Re: [IBIS-Users] Making Pullup curves VCC relative.


Subject: Re: [IBIS-Users] Making Pullup curves VCC relative.
From: Abdulrahman A. Rafiq (arafiqsi@yahoo.com)
Date: Wed Oct 16 2002 - 00:45:33 PDT


Ibis Gurus,
This brings up another question I have had for sometime now. I am wondering if there is a white paper on generating ibis models in the lab. I have upto now translated spice models to ibis, however if i ever needed to, how could i generate an ibis model from lab measurments?
Thanks,
Abdulrahman Rafiq Al Davis <aldavis@ieee.org> wrote:
On Tuesday 15 October 2002 02:58 pm, Ingraham, Andrew wrote:
> "VCC relative" means that 0V in the table, corresponds to Vin
> or Vout = VCC at the device pin or pad.

What confuses people is that the sign is flipped from what it
logically ought to be. If you just flip the sign of the
voltage, it makes sense.

>
> In addition, the tables are arranged so that positive table
> values mean I/O voltages mostly within the normal range
> between GND and VCC, and negative table values are voltages
> below GND or above VCC, toward the clamps. (Obviously, if
> you go far enough in either direction, you will go beyond GND
> and VCC. Large positive table values correspond to voltages
> beyond the opposite supply rail.)
>
> Vtable = Vcc - Voutput.

Read this carefully .... It means the ground lead of your
scope is hooked to the output, and the tip of the probe to the
power supply.

The pullup is just like the pulldown except for the sign flip.
It is a i/v table describing a nonlinear resistance,
representing the channel of the output device. For pulldown,
it is hooked from output to pulldown_reference. You measure
the voltage across it. It just happens that pulldown_reference
is often the same as ground, so the voltage to ground happens
to match. For pullup, that device is hooked from output to the
power supply (pullup_reference) and you measure the voltage
across it. Unfortunately, in the standard, the sign of the
voltage is flipped. Since the standard cannot be changed
without breaking compatibility, we are stuck with it.

> This formula is applied only to the [Pullup] and [Power
> Clamp] tables. No such adjustment is made to the [Pulldown]
> and [GND Clamp] tables.

except for ECL types.

> The step of subtracting the clamp currents from the simulated
> or measured currents, to get the values for the [Pullup] and
> [Pulldown] tables, is different. It has nothing to do with
> the fact that the [Pullup] and [Power Clamp] tables are VCC
> relative. They are two independent steps in the process.
> Subtracting the clamp currents prevents double counting when
> they are added back together by the simulator.

But be careful when you subtract. The way most people do it,
they subtract two very large numbers and come up with a small
number as the result. The pullup and pulldown curves are often
nonsense in this region, showing something non-monotonic, which
is usually wrong and sometimes causes trouble for simulators,
if your simulator handles ground bounce. If you are not
modeling ground bounce, they are in parallel and you don't see
the problem. It is only a problem if pulldown_reference and
ground_clamp_reference are different nodes, or likewise if
pullup_reference and power_clamp_reference are different nodes.

A better approach is to take the derivatives of both curves,
subtract the derivatives, smooth out the nonsense in the clamp
region, and then integrate it. Simply dropping points is the
best way to smooth the nonsense.

The data in this region is questionable anyway. Its main
purpose is to put the simulator back on track when a stray
iteration sends it there. It is important that it be well
behaved. It is not so important for it to be accurate. This
is fortunate, because you can't be accurate. If you try it
with a real device, you will destroy it. If you try it with a
simulator, you are operating devices in regions the models were
not designed for.

You should always look at the derivatives of your curves
anyway. The derivative is more important than the data itself,
and is often loaded with noise due to rounding and measurement
noise. Take the derivative of your actual data tables and plot
it. You may be in for a big surprise.
|------------------------------------------------------------------
|For help or to subscribe/unsubscribe, email majordomo@eda.org
|with just the appropriate command message(s) in the body:
|
| help
| subscribe ibis
| subscribe ibis-users
| unsubscribe ibis
| unsubscribe ibis-users
|
|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

---------------------------------
Do you Yahoo!?
Faith Hill - Exclusive Performances, Videos, & more
faith.yahoo.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



This archive was generated by hypermail 2b28 : Wed Oct 16 2002 - 00:56:45 PDT