RE: Questions on the relationship between IBIS model & datasheet

When Vol is specified at 0.55 volts and 24 mA it means the part will put out
less than 0.55 volts when 24 ma of current is being sunk into the output.
This is what the manufacturer "guarantees". Since the manufacturer wants to
ship all possible die from a wafer the output spec is conservative, all
parts should meet this spec. It is not unusual to see 2 to 5 time the spec
current at the specified output voltage. If the model shows less than the
specified current at the specified voltage I would be suspect of the model.
I have rarely seen a part that would not meet its Vol or Voh spec. I have
seen many parts that would product more current at a given voltage than the
model said it should (measured Iol at Vol was greater than the Max in the
model).

Tom Dagostino
IBIS and Tau Modeling Manager
SDD
Mentor Graphics Corp.
503-685-1613
tom_dagostino@mentor.com

-----Original Message-----
From: Ingraham, Andrew [mailto:Andrew.Ingraham@compaq.com]
Sent: Tuesday, December 05, 2000 7:54 AM
To: 'Su Ming TAI'
Cc: si-list@silab.eng.sun.com; ibis-users@eda.org
Subject: RE: Questions on the relationship between IBIS model &
datasheet

> I have a datasheet and an IBIS model of a same device. On the datasheet
> (see pg 4), the Ioh and Iol values are -24mA and 24mA respectively when
> Vcc
> is 3V.
 
Those particular values mean that they do not want you to sink or source
more than 24mA at Vcc=3V. The device may be capable of sinking or sourcing
more than that, but those are the recommended maximum output currents.

> Now, on the IBIS model, at the pullup and pulldown section, when the
> voltage is at 3V, the current are not anything near to 24mA or -24mA at
> all.
 
Careful here. The "voltage" that is 3V, is Vcc, the supply voltage. It is
NOT the axis or independent variable of the [Pullup] and [Pulldown] curves!

> My question is:
> 1) Why is that the datasheet values do not seem to correlate with the IBIS
>
> model?
 
The IBIS datasheet (IBIS model) hopefully represents actual performance of
the device. The Spec sheet (what you called datasheet but I need to call it
something else to differentiate it from the IBIS datasheet) describes how to
use the part. Some of the information, for example page 4, puts constraints
on how you may use it. Other sections, for example page 5, give the
absolute worst-case performance you can (hopefully) ever expect, even when
they migrate the part to a new IC fab which could result in a different IBIS
datasheet being generated.

There is almost always some "padding" included in the Spec sheet data. For
example, the minimum tpd (according to all simulations and projections made
by the vendor) might be 1.7ns, hypothetically, but the vendor publishes 1ns
to be safe. It's a marketing decision. They might not add the same amount
of "padding" to their IBIS models, which are intended to better represent
actual performance.

In other words, when comparing "apples to apples" (identical
characteristics), I would not be surprised if the Spec sheet was more
conservative than the IBIS datasheet.

But let's take a closer look at your example. The spec sheet on page 5 has:

  Voh(min) = 2.2V @ Vcc = 3V, Ioh = -24mA
  Vol(max) = 0.55V @ Vcc = 3V, Iol = 24mA

Voh = 2.2V means Vcc-Voutput = 0.8V. Checking the IBIS [Pullup] curve, when
Voltage = 0.8V, I(min) = -21.8948mA, which is close to -24mA. It doesn't
quite meet the Spec. At Ioh = -24mA, Vcc-Vout = slightly more than 0.9V, so
Vout = slightly less than 2.1V. Not quite 2.2V minimum.

For Vol = 0.55V, the IBIS [Pulldown] curve says I(min) = between 32.8243mA
and 38.6213mA. For Iol = 24mA, Vout = about 0.36V. This meets the Spec
sheet (max. 0.55V).

> 2) How do we know that the IBIS model is correctly generated based on the
> values from the datasheet?
 
I would hope that both the IBIS datasheet and the Spec sheet are generated
from simulations and/or measurements, not one from the other. Neither sheet
alone has sufficient information to generate the other.

The IBIS datasheet has considerably more data about I vs. V, so it could
never be generated from a Spec sheet alone, but the IBIS datasheet usually
lacks the Absolute Maximum and Recommended Maximum conditions that are
included on the Spec sheet.

Regards,
Andy

 
Received on Tue Dec 5 12:04:51 2000