Subject: RE: [IBIS-Users] Modeling input receivers with on-die termination
From: Muranyi, Arpad (arpad.muranyi@intel.com)
Date: Thu May 08 2003 - 14:52:17 PDT
Even though it may be somewhat more complicated to make a model that way,
it is better to separate the currents of the two clamps for the case when
you have power delivery network around the buffer's power and GND connections
and you are interested in the amount of noise in each. If you combined
everything into just one curve, say the GND clamp curve, all of your noise
will show up in the GND pad, and nothing in the power pad when your signal
makes the clamps work.
I completely agree with Mike Labonte, that you should extend the tables
-Vcc to 2 * Vcc, that way the tool will have everything defined. That's
why I developed the algorithms which I have mentioned before.
The only problem that you cannot avoid is that when your buffer has a
double termination, i.e. both pullup and pulldown together then the
"going to zero and extending" algorithm will still not account for the
correct currents from the power and GND pads, because it is not able
to separate the upper and lower resistance values. In order to do that,
you need to either be able to turn one or the other off, so you can
measure at least one of them separately, or you will need to start
separating things in the netlist (or on the die).
Arpad
===========================================================================
-----Original Message-----
From: Mike LaBonte [mailto:milabont@cisco.com]
Sent: Thursday, May 08, 2003 2:16 PM
To: Todd Westerhoff; Ibis-Users (E-mail)
Subject: RE: [IBIS-Users] Modeling input receivers with on-die
termination
There is a way to deal with any uncertainty over how simulators will treat
the "missing" data in power clamp and gnd clamp curves - don't have any
missing data! If both curves extended from -VDD to 2VDD there would be no
guessing. Of course if you are going to go that far, why not just model both
clamps in one curve (Proposal 1). But even if you do Proposal 2, don't end
the curves at the zero current point. Take the zero current all the way to
the other end of the voltage range. Don't make the simulators guess what the
remaining data is.
An observation: characterizing both clamps in one V/I curve is more or less
a black-box approach, in that it does not assume any knowledge that the
buffer has separate diodes connected to power and ground. It just measures
how the buffer behaves when tristated. I like the simplicity of that. Now if
we measured the static behavior of the buffer in 2 other states and added
dynamic info on how to make the transitions from each state to the others we
would have a truly behavioral model, with no notion of structure. In our
attempt to model the behavior of buffers at 3 ports (pad, power, ground)
IBIS became complicated.
Mike
|------------------------------------------------------------------
|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 : Thu May 08 2003 - 15:14:30 PDT