Your post doesn't say if +20V at pin 5 of the transformer is checked but WF5 on your web page looks OK so +20V seems to be fine. Still worth checking.
Question though. You mention that CR769 doesn't clamp if the voltage never
goes below 10V. I understand. But the voltage at R768 (on the side of C768)
definitely goes below 10V. So something on the other side of R768 (which is
the nexus of CR769, CR776 and R777) prevents that signal from going below 60V,
correct?
That is also how I see it. However, looking at the waveforms I don't think it is something hard clamping to 60V, more like there is an undesired load (pulling above 60v) that is using up the charge coming from C768 before cathode of CR769 has a chance to go below 10V.
Now that you removed the tube, you could try lifting R777 to figure out if the problem is to the right side of left side of R777.
Could this be related to the voltage shift at C768? C768 is pure AC centered
around 0 when R768 is lifted but biased to 100V when R768 is connected?
This is expected. When R768 is lifted only DC path from right side of C768 is your probe, which has a DC path to ground. So average value will be at ground (0V). When R768 is connected your clamps (and undesired leakage) are providing DC path. Let's look at right side of R768: Its maximum is limited by top clamp at 110V (from your measurement), bottom is clamped at ~65V. The waveform is not symmetric so the average is more that (110V + 65V)/2=87V. You can ask the scope to give you an average value but ~ 100V is probably reasonable (more time is spent at 110V). Since this is the average (DC) value at the right side of R768, left side is at the same DC voltage because C768 doesn't pass DC current.
Most likely one of the diodes CR776, CR782, CR780 went bad or leaky.
Ozan
