RE: [IBIS-Users] Buffer ramp rates & loading question.


Subject: RE: [IBIS-Users] Buffer ramp rates & loading question.
From: Peters, Stephen (stephen.peters@intel.com)
Date: Fri Feb 14 2003 - 13:04:26 PST


Hi John:

  Actually, a simulator uses both the I-V and V-T tables to perform the
simulation. In brief, V-T tables describe how a buffer output transitions
from one state to another, and if all four recommended tables are included
then a simulator knows exactly how to turn on or off a pullup or pulldown
when transitioning from one output state to the next. The [Ramp] keyword
supplies somewhat the same information, but obviously the V-T curves are
more detailed.

You are correct that the curve produced by the V-T table data can be used as
a reference; i.e. the simulator should reproduce the V-T table data when
simulating the buffer into the load documented by the [Rising
Waveform]/[Falling Waveform] keywords. Adding, for example, a waveform of
the buffer driving a capacitive load might be useful, but the simulator will
probably not use it to create the model. One still needs to include the
four waveforms into pure resistive loads.

On your third point you are correct - don't add the bond wire LRC data when
doing V-T extraction, or its effects will be double counted.

 Regards,
 Stephen Peters
 Intel Corp.

-----Original Message-----
From: John Phillippe [mailto:john.phillippe@motorola.com]
Sent: Friday, February 14, 2003 12:15 PM
To: ibis-users@eda.org
Subject: Re: [IBIS-Users] Buffer ramp rates & loading question.

Lynne Green wrote:

>
> 2. I have some slow I/O drivers that take a long time to settle for
> the rising and falling curves. To help the curves settle faster, I
> want to use 300 Ohms for the load. Normally I would use 50 Ohms load
> for the I/O drivers. I am wondering if I can use 300 Ohms load for the
> slow drivers? Is it going to be a simulator problem if some of the
> buffers use 50 Ohm loads, and others 300 Ohms loads?
>
>
>>>V-t tables should be built with the nominal load you expect to see in
>>
> an actual PCB trace. If the driver is intended for use with a wide
> range of trace impedances, you could build one set of V-t tables for
> the maximum load (say 300 Ohms), and another set of V-t tables for the
> minimum load (say 50 Ohms). IBIS supports up to 100 waveform tables.
>
>>>Simulators can use more than one set of V-t tables. A model is
>>
> generally more accurate if one set of V-t tables is "near" the actual
> load. (For example, if a driver could be used with any load from
> 50-300 Ohms, one could build a set of V-t tables at 50, another at
> 300, and perhaps others in between.)

I remember a while back that some people on this group had recommended
to just use the 50 ohm load and nothing else for V/T curves. Now what I
am sort of infering from the above post is to try and use a loading that
would accurately represent how the chip will be used in a board
environment. So this brings up a few new questions. FYI - I am using
S2ibis2 for ibis creation.

1. I have been under the assumption that in an IBIS model, the I/V
curves are what the simulator uses when doing board simulations. The
V/T curves are simply a nice little reference, but are not used for any
simulations. I.E. If I set my ibis model up in a simulator and then
put a bunch of 50 ohm loads to ground, and then simulate some edges, the
V/T curves are just there to "double check" the simulations. Is this a
correct way of thinking?

2. If the above is correct, than it would make sense to add more than
just the 50 ohm load models. I would want to add capacitative loads and
what not to simulate the original design specs. I.E. If a buffer was
designed to drive a 25pF load, than I would want to add that 25pF V/T
curve for reference.

3. In the V/T curve section (of s2ibis2 scripts) you can add bond wire
LRC values for the pin. It seems to me that I should leave this blank
when modeling, because when I put the chip together the LRC bond wire
values will be placed up in the pin list, and I would be "double
counting" the LRC values if I modeled them with the individual buffers
and also put them in the PIN list of the ibis model.

Thanks for all the good information so far.

-- 
John Phillippe
SPS, 32 Bit Embedded Controller Division, IC Creation
Motorola      -  512-895-1835
Austin, TX    -  john.phillippe@motorola.com

|------------------------------------------------------------------ |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 |------------------------------------------------------------------ |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 : Fri Feb 14 2003 - 13:10:51 PST