SOLVED + Surprise: SF2 WW with bad front sprites. HELP

[Lions3]

Hey guys, trying to dip my toes back into CPS stuff. It's been years.

Picked up a SF2 WW today with minor graphic glitch. Boards are clean. "A" has a Z80 repair but audio is fine. Also it's a CPS Dash board. Which is odd for a WW board, right? Everything plays fine. All background sprites load fine. It's only the front sprites, like the Intro title, characters, and health bar sprites. The character shadows when they move almost look like a hole into another layer of sprites. As I can see a pattern that doesn't move with the shadow. Does this sounds like a simple fix? or one of the dreaded bad A board chips. At the moment I don't have another A board to test with... hopefully I can talk a friend into bringing something over

Example of bad sprites:




Overall this looks like a pretty clean CPS1

 

[xodaraP]

If you've tested it on a known good A board, the mask ROMs fail on these. A lot. 27C400 are the direct replacement

 

[Lions3]

After some digging it appears to be a bad A board.

Also on a side note, while testing I discovered this was a previously undumped revision of SF2 WW. It's actually the final USA revision that matches the final JP release. Dumped and added to MAME the other week

 

[Hatsune Mike]

Since the sprite addressing and fetching looks good, I suspect it is the sprite framebuffer RAM, which is the big pile of ZIP DRAM ICs in the upper-right corner.

 

[Hammy]

Since the sprite addressing and fetching looks good, I suspect it is the sprite framebuffer RAM, which is the big pile of ZIP DRAM ICs in the upper-right corner.

I think the A= chip controls the input for the framebuffer ram?
For sure X+Y faults are the A chip, and this is the most common error.

ZIP ram does fail and there will be a small % of PCB's with that as the fault

This one is indeed a little interesting but from previous experience trying:

Weak lines from the A= chip to the ZIP ram. Removing the ram connection did not fix the signals...
Shorted out ZIP ram will make other chips die in a domino effect. (seen recently on a vulcan venture with custom toast from faulty ram)

I think the A chip can also disturb other ram data lines (22256 DIP ram below the ZIP ram) and cause strange errors.

I guess the way to find out is to get a working pcb and remove some ram and get a picture of the symptoms of each for reference.

 

[Elfito69]

Since the sprite addressing and fetching looks good, I suspect it is the sprite framebuffer RAM, which is the big pile of ZIP DRAM ICs in the upper-right corner.

12 ram chips are cheaper than a working A board, just saying