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit High frequency performance of IBIS modelDear Chao, What you have raised is a question many people are asking these days. Here is my two cents worth. I feel that all components of the system you are trying to model need to be modelled to the same degree of accuracy. It may be that your IBIS model is not accurate enough for some of the reasons mentioned such as (absence of) frequency dependent parameters, fixed Ccomp etc. However, you may be making bigger assumptions in the rest of your simulation. For example, you may not be taking into account the power/ground noise in your system or the characteristics of the power delivery system. And even if you had a model which had everything - frequency dependent dielectric constant, frequency dependent loss tangent, skin effect, nonlinear circuit elements with the latest MOS model, full wave effects, etc. etc. - you probably would not find a simulator which can handle all this. I believe that IBIS models are far from being the weakest link in system simulation. Of course, I would still ensure that my IBIS models are reasonable. I would then ensure that I am simulating the rest of the system (planes, traces, vias, package pins, connectors, etc) to a sufficient degree of accuracy before trying to improve on the IBIS models. Best Regards, Raj Raghuram Sigrity, Inc. "Achieve what others can't" raghu@sigrity.com http://www.sigrity.com 4675 Stevens Creek Blvd. , Ste 130 Santa Clara, CA-95051 PH: 408-260-9344 x116 CELL: 408-390-7614 FAX: 408-260-9342 -----Original Message----- From: owner-ibis-users@eda.org [mailto:owner-ibis-users@eda.org]On Behalf Of Chao Su Sent: Friday, October 19, 2001 12:41 PM To: ibis-users@eda.org Subject: High frequency performance of IBIS model Hi, I'm working in signal integrity of optical transceivers. I'm considering of using IBIS model for my simulation, our optical transciever is operated at 2.5Gb/s. Somebody told me that IBIS model is not good to predict high frequency behavior for the device. I would like you can give me advice about what issue I need to concern when I get my IBIS model from vendors and how I can make sure the model is good in high frequency simulations. Thanks Chao Su Optical Systems Design (Dept. QF11) Nortel Networks P. O. Box 3511 Station C, Ottawa, ON, K1Y 4H7 Canada ---------------------------------------------------------------- (Phone: (613) 765-7688 (ESN 395-7688) *E-mail: csu@nortelnetworks.com +Mail Stop: 045/12/E06 ------=_NextPart_000_0055_01C158B9.D6DC4B40 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable High frequency performance of IBIS model
Dear=20 Chao,
 
What=20 you have raised is a question many people are asking these days. Here is = my two=20 cents worth.
 
I feel=20 that all components of the system you are trying to model need to be = modelled to=20 the same degree of accuracy. It may be that your IBIS model is not = accurate=20 enough for some of the reasons mentioned such as (absence of) frequency=20 dependent parameters, fixed Ccomp etc. However, you may be making bigger = assumptions in the rest of your simulation. For example, you may not be = taking=20 into account the power/ground noise in your system or the = characteristics of the=20 power delivery system. And even if you had a model which had everything = -=20 frequency dependent dielectric constant, frequency dependent loss = tangent, skin=20 effect, nonlinear circuit elements with the latest MOS model, full wave = effects,=20 etc. etc. - you probably would not find a simulator which can handle all = this.
 
I=20 believe that IBIS models are far from being the weakest link in system=20 simulation. Of course, I would still ensure that my IBIS models are = reasonable.=20 I would then ensure that I am simulating the rest of the system (planes, = traces,=20 vias, package pins,  connectors, etc) to a sufficient degree of = accuracy=20 before trying to improve on the IBIS models.
 
Best=20 Regards,
 

Raj Raghuram
Sigrity, Inc.
"Achieve what others=20 can't"
raghu@sigrity.com
http://www.sigrity.com
4675 Stevens Creek Blvd. , = Ste=20 130
Santa Clara, CA-95051
PH: 408-260-9344 x116
CELL:=20 408-390-7614
FAX: 408-260-9342

-----Original Message-----
From: = owner-ibis-users@eda.org=20 [mailto:owner-ibis-users@eda.org]On Behalf Of Chao = Su
Sent:=20 Friday, October 19, 2001 12:41 PM
To:=20 ibis-users@eda.org
Subject: High frequency performance of = IBIS=20 model

Hi,

I'm working in signal integrity of = optical=20 transceivers. I'm considering of using IBIS model for my simulation, = our=20 optical transciever is operated at 2.5Gb/s. Somebody told me that IBIS = model=20 is not good to predict high frequency behavior for the device. I would = like=20 you can give me advice about what issue I need to concern when I get = my IBIS=20 model from vendors and how I can make sure the model is good in high = frequency=20 simulations.

Thanks

Chao = Su=20
Optical Systems = Design (Dept.=20 QF11)
Nortel=20 Networks
P. O. = Box=20 3511
Station C, = Ottawa, ON,=20 K1Y 4H7
Canada=20
----------------------------------------------------------------= =20
(Phone: (613) 765-7688 (ESN 395-7688)
*E-mail:       =20 csu@nortelnetworks.com
+Mail=20 Stop:     045/12/E06=20


------=_NextPart_000_0055_01C158B9.D6DC4B40--