Found it! It was the RAM chip at 10M. The chip itself was fine, but the trace to pin 16 was intermittent. I got the board with a socket already at that location, and I'm guessing that when the chip was pulled out the board got damaged. I also found that flexing the board would get it to work briefly.

Got a board in that when it booted up the attract screen was messed up. Colors of characters aren't quite right Flexing the board around the area of 1A cleared up the problem. Looking at the back of the board I see someone tried to reflow the solder around a bunch of the sockets. I'm going to start ripping them up one at a time and replacing them with new clean ones. Update: Replacing the socket at 2A fixed it.

Haven't yet been able to figure this one out. The player stats on the side have lots of funky colors. The messages in the game are blue instead of white. Also the high score table has all the wrong colors for the text. When I put the game into self-test the screen turns red instead of white, and a RAM error is displayed but I can't read it because the text is black on a black background. Once I skip past the RAM error, all the self-tests come up fine, including the correct colors on the...

Hooked up the board to my test bench and found the motion objects were the wrong color and displayed lots of sparklies and other glitches. Replaced three of the 2149 chips at 12WX and the problem went away.