RE: IBIS User Group Minutes - 4/23/98

Kellee,

I think the reason Hyperlynx was not on the list was that we were
tracking the progress of EDA vendors who had NOT yet developed a free
model development kit as Hyperlynx already has. Perhaps this wasn't
real clear from the minutes. I can certainly make a note of this in the
Accuracy Subcommittee minutes, which will come out this week.

Thanks for looking at the test loads. Getting rid of the transmission
line would certainly eliminate some variables. Electrically, an ideally
terminated transmission line and a resistive load are identical, but
real boards are not ideal. The reason we decided to put the probe point
at the end of a transmission line in our test board is that we couldn't
place all of our probe points right next to the DUT and still route
around them. (We have twice as many loads as necessary on the test
board to demonstrate different probing technologies.) I think we should
mention in our documentation for the board that placing a resistive load
right next to the DUT would be desirable. Good point about the probe
parasitics, too.

To add to what you said about the purpose of doing the test board, it is
really an example of how one might implement the IBIS Accuracy
Specification to demonstrate lab vs. simulation correlation. Other
people will come up with other ways, but this is a start. We're hoping
we can dodge the "simulation engine question" (i.e. where do the various
engines begin to disagree with each other and SPICE?) by sticking to
push-pull and open drain drivers at roughly IBIS 1.1 level for the first
version of the spec. The reason we'd like every EDA vendor to
distribute a model development kit is so that IC vendors can chose one
that meets their needs as well as the needs of their customers.

Thanks for taking the time to write. I really appreciate the input.

Greg
----------
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

        ----------
        From: Kellee Crisafulli[SMTP:kellee@hyperlynx.com]
        Sent: Friday, May 01, 1998 7:46 PM
        To: ibis-users@eda.org
        Subject: Re: IBIS User Group Minutes - 4/23/98

        Hi all,

        I have 2 items:

        1) HyperLynx was left out of the list of vendors supplying free
        simulation engines to IC vendors: We were the first to
        offer this back in Jan. at the Design Con IBIS meeting and
        I bet we have by far supplied the most free packages to IC
        vendors to date.
        I would very much appreciate being mentioned in the same list
        as those that also agreed at a later date to do the same.
        We have already supplied a large number of packages to IC
vendors
        free of charge and have been doing so to various degrees
        since IBIS V0.9 was released years ago.

> Request from Bruce Helbrunn to EDA vendors that free
simulation
>engine be made available to IC vendors has received
>a positive response from Quad Design, Interconnectix/Mentor and
Cadence

        2) I reviewed the test loads that Greg Edlund sent out and
           noticed there might be some room for improvement I offer
           the following for your consideration.

        All the following assumes this validation is against a real test
board
        you have. If the goal is to just compare the simulators than
ignore this
        but there may be a problem determining just which simulator is
correct.

        After listening to Greg's talk at the Jan. 1998 Design Con IBIS
        meeting I recall two goals for his test board:
        1) To validate the IBIS model data matches the real device under
test.
        2) To insure all the simulators work with the models and provide
           the same simulation results into real boards (i.e.
transmission
           lines).
        Hopefully I am not confusing his talk with one of the others.

        I believe the loads shown do a good job with item 2 but item 1
        might benefit from a change.
        I would like to see two additional cases for push/pull that
        have pull up/down resistors without a transmission lines.
        This allows the V/T table to be validated without the
interaction
        of the transmission lines.

        Also for the purposes real board validation the probing ground
lead length,
        probe capacitance and probe locations should be specified.

------------------------------------------------------------------------
-
        Have a great day...
        Kellee Crisafulli at HyperLynx
        kellee@hyperlynx.com http://www.hyperlynx.com

------------------------------------------------------------------------
-
Received on Mon May 4 05:22:46 1998