Thanks guys for the comments -- I dug into the manual, and all of the AC gains are set by laser-trimmed resistors in ?think-film? packs which HP called "fineline" resistors -- the AC tweaks appear to adjust the low-end of range linearity by adjusting amp offsets. The RMS converter is a log amp, an averager, and an anti-log amp. This circuit actually works very well. I'm going to open the 3456 up and check test points 4 and 8 and see where the gain differences actually are. I may have to actually measure the resistances of the packaged resistors and see how they compare to one another -- I may be able to get better full-scale accuracy by adjusting just one resistor... Those fineline resistors are supposed to be 0.01% or better -- which means that HP never expected the 3456 to be really good at AC. Yet it is a very excellent 6-1/2 digit meter for DC and ohms, holding very close readings when compared to my 3458A, so the AC performance is a little disappointing.