Just to report my experience with trying to reduce DC offset on an SA-400 I picked up at an estate sale. It was in excellent condition, just some scratchy pots/switches which cleaned up well with Deoxit. Offsets measured 60/55 mV on L/R channels so I thought I'd try to lower that by paralleling resistors on R613 and R614 as Elnado suggested. I was able to get it down to 0/8 mV by adding 560 ohms in parallel. The original resistors in the SA400 are 330 ohms (from schematic and measured), so the parallel combo gave about 210 ohms resistance. It seemed a nice improvement and I thought I was done.
But then I started noticing the output would start to cut in/out on one or both channels especially at very low volume setting (within 4 clicks of the bottom), sort of like a dirty contact. This often happened immediately after switching on, but also could be triggered by toggling low and high filter switches and sometimes loudness. Switching them would cause the level to randomly cut in or out with each switch. Wiggling the volume control to higher level would clear it up, then toggling the switches would trigger it again. Also powering off/on would cause both channels to stay quiet after relays clicked in, until the volume was wiggled. Or if I waited several seconds, one channel volume would gradually fade in. Changing Balance did would not have any affect.
Of course I figured this was due to still dirty switches or volume pot (although it didn't really make sense that wiggling the volume would fix dirty filter switches) so I Deox'd the bejubus out of them, even removed the volume pot from the board and and really doused it. None of this fixed it. I traced the signal thru and found there was no signal loss actually occuring until after the volume control, somewhere around the first PA diff transistors. Poking around that area (Q601/2, R613/614, etc) with a chop stick would cause the signal to drop in/out as it had with the filter controls. Of course this is also where the extra resistors were added. I removed the resistors and voila, no more dropouts. I assume the lowered resistance in the current mirror leg was causing a larger current flow, and making the diff pair and/or other components sensitive to "signal shock" from switching the filters, etc. At least that's my theory. Possibly the Tr's or R's are fatiguing after 45 years, but I'm tired of chasing the bug so didn't investigate further.
So hopefully this helps someone else to avoid several hours of hair pulling in a similar situation. If you add the Offset compensation resistors and begin having audio level dropouts, remove them and see if that helps.
Cheers,
Tom
