JVC DT-V1910CG /1710CG Unstable Image (Wobble)

[fernan1234]

after testing i can see that the mod (replacing the amp for h-sync with a clamp) eliminated the wobble at 480p but made it slightly worse in 240p.

So you had wobble in 480p before the mod? From what source? The only 480p source I recall using extensively on these monitors is a Dreamcast's RGBHV, which always looked 100% solid. I also used 480p RGBS from an N64Advanced board, which also looked perfect though it was using the custom D-Sub Retro-Access cables that solved wobbling in all scan rates for me (and that one in particular was also wired for c-sync and reaching the monitor with TTL, another factor worth exploring IMO).

And did you have a chance to also test 480i (or maybe you meant to say 480i rather than 480p)? Like I mentioned before, using an external RGBS (sync on composite) -> RGBHV sync separator made 240p 100% rock solid, but introduced slight but noticeable wobble for 480i, whereas the original RGBS input was perfect on 480i but had terrible wobble with 240p. Maybe this can be a hint for figuring out this puzzle.

Another hint may be to find out the sync cleaners that Retro-Access uses (or used to use) in their D-Sub cables. I don't know if they alone fixed the problem, or if the Extron matrix that I hooked them up to was needed for raising video level sync to TTL as well, but with that setup I had no wobble on any scan rate. It'll be another month or two until I have access to those cables again, but once I do I'll try to see if I can open the hood on one of them to check.

[famiac]

i tested 480p with a fujitsu fm-towns and with a nintendo gamecube.
I tested 240p with an NESRGB-modded famicom and with a sega master system. the NESRGB does not have the "de-jitter" feature. -- these are both problematic sources. i should try with an N64.

retro-access uses the LM1881. I've compared the LM1881, the EL1883, and ISL59885. they all output supply voltage level sync (5V in my case) and the ISL performs best. I have one in the cable i'm using to split RGBS into RGBHV, and that cable goes to an extron 203rxi. the extron boxes output TTL-level sync.

I've tried foregoing the extron but it seems not to make a difference.

i haven't yet tested 480i

[famiac]

i can now recommend the signal board modifcations -- with certainty -- to all DT-V owners.

Two problems with the tests i did earlier:
1- the famicom and master system both output flawed sync signals that don't play nice with the DT-V. the NES sync jitter is widely documented, and the master system sync signal doesn't have any serration pulses.
2- my signals were going through an extron 203rxi, which adds noise/jitter of its own on the sync outputs, especially when DDSP is enabled.

after testing with a directly connected N64 outputting 240p/480i and comparing side-by-side with an unmodified DT-V, i can say that the jitter problem goes from being highly distracting to almost unnoticeable. I also did the same comparisons with a gamecube at 480p and the modified DT-V was a clear winner.

improvements to the circuit:
by adding a BAT42 Schottky diode in series with one of the signal diodes, you can further improve performance by increasing the clamp voltage. It seems the M52346SP behaves at its best with input voltage around 1.5Vpp, which is consistent with what the datasheet says.
I also changed the 4.7uF nonpolar capacitor from a ceramic type to a 0.1uF film type.

perhaps the jitter problem hasn't been eliminated completely, but the sync amp bypass + clamp restoration on the signal board definitely brings us a step closer to DT-V perfection. it's good enough for me, at least.

EDIT: after tracing the sync line all the way back to the RGB cards, it's clear that JVC was trying to err on the "safe" side by putting in oversized coupling capacitors.
these caps are all charged by the sync signal, and because they're oversized they take too long to reach steady-state. swapping them with 0.1uF caps should much improve the "wobble."
the caps in question are C103, C106, C107, and C108 on the "SIGNAL PWB" and C125, C124, and C126 on the tall "MOTHER PWB".
you may wish to swap the 10uF input cap on your RGB input card to a 0.1uF cap as well.