Kellee,
I notice that the 1994 Pentium book, a clamping diode is now represented as
an equivalent Spice formatted diode with "TT" for transit time. This may
be the critical item necessary for correct turn-on specification and it is
up to simulator companies to provide the "delayed" reaction in an appropriate
manner. However, I am also interested in more discussion for better
understanding and to sort out all the options.
Bob Ross,
Interconnectix, Inc.
> Bob I saw your Email, and I am also very interested in your items 1 and 4.
> (1) More details on the diode stored charge effects in clamping circuits and
> how to model.
> (4) More details on possible simplified package models with coupling and
> delay.
> I feel that the diodes are a very badly needed improvement to the specification.
> Many of the devices out there don't exhibit the instant turn-on times implied
> by the present IBIS specification. This has little effect on CMOS output buffers
> but has dramatic effect on all types of input buffers. Termination effects may be
> much more pronounced in a real circuit than simulators are showing.
> I would like to see something very simple to start with. Perhaps a turn on curve
> of percent ON versus time, where the percent on is relative to the diode current
> in the table. This could be a simple "turn-on-time". Another approach might be to
> use more of a real charge storage model. The trick is to make it measurable with
> device testers, and keep the amount of information that needs to be collected and
> simulated as small as possible. As IBIS models get larger the simulation speed goes
> down.
> Kellee Crisafulli,
> HyperLynx Inc.
Received on Tue Dec 6 18:55:38 1994
This archive was generated by hypermail 2.1.8 : Fri Jun 03 2011 - 09:52:28 PDT