Re: Clarifing example for Packaging EGG

J Eric Bracken <bracken@kevily.ece.cmu.edu> writes:

  The one thing I _do_ worry about here is how the model MAKERS will
go about producing the coupling information...using field solver software?
Doing cut-and-try on measured data? With a dart board? This could
really affect the accuracy of results, particularly when different vendors
are doing their own thing.

        Aye, there's the rub. Package effects (crosstalk, ground/power
        bounce, etc.) often dominate or determine the device behavior,
        so a discretionary approach to packages can make the models useless
        if the modeller is not careful. Drivers and receivers cannot be
        modelled correctly without their packaging, unless the packaging
        is so good that its effects are negligible. Maybe vendors should
        put more effort into making their packaging "invisible"... Until
        that happens, for IBIS to be useful it must include a prescription
        for package modelling.
        
        In my experience, the only useful models are based on measured
        _AC_ data from actual parts. No 2-1/2D or 3D modeller I know of
        gives accurate results to useful precision in reasonable time,
        and quasi-DC I-V curves do not reveal package effects.
        
        
        Richard Schumacher
Received on Mon May 8 11:00:18 1995