Subject: Re: ibischk3 V3.2.8 [Warning]
From: Ross, Bob (bob_ross@mentorg.com)
Date: Mon Feb 04 2002 - 15:55:48 PST
Andy:
I agree with you and everyone else. Both the [Gnd Clamp]
and [Power Clamp] tables should have been extended to cover
the whole range for consistency. In fact, this is necessary
for correct representation when the buffer has internal
terminators that need to be isolated from the [Pullup]
and [Pulldown] tables.
When IBIS was originally formulated for 5 V CMOS technology,
the [Gnd Clamp] and [Power Clamp] tables had negligible
currents in the 0 to 5 V region. These tables were both
truncated and joined at Vcc (5 V). The fact that the currents
were nearly 0 ma AND the slope with respect to voltage
was approximately 0ma/V meant the extrapolation by slope
or by value both worked in EDA tools. Several tools and ibischk3
do extrapolate the end-points, so this is a potential source
of waveform mismatch if the clamp tables do not have zero slope.
Bob
"Ingraham, Andrew" wrote:
>
> I don't disagree that models CAN be made to cover the full range from
> -Vcc to +2*Vcc.
>
> I just don't think that a model should ever be made that doesn't cover
> this range. I can think of no valid reason for any I-V curve to cover a
> smaller range.
>
> I'm not sure precisely what s2ibis* does (since I've not used it), but I
> believe it does just the minimum required by the IBIS spec. Since most
> s2ibis* users probably don't know much about IBIS, this is bound to
> cause problems.
>
> Regards,
> Andy
This archive was generated by hypermail 2b28 : Mon Feb 04 2002 - 16:20:06 PST