I 'think' one of the problems is the O2 inputs to the PCM are kinda 'floating' (I might be wrong), but they are not just a simple GND and signal input. The 'Lo' side of the NBO2 output is not tied directly to GND.
Most of the simulated O2 outputs don't account for that, this can cause problems. I must confess, I've never tried the PLX simulated NBO2 output into a PCM, I have logged it and was impressed how close it tracked to a real NBO2 sensor.

Cheers,
Ross