Psychic5 video distortion. Input needed! };-S

[Elgen]

Hey Mates

I currently have a bootleg Psychic5 on the table. I've made some repairs in the controls/input section and made a JAMMA-adaptor, and it now plays perfectly.

HOWEVER, there are some small, tho rather annoying, distortions in the upper part of the screen.
It looks like small white stripes about ½-1cm long (in the direction of the scanlines) and 1pixel wide. They only apear during gameplay (or gameplay-part of the attract-mode), and not all the time.

I've to take 2 stills of it with my iPhone





I've also shot a small video of it

http://www.youtube.com/watch?v=MOPsOnKEqCM

The board is 2 PCBs with ribbons, and video is generated on the secondary PCB. The fault is definitly on the second board, cause I've got another Psychic5 (maby original?) with identical secondary PCB. And if I swap the PCBs, the error swaps as well.

I've traced the R/G/B and SYNC lines back på this section on the secondary board; actually they go directly to some of the resistors



Another piece of info is, that this game can flip the screen via dip switch, and when I do so, the interference still appear in the same physical place on the screen (top); it doesn't flip along with the picture.

I'm pretty sure, that this interference is induced into the video signal during the very last steps of the video generation, as it looks like affects both back grounds and sprites, but that is just my feeling...could be wrong.

This is really starting to annoy me, so any input would be nice.

Also, should anyone have a scan of the schematics for the game, I would be very glad to get a copy; haven't been able to find it via googling.

[f4brice]

Hello.

Resistors are probably the last RGB digital to analog convert step.
See the two 74LS174 latches close to the resistors ?
They bring 2 x 6 latches = 12 latches and you have 12 resistors...
12 / 3 = 4 bits for each color.

In my opinion, you have a faulty video RAM.
Not palette RAM (that stores color for each palette index), but screen content RAM, that stores indexes to palette RAM.
When you change the "screen flip" dip switch, you see that the glitch doesn't flip.
For me it means the internal way of handling screen flip is managed by changing how the screen RAM is written, and the read process is unchanged.

If ever you have a logic analyzer, you may solve this easily.
On this message, I explain how to user advanced triggering to capture the exact needed scan line.

Regards,

Fabrice

[Elgen]

Thnx for the reply. I now have some more to try. I would not have guessed on a RAM-error myself.

I've tried piggy-backing all the RAMs on the board with known goods, but I think I'll desolder and test them now.

[Elgen]

Okay, a bit of new info. I've desoldered and tested the 3 ROMs on the secondary board, and they all came out good.
Next I think I'll desolder and test the RAM, but when having one of the ROMs out, I discovered that the error displays more clearly, because the background now just displays as solid blocks.







From the above pictures it is obvious, that this is NOT white stripes, but that one line in some blocks sometimes get misplaced "one block to the right" (look at the colours of the stripes).

I think that this could be a problem with addressing that sometimes glitches, but I'm not certain.

Where would you pplz go from here? Still test the RAM, of maby focus on the logic ICs?