I have done it. I don't recall the exact pins, but you must jumper them
together. IIRC they are the negative supply pins for the output stages.
You also have to remove the two "bootstrap" caps.
This is all easily identifiable if you look at the datsheets. The big
problem is the case size. The 391 is wider. Thus you can only use it
when the ICs are clamped to the heatsink, although one time in a Zenith
(I was really in a pinch) I went ahead and changed it over to the newer
heatsink from a junk board.
It is still easier to just use the right part, but as we all know, ICs
get discontinued. Then you have to buy them aftermarket and you can't
really be sure you're getting good ones. At that point it'll be better
to change over to the 391 rather than changing 4274s trying to find a
good one.
I just looked, you remove the caps between 5-6 & 11-12 and install a
jumper between 4-13. Simpler than many factory mods, and quicker than
changing an IC three or four times or troubleshooting a waveform
generator that is working perfectly.
Incidentally, in the past I would use standard OP AMP protocols to
determine if the STK is bad. That means either a null point at the - in
or the same waveform at the + and - in. This is not always conclusive I
have found. I've had the indication that the channel was bad when it
was in fact working perfectly. Apparently in some sets they let the
output clip, but not bad enough to throw the convergernce off.
JURB
