Help me turn my parts board Super Kaneko Nova into a working board: Graphics issues + "Check B NG"

[adgenet]

I bought a Kaneko Nova motherboard a while back as a future donor for spare parts in case my good board fails, and it was advertised as having graphical glitches.
Of course, it arrives and it does actually have graphical glitches, just as described:

Of course, not being content with just leaving it be as I intended, I started poking around.

I noticed one of the QFP208 chips, SPCII-A, had a few suspicious looking legs (discolored and slightly bent legs, and some corrosion peeking out from underneath) so I pulled the chip off to inspect the pads further, and the traces underneath.
It turns out one of the pins was bad on the chip (broken off at the base), but it was a redundant ground pin. There was no trace damage anywhere underneath the chip, and the corrosion was just from the minor remnants of the broken pin - Everything cleaned up nicely.
Reattaching the chip and repairing the damaged pin caused no change to the graphical glitches.

Further inspection around the board reveals no other obvious physical problems.
I tried swapping the BIOS, and the Lattice chip next to it from my working board with no change in behavior.

The bad board on the bootup "check" screen shows "Check B: NG".
I had always assumed this boot check was to check the cartridge ROMS, but it seems like that may not be the case, as the same cartridge obviously boots with "Check B: OK" on my working board.
I wonder what is actually being checked? Can I figure this out somehow from the MAME driver source?

I have included links to video captures from the bad boards, and a comparison with my working board. Ignore the timer and backup errors as they are simply due to having no clock battery installed.
You will see in one of the videos of the bad board (video 3), that there is a red colored weirdness that starts entering the screen. This only begins after a few minutes of operation:

Also strangely, some graphics have no problems at all, as you see with some of the portraits and characters, while the rest of the screen is glitchy as usual:

I'm obviously 100% ok with this staying a parts board as I originally intended, but if anyone has ideas on how to fix this, I'd much rather resurrect it.

Videos:

Video 1 - Good Board, Boot - Gameplay: https://streamable.com/8pc1l4
Video 2 - Bad Board, Boot - Gameplay: https://streamable.com/ogkgzf
Video 3 - Bad Board, Red weirdness: https://streamable.com/3hjw0x

Those watching closely will also notice that the boot sequence graphics are NOT glitched in any way...

Anybody know which chips are responsible for what on the Super Kaneko Nova?
There's not really much documented out there.
Also, trivia for those interested at all: The board will boot to the Kaneko Nova BIOS logo even without the SPCII-A chip installed, but it stops on a black screen and doesn't progress past it.

 

[ic3b4ll]

There's a *lot* of RAM chips on that board and they all seem graphics related, i.e. connected to scary huge custom chips. If I had to guess, and after looking at the MAME driver, I'd say the B thingamabob at boot actually refers to the chip marked SPCII-B. Maybe. I'm guessing.
This would therefore point to an issue with it or the RAM connected to it.
How to tell? Your guess is as good as mine but I'd try to find some vias underneath the board that are connected to the relevant RAM and check them with a probe or scope.
Or just remove RAM chips and see what happens. If it gets worse the RAM you pulled should be OK. If it gets worse but similar you found your culprit.
Oh! Check the pin connectors with a probe before anything else.
Maybe I've should have said first... Erm.

 

[adgenet]

Thanks for the reply.
As I said, there isn't much information out there regarding this system.
I figured the problem might be with the RAM, but was unsure of where to start.
I have a logic probe on order but shipping is taking a loooong time so I may just resort to playing with the RAM like you suggest. The soldering isn't a big deal at all.

 

[adgenet]

I had nothing better to do so I went ahead and swapped all of the chips around randomly within the sections highlighted in RED (DRAM) and GREEN (SRAM).
YELLOW check-marked chips have been repaired (SPCII-B) and reflowed (SPCII-A/B + VIEWIII-A). There is no change after doing any of this.
I might play with the other SRAM highlighted in BLUE next to the VIEWIII-A chip later to see if I get any change in behavior. These are the same SRAM chips as those in the GREEN section.

And just for some more photos to add to the thread because why not, here's my chip leg repair I made to SPCII-B.

And here's a shot of the GREEN highlighted area with the SRAM removed:

 

[adgenet]

I went ahead and swapped the SRAM around from the Blue section with the Green. No Change.
The board will also run with only one or two SRAMs left installed in the blue section. It seems that area is responsible for drawing parts of the screen, including the boot screen, as if you remove SRAM chips, you end up with the boot screen missing parts as well as occasionally dropping video sync. Removing them all will finally kill video output entirely.

Also, at least the test game I used, Jan Jan Paradise, can run without the 28 Pin DIP RAM next to the Green Section installed without any changes in behavior.
I suppose it could be the culprit causing all of the issues if it can run without it with no change if it's entirely dead to begin with, but I'm not about to remove one from my working board, and I couldn't find an easy replacement available to order online.

So my conclusion is: possibly bad DIP package RAM, or a bad custom (most likely SPCII-A or B?)

 

[ic3b4ll]

I've been paying attention to your posts and I agree with your analysis. No change in behaviour would imply that the DIP RAM chip is the culprit or, indeed, that there's a bad custom.
The DIP is just a common SRAM chip. It's either a 6164 or 61256. Cypress has weird numbering scheme. If it's xx185-yy it's a 6164, if it's xx199-yy it's a 61256. You can replace it with any compatible part but make sure it's fast enough.
Congrats on your soldering!

 

[adgenet]

It's a Cypress CY7C425-20PC. My other board has a Sharp LH540202 instead.
Sharp's datasheet says it's compatible with Am/IDT/MS7202.
Time to do some searching I guess.

 

[ic3b4ll]

Oops, my bad. Really looked like a common SRAM chip.