typ outside min,max

At first I thought this was a bug, that is to say, a
defect in the IBIS models I had. But now that I look
closer, it seems to be more prevalent than I would
have thought possible for a "bug". Maybe it is
a feature. :)

What I have found is that "typ" data is outside the "min"
and "max" values.

Consider this model fragment (greatly reduced for brevity)
provided by a prominent silicon supplier:

[Pulldown]
|Voltage I(typ) I(min) I(max)
| [lines deleted for brevity]
-1.00E+00 -3.22E-02 -1.96E-02 -2.18E-02
-9.00E-01 -2.96E-02 -1.50E-02 -1.91E-02
-8.00E-01 -2.73E-02 -1.12E-02 -1.66E-02
-7.00E-01 -2.50E-02 -8.54E-03 -1.46E-02
-6.00E-01 -2.26E-02 -6.72E-03 -1.28E-02
-5.00E-01 -2.02E-02 -5.41E-03 -1.09E-02
-4.00E-01 -1.77E-02 -4.39E-03 -9.05E-03
-3.00E-01 -1.51E-02 -3.40E-03 -7.20E-03
-2.00E-01 -1.26E-02 -2.42E-03 -5.37E-03
-1.00E-01 -1.01E-02 -1.44E-03 -3.55E-03
0.00E+00 -4.67E-03 -4.60E-04 -1.43E-03
1.00E-01 1.82E-02 4.97E-04 1.33E-02 | typ out of range of min/max
2.00E-01 5.53E-02 1.43E-03 5.37E-02 | typ out of range of min/max
3.00E-01 9.70E-02 3.27E-03 9.92E-02
4.00E-01 1.41E-01 2.23E-02 1.44E-01
5.00E-01 1.85E-01 6.56E-02 1.95E-01
6.00E-01 2.29E-01 9.21E-02 2.50E-01
7.00E-01 2.72E-01 1.03E-01 3.01E-01
8.00E-01 3.12E-01 1.07E-01 3.49E-01
| [lines deleted for brevity]

I expected that min and max represent the extremes of the
device behavior. It seems that in some cases, it is not.
Should this be expected? How is it explained?

Should the model be changed to put the "extreme" value
into the appropriate min or max column?

This does not seem to be a singular case either, I have
observed this problem in more than 30 models from at
least 6 suppliers.

Your comments?

 -- Paul Gregory

   phone: (208) 396-5086 USmail: Hewlett-Packard
     fax: (208) 396-4122 M/S 143
   email: paul_gregory@hp.com 11311 Chinden Blvd.
                                               Boise, ID 83714
 
Received on Fri Nov 14 08:55:11 1997