Re: AC warning messages

Kim and Mike and All:

The second warning implies to me that there is a major
problem with the model. As Mike points out, the current
appears to be zero. However, this should not be the
case for the AC Waveform test, since the given waveform
spans a realistic range of -1.72 V to -.98 V with the
R_fixture = 50 and V_fixture = -2 V. A correctly
configured ECL output buffer does require a termination
resistor to operate since current is sourced in both
states. This termination is correct and is used by
ibischk3. The default termination when only the [Ramp] is
used for Model_type Output_ECL is R_load to Vcc - 2 V.

We would have to see the model to investigate what the
problem is since there could be a number of causes:

  Bad table or non-monotonic I-V table with bad endpoint or
    repeated voltage points somehow causing ibischk3 not to
    converge to the end-points,
  Wrong Model_type,
  Extra clamp table (or extrapolation) somehow internally
    canceling the external load,
  Voltage Range or internal Reference voltage problem,
  Etc.

A newer release of ibischk3 is planned, but we do not have
an expected date yet. It will include an additional check for
repeated voltages. Also, a pending BUG47 under discussion would
now issue an Error for the second Warning message since
the waveform endpoints are outside the pending 10% range.

I would expect different simulators would probably handle
the model differently. It depends on what the problem is.
Some may produce no waveform or a badly distorted waveform
while others might try to "correct" the problem and produce
a realisticly looking result. Still others may actually function
correctly if the problem is in an I-V table region outside of the
actual simulation range (approximately -.98 V to -1.72 V range).
Such differences might just be a consequence of the details of
the internal DC operating point algorithms.

So the cause of the second Warning needs to be investigated,
understood, and resolved. The first Warning indicates that
there final value of the [Rising Waveform] table is not
consistent with the high state voltage that would be calculated
by the DC I-V tables. It should also be investigated, but
the current deviation is actually about 4-5%. This is still within
the planned Warning range.

Bob Ross
Mentor Graphics

Mike LaBonte wrote:
>
> Kim,
>
> Ibischk3 figures out what voltage the buffer will settle at when
> only the VI curves are used to drive the test loads. That is where
> 0.01V (pulldowm mode) and 2.37V (pullup mode) come from. The voltages
> at the beginnings and ends of the waveforms should match these DC
> levels.
>
> Your second warning implies that the VI curves are delivering no
> current at all. That makes sense to me, because ECL requires a
> pullup, and this test fixture is giving it just a pulldown. Does
> anyone know what the test fixtures should be for ECL?
>
> Mike
>
> Kim Helliwell wrote:
> >
> > When ibischk3 issues a warning like:
> >
> > WARNING - Model 'LVT162245_IO': TYP AC Rising Endpoints ( 0.01V, 2.45V) not within
> > 0.049V (2%) of ( 0.01V, 2.37V) on VI curves for 50 Ohms to 0V
> >
> > where do the expected limits (0.01V, 2.37V) come from? The reason I ask is
> > that I am getting a warning more like:
> >
> > WARNING - Model 'ECLout': TYP AC Rising Endpoints (-1.72V, -0.98V) not within
> > 0.015V (2%) of (-2.00V, -2.00V) on VI curves for 50 Ohms to -2V
> >
> > where the upper and lower limits are the same, and that simply doesn't make sense.
> >
> > Also: how serious a problem is this? Do some or all simulators have problems with
> > this sort of warning?
> >
> > --
> > Kim Helliwell
> > Senior CAE Engineer
> > Acuson Corporation
> > Phone: 650 694 5030 FAX: 650 943 7260
 
Received on Thu Oct 19 10:16:32 2000