Re: [SI-LIST] : newbye

Francesc Moll (moll@binghamton.edu)
Tue, 24 Mar 1998 15:40:05 -0500

Hi Fabrice,

I think your question does not have a straight answer, because it
depends on which kind of coupling you have between your lines.

If the coupling is mostly capacitive, (for example, high driver's
equivalent resistance), a rising transition induces a positive-going
glitch in the neighboring line. If this glitch is superposed to a rising
transition, the resulting effect may be a decreased transition delay. If
the glitch is superposed to a falling transition, the resulting effect
may be an increased delay. This depends on the waveform, on the relative
time between transitions in both lines, and also on the receiver's
switching threshold. Similarly happens for a falling transition.

However, if the coupling is mostly inductive, the glitch 'polarity' is
reversed with respect to the capacitive case: a rising transition
induces a negative-going glitch, and so on.

You could also add effects due to switching noise. Sure it is not easy
to say if the global result gives a 'bad' signal, which is probably the
reason why such tools do not exist (to my knowledge) and also why it
would be very interesting to develop this tool.

In any case, my opinion is that the main effect is a variation in the
expected transition delay. We published recently some measurements of
this variation for the case of coupling inside an integrated circuit,
which is mainly capacitive. The reference is:
F.Moll, M.Roca, A.Rubio "Measurement of crosstalk induced delay errors
in integrated circuits", El. Lett. vol.33 no.19 (11th Sep. 97).

Hope this helps. Maybe someone else has some more comments...

Francesc

-------------------------------------
Francesc Moll e-mail: moll@eel.upc.es
Dpt. of Electronic Engineering
Universitat Politecnica de Catalunya
(Until end of July at SUNY-Binghamton
Dpt. of Electrical Engineering)

Fabrice ILPONSE wrote:
>
> Hi!!
>
> I'm pretty new in this aspect of chip design and i'm more CAO
> oriented than electronician. I'm working on a tool that is suppose to
> analyse a net list and give out a list of "bad" signals.
> What i would like to know is the different phenomena appearing in
> different signal switching direction. I mean when both signals rise,
> one fall and the other rise, etc...
>
> If someone can help...
>
> [foregive me for the eventual mistakes :) ]
>
> --
> ^ ^ ^
> | | |
> +-+-+ Fabrice ILPONSE
> | email: fabrice@asim.lip6.fr
> |
> |
> -
>
>