We have a word for problems like this one, but I'll refrain from using
it in public.

On a high brigthness signal it jitters about an inch back and forth. On
a test pattern there is a slight wave in it, but it is hard to see on
normal video. That is so slight that it almost could be left alone, but
it's surely related to the jitter.

I changed C 338 which is the low freq portion of the AFC filter
network. Also changed IC 301 to no avail. I also bypassed C333 which
didn't do anything.

I see pin 38 is also fed by Q 314. It seems to get an inverted ABL
signal there. To eliminate that as a suspect I disabled it temporarily
by shorting Q 314 E to B. Still doing it.

Looking further, with the spec sheet for the CXA2025AS I find pin 37 is
called L2LFIN. The spec sheet says to bypass it and that you can use DC
to shift the H sync phase. I have also bypassed that with a .01, which
is plenty since it is fed through a 470K resistor (R 375).

There is also something not quite right abut the picture. It looks like
the video is slightly clipped on the whites. Naturally I checked the
signal at the HSIN pin 44. From the sync tips to the top of a staircase
waveform it is exactly 2.0 V P-P, as called for in the print.

Anything you might have on this would be greatly appreciated.

JURB

Problem solved.

Turns out it was a two pronged problem. I had already bridged C333, a
..01 uf. It had an effect on the symptom but I dismissed it because it
did not fix the problem. I went back to the spec sheet om the CXA2025AS
and found out I had a misconception.

Pin 36 is marked AFCPIN. Just on the sony schematic it would seem to be
something to do with pincushion, but no. It's AFC pulse in. This is
actual feedback from the horizontal, which is what I was looking for at
that point. The print calls for 4.8 V P-P and it had about 6, this was
traced to an open D527, a 5.1 zener. Replacing that affected the
geometry and convergence so I guess the set was aligned in this state.
Thing is it didn't fix it !.

So with the print in front of me AND the PDF on the screen of the
jungle I looked for ANYTHING the hell else that could cause this. It
was damsure that it had something to do with the jungle because the OSD
was stable. Well almost damsure.

Incedentally I had to initially build a zener as we were out 5.1s, and
I can tell you from experience that if the pulse amplitude goes down to
about 4.4 V you lose all OSD. A symcure for your database that never
really happened.

So actually it was both the open zener and the .01 uf. The cap is very
close to the jungle IC and might have got hit when the IC got changed.

This zener, one other thing about it, if it goes bad, that pulse
amplitude changes of course, but when that happens it also affects
geometry and convergence.

JURB