Re: Simulating Pulldown and pullup curves.

Andy Ingraham wrote:
>
> Eitan Medina wrote:
>
> > I have an I/O pad with clamping-high and clamping-low circuits.
> > When simulating the whole pad-circuit I get non-monotonic data (My
> > checker wanrns about this).
> > Should I simulate the pullup-curve on just the pullup-circuitry (eliminating
> > the clamping and other structures that exist in the pad-circuit) in order to
> > get monotonic curves ?
>
> IBIS does not require monotonicity.
>
> If your device is non-monotonic (if the total current is
> non-monotonic), there is nothing you can do to the data to make it
> seem monotonic. Some simulators will have problems simulating with
> your device, others will not. You cannot help that, except by
> redesigning your IC.
>
> When separating the total device data into pullup/pulldown/clamp
> curves for IBIS, even though the total current may be monotonic, some
> of the resulting curves by themselves may become non-monotonic. This
> is unimportant, because the simulator will sum the currents and the
> non-monotonicity should go away. The IBIS checker complains because
> it doesn't try to sum the currents back together. It is just being
> extra cautious.
>
> I would not bother trying to make the separate curves monotonic just
> to eliminate the warning message; especially if doing so results in
> a less-than-correct representation of your device. If you simulate
> the currents separately, they may not add up to the total current
> when you recombine them.

As a point of reference, some of our drivers have
non-monotonic I/V curves due to circuitry which
ensures that the outputs remain high-impedance when
the power is OFF; the weak devices used for this also
cause some loss of gate drive when the device drives
into a heavy load.

-- 
D. C. Sessions
dc.sessions@tempe.vlsi.com