Well thanks everyone for the input. Turned out to be a bad op-amp (A4). The unit was an RCA 4136 quad op amp and is used in the shunt amplifier and the summing amplifier that generates the error signal that defines the duty cycle of the SCRs. I was able to get a near mint condition SPM-353B manual off Ebay as the .pdf downloadable from Miller were not scanned at a high enough resolution so as to be readable. What is strange about this board is that it looks like they were modified in production: there is an added capacitor to a ground lug and several op-amps using much lower resistor values than in the schematics. My guess is that they probably had problems with noise transients killing the op-amps and so they reduced the input impedance of the amplifier circuits. Unit works great now, I can go from 4A all the way to full current and the dial stays calibrated within about 10%. Perfect for the kind of work I do.