:-) Yes, I would expect the IP3 and IP2 to be measured with signals in
the -10dBm to 0dBm range. AFAIK, testing IP3 and IP2 is pretty much
always done at levels way below the intercept points. It would be a
pretty non-linear amplifier that had IP3 and IP2 below its maximum
output before clipping. With the performance I'm looking for, it means
detecting distortion products down more than 100dB from the inputs, but
that's something I'm up for. I do need to work on a little better
combiner for my sources, though, so they don't intermodulate before
going into the D.U.T.. I can get fairly close to the sort of
performance I asked about with some modern op amps, but not across the
full HF band.

Cheers,
Tom