If you test the output for a microsecond, or ten microseconds, it probably won't damage the part as much as loading it in the same way for 1 second or more, isn't that right? I haven't checked, but there is a maximum duration on short-circuit to GND or Vdd, isn't there?

What concerned me was that it appeared that you set the output to a zero and then measured the behavior with respect to Ioh, then measured it with respect to Iol with the output set to a one. The P and N channel devices in the totem pole are probably pretty similar, but it seems as though you're testing the wrong ones.

Don't get me wrong here ... I'm not trying to beat you up, I'm just trying to understand the characterization methodology, because it may help me to understand the spec's better. My undestanding heretofore has been that if the datasheet said that the device would supply 2.7 volts at 2.5 mA, it would happily do that forever, from each and every output, up to the total package power dissipation limit. I've never assumed that one should attempt to drive a load consuming more than the rated current. In the process of characterizing an output's capacity, it's probably risky to do it too slowly. I'd have though that you'd set up the test such that it tests it, say, 100 times for a 10 microsecond interval, with, say, 100 microseconds for the device to cool. That way you get a picture of what it does as it warms up.

Now, let me ask you this ... at what timing and with what loads would YOU use the Intersil part that you tested, if your paycheck depended on it?

RE