Wow! Thanks, Modemhead for those pics. You must have done that before, as you posted everything I needed to interpret them, the baselines, the settings, the input signal, etc. I'd always wanted a Fluke, and while reading on the web I came across your posts about converting an 8050a to LEDs. It was that info that caused me to buy the (for parts) 8050a I found on EBay at a price I could afford.
So back to the repair. The 8050a manual says the input of the U32 board goes to an absolute value circuit, and studying those circuits from the 1980s, many use a pair of diodes (A7- which my notes label "D1" and "D2" for convenience) between a pair of op amps (almost certainly LF351N, "U1") and possibly CA3150E ("U2") with multiple precision resistors- often laser trimmed (the thick film U32 board seems perfect for this).
I set my scope to match yours, except I used 60 Hz input and 5ms/div. The same baselines, the same 1V RMS input, the same .5V/div on the upper trace and 1V/div on the lower trace . My T1, T2 and T4 are the same as yours, so I won't post them, but my T3 differs. The signal between the positive going half wave is not zero, it's got some weaker positive signal there.
I then traced to the CA3150E ("U2") and the inverting input pin 2 shows something similar - close to full wave rectification, but not symmetrical - see below with 0.2V/div on the upper trace. The output (pin 6) shows pulses at the zero crossings - see below with 2V/div on the upper trace, still DC coupled. (I've got some better traces AC coupled and showing the pulses more clearly, but you can see the basic signal below - alternating size positive going pulses from a negative DC baseline.) The offset null (pin 1) of that chip is near -5V and comes from the trimmer pot on the U32 board. The NI input (pin 3) is at ground.
Any ideas?