To me this kind of wonky oscillatory behavior would make me want to look for defective and/or overheated op-amp (or some component in the feedback portion of the op-amp circuit) or an open somewhere in the path.
Here's the thing I notice about your pic: the oscillations aren't centered on top of where the lines normally appear (sans-oscillation.) It's like your keystone adjustment dies when oscillations start. (See pic below.) Also, even though your h-scan is obviously messed up, except for the oscillations the horizontal scanning seems perfectly normal. Thus, I'd want to trace the keystone circuit and blame whatever it consists of...It occurred after writing this that it's easy to know if this is the part of the circuit that's giving you problems. (...so here I am inserting this line.) Turn on the PJ, wait for this problem to hit, and try adjusting your keystone setting. Does the keystone still work? (Obviously there will be oscillations, but you should still be able to see obvious keystoning adjustment if it is still working properly.) If so, I'm almost certainly wrong and you can forget this post. If not, your problem is here... read on!
Acording to the service manual, the keystone signal is actually controlled by the VDM. (Page HDM-5). This signal is common to all three tubes, so that feature is consistent.
The keystone and pincushion signals are sourced at DAC U8 and get summed on the VDM using op-amp U701. (00-263330-02P.pdf) One possibility is that this op-amp is defective. Another is that the connection between the OP-AMP and DAC is bad leaving an input to the op-amp floating.
The signal then leaves via pin KEY-PIN on PA13/PC13 pin 13 to the Upper Mother Board and travels to the HDM via pin JC12/JA12 pin 15 (00-260358-02P.pdf)
On the HDM, JC12/JA12 become PC12/PA12 pin 15 and appears as KEY&PIN instead of KEY-PIN.
It is fed through multiplier U18 and summed with other signals using the 4 amps in U17. Again, U17 could be bad or have a floating input.
After that it is summed with H-DRIVE using U6, and as I've already said, your overall H-pattern seems fine, so my guess is the problem is not in U6 or beyond. A situation where a component superimposes an oscillation on top of an otherwise good signal seems unlikely to me, and, as you pointed out, it has to be somewhere in the signal chain that effects all three tubes in the same way. This whole chain ends with the generation of the B- supply, common to all three tubes; after that, I don't think there's anything common between them.
What's more, the multiplier RC4200 could be the cause of the oscillation amplitude decreasing as the scan goes down the screen. (I don't know what B-FDBK looks like, but if it's a ramp from top to bottom and KEY&PIN is oscillating, the output you are getting would make sense.)
If this is the problem, I'd start by cleaning and reseat the VDM harness. (I don't know if a floating KEY-PIN could lead to oscillations, but it's worth a shot.)
If I had swap boards available, I'd swap the VDM to be certain the board wasn't feeding the bad signal via KEY-PIN in addition to the obvious HDM swap.
If I didn't have swap boards available, I'd scope the KEY-PIN output from the VDM as it's accessible without an extension board. Without a scope and with no possibility of swap boards, I'd resolder VDM U8 and U701 and U18 on the HDM. If that didn't do it, I'd replace U701 on the VDM and U18 on the HDM.
Trickier is to check the intermittent connectivity issues between the VDM harness and JC12/JA12 pin 15. You'd have to scope the HDM KEY&PIN line with the PJ on. Not sure how I'd go about this, frankly... probably solder a shielded wire to the line and run it outside the PJ to scope it.
Hmm. Quick calculation. C93 is 47 uF. It looks like there are about 10 lines per oscillation. If your horizontal scanrate is 45 khz (720p), that's an oscillation frequency of about 4.5 khz or about 28k rad/sec. 28000 = sqrt(1/LC)... 1/(28000^2) = LC, if C = 47uf, L = .00003H, or 30uH. That's seems high, like you'd need about 30 meters of wire for the VDM harness to get that. Then again, I did just pull about 4 numbers out of my butt. So...
A pure SWAG would one of the two op-amps is defective. It doesn't seem like the inductance would be high enough for a loose connection to form the right frequency oscillation with a 47uF cap. I don't know for sure. I'm just sloshing numbers around to see if I hit anything interesting.
One way to solve this is to go looking through the relevant circuit sections looking for something that could reasonably form a time constant of about 30k rad per second, but I don't think that's less work than what I've already outlined.
Out of pure curiosity I'd love to know what the source horizontal rate in the picture you took is and how many scan lines per oscillation it is so I can accurately compute the frequency of oscillation. Of course, I'm a geek that way.
ugh OMG I just realized you bumped a two year old thread.