Subject: RE: [IBIS-Users] (lv)pecl
From: Henrik G. Madsen (hgm@vitesse.com)
Date: Wed May 14 2003 - 07:49:36 PDT
Relating to garbage lines
We are using the s2ibis2 (windows version) and having similar problems. They starts when the supply voltage is reduced below 3.3 V and gets worse and worse as the supply voltage drops. So the issue is worse for 1.8 V models than 2.5 V models etc.
I brought up the issue around a year ago on the list with no replies.
The Spice results files are OK, so the problems relates to the way that s2ibis2 reads them.
The 'workaround' we use it to extract the missing/erronous points from the simulation files and adding them to the IBIS file manually.
Regards
/Henrik G Madsen
-----Original Message-----
From: erik.van.der.ven@philips.com [mailto:erik.van.der.ven@philips.com]
Sent: 14. maj 2003 15:25
To: ibis-users@eda.org
Subject: RE: [IBIS-Users] (lv)pecl
Tom,
Do you need the netlist as well, or is the plain s2i file good enough for you? (because I cannot send you the spectre file).
Birendra seems to have a similar problem.
Alex,
I am using Spectre for simulations.
With kind regards,
Erik.
Ir. Erik van der Ven
Room DB1032
Business Line Networking Infrastructure
Philips Semiconductors BV
Gerstweg 2
6534 AE Nijmegen
The Netherlands
Phone: +31-24-3534334
"Tom Dagostino" <tom@teraspeed.com>
05/12/03 05:53 PM
Please respond to tom
To: Erik van der Ven/NYM/SC/PHILIPS@EMEA2
cc:
Subject: RE: [IBIS-Users] (lv)pecl
Classification:
I've seen some other models with this kind of issue in them from s2ibis. I'm not sure what causes this, it was a while ago and the half life of old problems is relatively short. I would first start by looking at the input file that is generated. Be sure you do not have any typos in it, that the voltages are correct, etc. Send me a copy if you want and I'll take a look at it.
Tom Dagostino
Teraspeed Consulting Group LLC Teraspeed Consulting Group LLC
2926 SE Yamhill St. Device Modeling Division
Portland, OR 97214 13610 SW Harness Lane
Beaverton, OR 97008
http://www.teraspeed.com 503-430-1065
tom@teraspeed.com
-----Original Message-----
From: owner-ibis-users@server.eda.org [mailto:owner-ibis-users@server.eda.org]On Behalf Of erik.van.der.ven@philips.com
Sent: Monday, May 12, 2003 12:37 AM
To: ibis-users@server.eda.org
Subject: RE: [IBIS-Users] (lv)pecl
Tom,
About the garbage lines:
I use s2ibis for model type output_ECL.
Voltage range has been left out (since this created a voltage source of 1.9V to the ground node in the spectre simulation files).
Pull-up and pull-down references have been set to 3.3V.
The result is an IBIS file that contains a pulldown-curve starting at 1.1V and continuing up to 3.3V.
Then the next voltages are: 2.05, 9 times 0V and 3.3V. Most currents are equal to the 3.3V current except for the max current, which is increasing for every line (note that I filled in the typical values for the the min and max cases).
The pull-up curve ranges from 0V to 2.2V and is then followed by 10 times 0V and 2.2V. The current for the 0V lines is always 0.000A, except for the first one.
So I get some strange values and the range is from 0-2.2V instead of -3.3 to 6.6V.
With kind regards,
Ir. Erik van der Ven
Room DB1032
Business Line Networking Infrastructure
Philips Semiconductors BV
Gerstweg 2
6534 AE Nijmegen
The Netherlands
Phone: +31-24-3534334
"Tom Dagostino" <tom@teraspeed.com>
05/09/03 07:11 PM
Please respond to tom
To: Erik van der Ven/NYM/SC/PHILIPS@EMEA2
cc:
Subject: RE: [IBIS-Users] (lv)pecl
Classification:
Erik
In the old days (in the 60s) when Motorola first developed ECL They found that running ECL from 0 to -5.2 volts worked better because the ground planes on boards at the time was cleaner. ECL is referenced to the most positive supply, thus a clean reference was important. In those days the ground could have been a complete plane of the board but power was in many cases distributed via traces on one of the layer. In today's modern circuit boards both the power and ground are distributed in planes so this is not an issue.
The voltage difference you see is not important. When you are running the SPICE simulation you just need to connect the Vcc to 3.3 and the Vee to ground to get the PECL model. Remember the 50 Ohm termination needs to be connected to Vcc - 2 volts or 1.3 for the typical case when running the AC simulations.
I do have concerns about the table range you are seeing. It should be between -3.3 and +6.6 Volts for all the IV tables.
The simulator does several things with the data. First of all it will check the data to ensure it is at least consistent, has not syntax issues, etc. Then, for outputs, it builds a set of coefficients that describes how the buffer transitions from pullup to pulldown and from pulldown to pullup. This table is based on the VT waveforms and the IV curves.
The output impedance of the buffer is composed of the sum of the pullup or pulldown IV table plus the power and ground clamps plus the component of C_comp.
If the buffer is an input the the power and ground clamps are summed and put in parallel with C_comp.
Because the model is designated ECL, the simulator makes one different calculation for the pulldown table than it does for CMOS, see below.
What is the 10 lines of garbage you are seeing?
Hope that helps.
Tom Dagostino
Teraspeed Consulting Group LLC Teraspeed Consulting Group LLC
2926 SE Yamhill St. Device Modeling Division
Portland, OR 97214 13610 SW Harness Lane
Beaverton, OR 97008
http://www.teraspeed.com 503-430-1065
tom@teraspeed.com
-----Original Message-----
From: owner-ibis-users@server.eda.org [mailto:owner-ibis-users@server.eda.org]On Behalf Of erik.van.der.ven@philips.com
Sent: Friday, May 09, 2003 12:41 AM
To: ibis-users@server.eda.org
Subject: RE: [IBIS-Users] (lv)pecl
Tom,
Thank you for your reply. I am starting with a Spectre netlist. I have tried s2ibis to make a model for LVPECL.
I noticed some differences. ECL has a supply of 0V (plus) and -5.2V (neg). My circuit has a supply of 3.3V (plus) and 0V (neg).
The pulldown table ranges from 1.1V to 3.3V (the pull down reference has been set to 3.3V, the pull-up reference as well), followed by ten lines of garbage. A collegue of mine has changed to pragram a bit so the pull down curve starts at 0V and ends at 2.2V (still followed by ten lines of garbage).
Is this a correct range?
What does the simulator that uses the IBIS model do with the information that it is an output-ECL model type? And how are the data tables interpreted in conjunction with the ECL_model type?
With kind regards,
Ir. Erik van der Ven
Room DB1032
Business Line Networking Infrastructure
Philips Semiconductors BV
Gerstweg 2
6534 AE Nijmegen
The Netherlands
Phone: +31-24-3534334
"Tom Dagostino" <tom@teraspeed.com>
05/08/03 07:16 PM
Please respond to tom
To: Erik van der Ven/NYM/SC/PHILIPS@EMEA2
cc:
Subject: RE: [IBIS-Users] (lv)pecl
Classification:
Erik
What are you starting with? A SPICE description or are you going to make the model from measurements? If you start with SPICE there is a spice to ibis program that you can download from the ibis home page. If you are starting from measurements then you will need a lot of equipment to make the measurements. You will have to manually convert the measurements to the ibis data.
Simulators use the output_ecl line to know how to handle ECL. In IBIS ECL models are slightly different than standard CMOS or TTL drivers. In CMOS or TTL the pulldown current is sunk into the ground plane and the pullup current is sourced from the supply. With PECL both the pullup and pulldown are sourced from the supply. So in PECL the pulldown is referenced to the supply, not ground. The simulator needs to know this when it is using the IV tables.
Tom Dagostino
Teraspeed Consulting Group LLC Teraspeed Consulting Group LLC
2926 SE Yamhill St. Device Modeling Division
Portland, OR 97214 13610 SW Harness Lane
Beaverton, OR 97008
http://www.teraspeed.com 503-430-1065
tom@teraspeed.com
-----Original Message-----
From: owner-ibis-users@server.eda.org [mailto:owner-ibis-users@server.eda.org]On Behalf Of erik.van.der.ven@philips.com
Sent: Thursday, May 08, 2003 12:04 AM
To: ibis-users@server.eda.org
Subject: [IBIS-Users] (lv)pecl
Dear all,
I think I mailed my question too early since I subscribed to the user group at the same time. So I'll post it again, please see below.
Ir. Erik van der Ven
Room DB1032
Business Line Networking Infrastructure
Philips Semiconductors BV
Gerstweg 2
6534 AE Nijmegen
The Netherlands
Phone: +31-24-3534334
----- Forwarded by Erik van der Ven/NYM/SC/PHILIPS on 05/08/03 09:03 AM -----
Erik van der Ven
05/06/03 02:01 PM
To: ibis-users@eda.org
cc:
Subject: (lv)pecl
Classification: Unclassified
1. How can I make an IBIS model of a (LV)PECL model?
2. What does an IBIS simulator do with the model type information (like output_ECL)?
Can anyone help me?
Thanks.
With kind regards,
Ir. Erik van der Ven
Room DB1032
Business Line Networking Infrastructure
Philips Semiconductors BV
Gerstweg 2
6534 AE Nijmegen
The Netherlands
Phone: +31-24-3534334
|------------------------------------------------------------------
|For help or to subscribe/unsubscribe, email majordomo@eda.org
|with just the appropriate command message(s) in the body:
|
| help
| subscribe ibis <optional e-mail address, if different>
| subscribe ibis-users <optional e-mail address, if different>
| unsubscribe ibis <optional e-mail address, if different>
| unsubscribe ibis-users <optional e-mail address, if different>
|
|or email a written request to ibis-request@eda.org.
|
|IBIS reflector archives exist under:
|
| http://www.eda.org/pub/ibis/email_archive/ Recent
| http://www.eda.org/pub/ibis/users_archive/ Recent
| http://www.eda.org/pub/ibis/email/ E-mail since 1993
This archive was generated by hypermail 2b28 : Wed May 14 2003 - 07:59:59 PDT