RE: Problem with clocked driver

> Thanks for your clarification. Just to be clear with your suggestion: are
> you saying that I should connect at the output a transient source with a
> rather slow ramp that changes the voltage (over time) in the range
> required by IBIS (e.g. -Vdd to 2Vdd for pulldown) and measure output
> voltage and current at several time points?
 
Yes, that's the idea. If you (or s2ibis) would have connected a DC voltage
source to the output, and swept it using the .DC statement, instead you
should set the voltage source to generate a PULSE or PWL waveform. Then
measure its current into the driver's output pin. Once you extract I(t) and
corresponding V(t) points, ignore t.

> The other question: when should I start taking the data?
 
Toggle the clock to get the flip/flop into the state you want it, and wait a
few nanoseconds for it to stabilize. Then start ramping the voltage source.
You just want to wait long enough after the clock for all transient effects
to die out, and then ramp it slowly enough that it can almost be considered
steady-state, compared to the speeds at which the device usually switches.
If in doubt, try doubling or halving the ramp rate and see if it changes the
resulting I-V characteristics.

I am not sure if pulling the output pin to -Vdd (especially during SPICE's
initial transient solution phase) might cause any problems. If so, start
with the voltage source set to something more benign (like 0V), then ramp it
to -Vdd, then up to 2Vdd, for example.

Andy
 
Received on Fri Sep 1 09:43:31 2000