IBIS Accuracy Subcommittee Minutes 5/21/98

IBIS Accuracy Subcommittee Minutes

Thursday, May 21, 1998
Held at Digital Equipment's PKO3 building, 129 Parker St., Maynard, MA.

Attendees

Greg Edlund, Digital Equipment (chair)
Fawn Engelmann, EMC
Bob Haller, Digital Equipment
Bruce Heilbrunn, Stratus Computer
Peter LaFlamme, Fairchild Semiconductor
Harvey Stiegler, Texas Instruments (by phone)

Next meeting is Thursday, June 18, 1998 from 3:00 to 5:00 pm at Stratus
Computer in Marlboro, MA.

MILESTONES

The test board milestone has slipped again.

July, 1998: Post the IBIS Model Test Board to the web
September 18, 1998: Distribute the first draft of the IBIS Accuracy
Specification
October 1998: PCB West Conference, IBIS Summit, and first IBIS Class
February 1999: Present the IBIS Accuracy Specification at DesignCon99

EDITING

1. Scope

The subcommittee reviewed the pre-rough draft document that Bob Haller
is writing. Bob has compiled the material generated by the subcommittee
placed it in a Word document as a means of recording and benchmarking
our progress. It is the consensus of the subcommittee that the Scope
section of the specification is in good enough shape for us to focus on
the following sections.

We agreed to add to the Introduction some language that addresses the
general topic of "uncertainties." Differences between simulators has
been a long-standing source of concern. While we will not directly
address this topic in this version of the specification, we feel we need
to make the reader aware that it exists. Another source of concern is
that most sample parts that may be used to verify model accuracy are
really "random samples," i.e. little or nothing is known about the
semiconductor processing conditions in their lot of origin.

Footnote related to [Waveform] keywords: Discussion with Bob Ross
following the subcommittee meeting revealed an important fact that we
had not previously understood. We were under the impression that the
[Rising Waveform] and [Falling Waveform] keywords were added to IBIS to
cover more advanced driver types such as GTL, which can employ a
feedback circuit for edge rate control. For this reason, we wanted to
steer away from the [Waveform] keywords because we felt they would add
too much complexity for us to handle in revision 1.0 Bob stated that
the two primary purposes of these keywords are as follows: 1) to allow
a simulator to adjust its internal ramp stimulus to better fit actual
waveforms under a specified load and 2) to more accurately represent the
phase relationship between gate waveforms for the pull-up and pull-down
devices in a push-pull driver. With this understanding, it is necessary
to review the [Waveform] keywords for possible inclusion into the scope
of the IBIS Accuracy Specification 1.0.

2. Required Measurements

Peter La Flamme identified a component which we can use as an example
open drain component for verifying our test loads. This component is
Fairchild's GTLP8T306, an 8-bit LVTTL to GTL bus transceiver. The
output buffer does not employ a feedback circuit and would be
well-suited for use in the IBIS Accuracy Specification 1.0. SPICE
models are available.

There are three essential sub-sections under Required Measurements:
test loads for transient waveforms, capacitance, and IV curves. We
discussed test loads thoroughly during the last meeting; this topic
seems to be well under control for now. The goal of this meeting was to
decide on what measurements to require for capacitance and IV curves.

We discussed two possible approaches to capacitance measurements. One
would be to choose the pins which the modeling engineer expects to yield
the minimum and maximum capacitance and measure these only. The other
would be to measure the capacitance of every pin in a die-less package;
obviously this method isn't of much use with high pin-count-packages.
The ideal situation would involve predictions made by a 3-D field
solver, although HOW the modeling engineer gets the capacitance data is
not really relevant to the specification as it is defined today. The
important point is that the modeling engineer measures capacitance and
documents the measurements.

A more general topic arose related to coverage of each unique I/O cell
design. Do we require a complete set of transient waveform,
capacitance, and IV curve measurements for each unique I/O cell design
present in an IC? This is quite feasible for a simple part such as a
244 buffer, but it is totally unrealistic for an ASIC that may have a
hundred or more I/O cells in its library.

We did not close on the topic of Required Measurements during this
meeting. We will attempt to carry on the discussion electronically so
we can continue to make progress toward our milestones.

Bruce Heilbrunn requested that we table the topic of requiring 10 Ohm
and 100 Ohm test loads until a later date. He is looking for an IBIS
level 1.1 driver that exhibits a good fit between simulations and lab
measurements at 50 Ohms but a poor fit at extreme impedances.

3. Measurement Techniques

We did not discuss measurement techniques at this meeting. The June and
July meetings will be dedicated to this topic.

4. Metrics

We did not discuss comparison metrics at this meeting. The August
meeting will be dedicated to this topic.

RELATED ACTIVITIES

IBIS Accuracy Test Board - The test board is in routing at EMC. When
routing is complete, we will circulate 1:1 layer plots for review prior
to going for etch.

IBIS Developers Tool Kit - No progress to report.

IBIS Cookbook - No progress to report.

HOMEWORK

All - Continue and close the discussion of Required Measurements via
email.

Greg Edlund - Finish routing of the IBIS Accuracy Test Board.

Fawn Engelmann - Finish routing of the IBIS Accuracy Test Board.

Bob Haller - Finish Required Measurements section.

----------
Greg Edlund, Principal Engineer
Server Product Development
Digital Equipment Corp.
129 Parker St. PKO3-1/20C
Maynard, MA 01754
(978) 493-4157 voice
(978) 493-0941 FAX
greg.edlund@digital.com
Received on Mon Jun 1 10:01:21 1998