Re: Switched clamping diodes

Text item:

John,

You might understand the situation of I-V curves a little better if you compare
the case of the I/O and Output only (non 3-statable) buffers.

If you make a model for an output only buffer, the clamp curves will have to be
part of the pullup/pulldown I-V curves. This is because you cannot 3-state the
device to measure the clamps separately by themselves. (In a way both pullup or
pulldown are always present together with the clamps).

The quote you cited below refers to 3-statable buffers or I/O-s, in which the
clamps are always present, while the driver section of the buffer may not be
there all the time (when the device acts as a receiver, for example). The
intention in IBIS was to separate the static "constantly present", and the
switched "when needed" characteristics of the buffers somehow.

Now, if you have a buffer that has a switched clamp, I don't see anything wrong
with combining the switched clamp I-V curves with the pullup or pulldown I-V
curves, as long as they are switched together. If they are switched separately
in time, you might need to model things differently, which might not be possible
with IBIS directly. If the simulator tool you are using is flexible enough, you
might be able to get around this problem by making two models, one to represent
the pullups/pulldowns and another to represent the clamps and then driving them
with two independent clocks or enables.

By the way, some 5 V safe 3.3 V buffers do have constantly present clamps, but
they clamp above 5 V only. If a CMOS device has a p-channel pullup, the pullup
I-V curve will cross 0 mA at 3.3 V and draw current between 3.3 and 5 V due to
the symmetrical nature of the CMOS channel. I would not call this a clamp,
however, since this current is part of the pullup transistor's characteristics
and not it's parasitic diode or a separate ESD structure

I hope this helps in clarifying the usage of I-V curves.

Arpad Muranyi
Intel Corporation
===============================================================================

All,

Thanks to everyone who replied to my earlier mail to ibis-users.

>From the answers, I don't think my 2nd point was
expressed clearly enough

> 2) One family of components switches out the power clamp
> diode whenever the output buffer is in tri-state.

This family is 5V tolerant LVC from Philips (and TI). When in
3-state, the clamping diode to Vcc is "removed". However, in
active state, it is present.

In the IBIS spec, it is assumed that:
   "the data in the clamping curve sections are handled
    as constantly present curves and the pullup and
    pulldown curves are used only when needed"

This assumption is false for LVC. The clamping curves are not
constantly present!

A work-around for 2.1 is to put the clamping curve in with the
pull-up and pull-down curves. As long as the 3-state curve is
truly high-impedance, the model will be very good.

For 3.0, perhaps it would it be an idea to have separate curves
for active and 3-state modes?

2.1
       3-state: [3-state] = [GND_clamp] + [POWER_Clamp]
        Active: ( [Pullup] or [Pulldown] ) + [3-state]

3.0 Proposal:
       3-state: [3-state] = [GND_clamp] + [POWER_Clamp]
        Active: [Pullup] or [Pulldown]

In order to reduce confusion, it would be better to rename
[GND_clamp] and [POWER_clamp] to [3-state pullup] and
[3-state pulldown]. Then, if the simulator detects the new keywords,
it will know that the definition of [Pullup] and [Pulldown]
has changed. In this way, a 2.1 description would still be valid
in 3.0.

Any comments?
John

--
John Fitzpatrick   <John.Fitzpatrick@ln.cit.alcatel.fr>
Alcatel CIT, 4 rue de Broglie, 22304 Lannion, France
Tel: (+33)96.04.79.33  Fax: (+33)96.04.85.09
Text item: External Message Header
The following mail header is for administrative use
and may be ignored unless there are problems.
***IF THERE ARE PROBLEMS SAVE THESE HEADERS***.
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=us-ascii
Subject: Switched clamping diodes
To: ibis@vhdl.org
Mime-Version: 1.0
X-Mailer: Mozilla 2.0 (X11; I; SunOS 4.1.3_U1 sun4c)
Organization: Alcatel CIT, 22304 Lannion, France
From: FITZPATRICK John <John.Fitzpatrick@ln.cit.alcatel.fr>
Date: Mon, 22 Apr 1996 12:30:55 +0200
Message-Id: <317B5FDF.1CFBAE39@ln.cit.alcatel.fr>
Sender: fitzpat1@ln.cit.alcatel.fr
Received: from las41247 (localhost) by las41247.ln.cit.alcatel.fr (4.1/SMI-4.1)
     id AA00371; Mon, 22 Apr 96 12:30:57 +0200
Received: from las41247.ln.cit.alcatel.fr by ln.cit.alcatel.fr (4.1/SMI-4.1)
     id AA07812; Mon, 22 Apr 96 12:31:00 +0200
Received: from ln.cit.alcatel.fr (las40304.ln.cit.alcatel.fr [155.132.225.2]) by
 nsfhh5.alcatel.fr (8.7.3/8.7.3) with SMTP id MAA24775 for <ibis@vhdl.org>; Mon,
 22 Apr 1996 12:28:07 +0200 (METDST)
Received: from alcatel.fr (gatekeeper-ssn.alcatel.fr [155.132.180.241]) by mailg
ate.alcatel.fr (8.7.3/8.7.3) with ESMTP id MAA29583 for <ibis@vhdl.org>; Mon, 22
 Apr 1996 12:27:34 +0200
Received: from alcatel.fr (mail.alcatel.fr [193.104.30.131]) by vhdl.vhdl.org (8
.7.3/8.7.3) with ESMTP id DAA07877 for <ibis@vhdl.org>; Mon, 22 Apr 1996 03:37:4
0 -0700 (PDT)
Received: from vhdl.vhdl.org (vhdl.vhdl.org [198.31.14.3]) by mailbag.jf.intel.c
om (8.7.4/8.7.3) with ESMTP id DAA22346; Mon, 22 Apr 1996 03:35:11 -0700 (PDT)
Received: from mailbag.jf.intel.com (root@mailbag.jf.intel.com [134.134.192.4])
by relay.jf.intel.com (8.7.4/8.7.3) with ESMTP id DAA20722; Mon, 22 Apr 1996 03:
35:55 -0700 (PDT)
Return-Path: owner-ibis@vhdl.vhdl.org