I'm assuming you fed the regulator ignition feed terminal directly from the battery for your test? Is this what you did and the problem was solved? If so, the circuit starting at the battery, going thru the fire wall, ignition switch, firewall again to the regulator has an issue. The ground side has to be good as well. This includes the regulator, alternator, and battery all seeing the same ground. But since the problem was solved without changing any grounds, they should be OK. The regulator uses these two circuits to determine field current. Field current adjusts alternator output. When you have what we refer to as voltage offset is where the issue can occur. All the points mentioned need to see very close to the same reading. All grounds the same. reading. All ignition circuits the same reading. Both within a couple of .10s under load. I solved a similar mystery while working at the FCA electrical lab a few years ago. Though the system was a little more complex. But the problem was ground offset. The PCM ground was good enough for the vehicle to run. But poor enough that the PCM thought the sense voltage being sent from the alternator was varying at the PCM itself. This caused the PCM to vary the field current output. Turned out heat shrink glue had wept between the ground terminal and the wire itself. This car had been at the dealer many times before we got it at CTC. It had a new PCM , alternator, wiring between the PCM and alternator. Nobody ever checked for offset.
Doug