What's new

WTF is this CPS2 display output doing

GoldenGlover

GoldenGlover

Champion
Joined
Jul 10, 2016
Messages
671
Reaction score
1,148
Location
New England (USA)
This is a new one for me…see below for a screenshot of a suicided CPS2 B board I bought as untested & with Raz’s suicide tester rom in 03:



IMG_9272.jpeg


All roms on the suicded board are present, seated correctly & where they should be.

So far, I’ve replaced the DL-2027 with a working pull (the old one had three broken legs on it) & have replaced the two 6264 rams. I’ve also reflowed the DL-1525 & have cleaned the other chips to rule out corrosion/bridged traces. No chips are overheating when this stays powered up as shown. I also replaced a busted resistor at location RA8 that also made no difference with output when removed and replaced with a new one.

Whats odd is that the top two pixel rows appear to be missing? I’ve never seen this before…it’s not as if the B board’s connectors aren’t making good contact (they are). Between this & the qsound & object ram errors, I’m truly stumped. There is also no music coming on at bootup with this tester rom.

Any CPS2 gurus have any idea on what I’m missing/where I should look next (considering I don’t have an oscilloscope & this is supposed to be a relaxing hobby for me lol)? TIA for any insight that can be provided…
 
Last edited:
You shouldn't expect the font to look 100% correct when using the suicide tester. While each game is going to have a text font, the palette colors the individual pixels map to aren't the same across games. Ie a game might have all pixels in the font mapped to a single palette color while another game might have a rainbow effect on their font where each row of pixels is a different palette color. The suicide tester tries to do its best to make the text at least readable for most games by setting most of the palette colors to the font color.

I would suggest running the suicide tester in mame for that romset and compare to see if there is an actual fault.
 
This may also give you more specific info on why the obj/qsound ram is bad.

https://www.mvs-scans.com/mad/mad-main.zip

If you drill down into machine/cps2/main/build/ in that zip. If your specific game happens to by listed you can use that otherwise use the generic/suicide-generic.03 rom. If you ended up using the generic version it may have similar font issues as Raz's suicide tester and there maybe other anomalies with the tile/sprite viewers. I think I also disabled beep codes on the generic version since I wouldn't know what sounds to play.

There is more info here about the cps2 version of mad

https://github.com/jwestfall69/mad/tree/main/machine/cps2
 
@ack Here’s the screen I get when I run the AvP encrypted program roms with your suicide generic.03 MAD rom burned @ 03:

IMG_9637.jpeg

This is the only error I’m getting.

Any pointers on what I should triage next?
 
Does it always shows that same error (MARCH) and that address/data?

It feels a bit suspect, the march test is the last test that gets run on the ram. I would expect a crazy actual value like that to cause a failure on the data pattern test.

My notes have the lower object ram chip being at location 4A on B Board.
 
I’ve replaced 4A already with a good 6264 replacement. I’m totally stumped at this point. Yes, it’s the same message every time I’m running it.
 
Object ram is setup for banking, where the lower half of each chip is one bank and the upper half the other. Check pin 2 on each object ram chip and see if anything weird is going on with it. That pin should be what controls which bank is active and comes from that 74LS373 that is right below the ram. Also check the input on that 74LS373 that's associated with whatever pin is connecting to the object ram chips, and also its LE pin. If any of those were floating it could cause the bank to randomly flip and cause problems.

Beyond that it doesn't look fun, schematics says the path from the CPU to object ram is just DL-1525 <=> DL-1827 <=> OBJ RAM

If you don't find anything I can make a modified version that skips the startup memory tests and just dumps you to the menu system. Then you could use the memory viewer to poke around object ram and see if that gives any hints on wtf is wrong.

edit: also what game is this?
 
Hi @ack ,

Thank you for the response & your suggested help. It’s Alien Vs. Predator.

Will try to source a replacement LS373 from my parts boards & see if this might resolve the issue.
 
I would advise checking the signals before just going straight to swapping the LS373. It could be broken trace between the LS373 and obj ram or perhaps a bad input signal going into the LS373, which would mean the issue is upstream from it.
 
You must log in or register to reply here.
Back
Top