IBIS Accuracy Subcommittee Minutes
Friday, December 4, 1998, held by conference call.
Participants
Greg Edlund, IBM (chair)
Bob Haller, Compaq
Bruce Heilbrunn
Peter LaFlamme, Fairchild Semiconductor
The next meeting is Friday, December 11, 1998 from 11:00 to 12:00 am EST
BUSINESS
The subcommittee has decided to conduct its business during a one-hour
conference call on Fridays opposite the IBIS Open Forum call.
Bob Haller reported that the final copy of our DesignCon99 paper, "A Tour
of the IBIS Accuracy Specification," was submitted and accepted.
List of work to be done:
1. Prepare the test board design files for publication – Greg, 12/31/98
2. Write an example "accuracy trailer" using test board data – Bob, Peter,
12/31/98
3. Finish writing rev 1.0 of the spec – Greg, DesignCon99
4. Write an example "accuracy datasheet" using test board data – Greg, Bob,
Peter, DesignCon99
5. Write the "IBIS Accuracy Test Board Application Note" – Greg, Fawn,
DesignCon99
6. Write a Perl script to compute figures of merit – Bob, Fawn, DesignCon99
TECHNICAL
The subcommittee agrees on the need for some kind of accuracy trailer that
would be appended to an IBIS datasheet using comments. In the very least,
the accuracy trailer would contain the accuracy figures of merit for
various measurements, i.e. the results of applying the correlation metrics.
Peter proposed indicating the existence of an accuracy trailer in an IBIS
datasheet using the [Source] keyword, which has an open-ended format. Greg
favors adding a new IBIS keyword (in the long run) that could be used by
EDA vendors to set a flag somewhere in the GUI, indicating to the user that
the IBIS datasheet has been lab-verified.
In addition to an accuracy trailer, we may also put together an "IBIS
Accuracy Datasheet" that contains some of the same information as the
trailer but has the potential for more detail, such as simulation vs. lab
overlay plots. Examples of both the trailer and the datasheet would be
based on data taken from the VCX16244 on the test board.
Bruce proposed adding some text to the disclaimer in the accuracy trailer
to the effect of, "accuracy does not imply syntactical correctness." If a
user does not read the IBIS Accuracy Spec, he or she may have the notion
that accuracy and syntactical correctness are one in the same.
The subcommittee discussed the idea of amending the spec to include VT
tables. This has been suggested by several people as a means for
circumventing simulation vs. hardware correlation. We all agreed that we
would like to add VT tables to the spec in the future, but we don't have
the bandwidth to do it before DesignCon99. VT tables would not replace
simulation results; they would be an alternative to simulation results.
The IBIS datasheet developer would be free to choose and would be
responsible for documenting this choice to the user. Adding VT tables
would necessitate a BIRD to allow the three extra test loads called out in
the IBIS Accuracy Spec. The subcommittee feels strongly that these loads
are necessary for insuring IBIS datasheet accuracy and has data to support
this statement.
Agenda for the next meeting:
Business
1. DesignCon99 paper and booth – 5 min.
2. Opening the conference call to interested parties after DesignCon99 – 5
min.
3. The need for EIA approval of the IBIS Accuracy Specification – 20 min.
Technical
Review the accuracy trailer line by line. Please read and offer comments
before hand. – 30 min.
Greg Edlund
Advisory Engineer, Critical Net Analysis
IBM
3650 Hwy. 52 N, Dept. HDC
Rochester, MN 55901
gedlund@us.ibm.com
Received on Tue Dec 8 14:56:48 1998
This archive was generated by hypermail 2.1.8 : Fri Jun 03 2011 - 09:52:29 PDT