On Mon, 10 Apr 2000, Shaghayegh Azgomi wrote:
> I have a question regarding IBIS models. When I try
> to use IBIS checker to verify the models IBIS models
> (Hyperlynx simulator) I get warning messages regading
> to 'non-monotonic' nature of some signals.
If you think of the pullups, pulldowns, and clamps as being
like nonlinear resistors, a lot clears up. "Non-monotonic"
means there is a negative resistance region. In a real
circuit, negative resistance regions usually mean the
circuit has stability problems. It will oscillate with
some loads.
In a sumulation, it means the same, but most simulators
don't handle this oscillation nicely. Whether you actually
have a problem depends on how it is used in the circuit.
If the device is loaded heavy enough, the negative
resistance in one device can be swamped out by another,
leaving a net positive resistance or "passive" circuit, in
which case you would not have a problem.
In Ibis models, this often comes up in the clamp region,
where the pullup current is swamped by the power clamp
current. (and likewise for pulldown and ground clamp)
Since both are essentially in parallel, it is impossible to
measure them separately, so the modelers often do an
arbitrary separation. The result is that one has a
negative resistance region that is swamped by the other,
leaving a net positive resistance. In this case, it is
usually not a problem, although it is still bad practice.
It can be a problem with waveform generation, and some
loads.
One point on the v/i tables that is often forgotten ....
The slope of a v/i characteristic is incremental
resistance. If the device is a receiver, this is the load
that the driver and its wire sees. If it is a driver, it
is the driving impedance of the line, and it serves to damp
reflections. This resistance (derivative of the table) is
actually more important than the values directly stated by
the table.
Received on Tue Apr 11 02:18:28 2000
This archive was generated by hypermail 2.1.8 : Fri Jun 03 2011 - 09:53:46 PDT