A debutante at the 2018 show, the game was working OK after setup through to Friday before I left. However by Sunday the game board was no longer booting, only getting as far as displaying corrupted text & graphics. I had a spare game PCB already prepared prior to the show that I swapped in on Sunday before opening to resurrect the game. With the replacement game PCB the game ran through to closing with no further issues.
On the bench the game behaved the same as it did in the cabinet. Inserting the Arduino ICT and running through the system tests found one issue, "E:12H 1802 02 00", indicating that RAM IC 12H (2102) was bad. 12H is one bit of the three bit colour RAM and though it is read-write I've not seen a bad one cause any boot issues. Despite the ICT not flagging any significant issue the game didn't boot with the 2650 CPU installed.
I also found that the CPU socket was very loose (having to press on the ZIF socket to get a good test run) so I decided to change the CPU socket before continuing to eliminate that as a factor.
With the new CPU socket, the tests were running reliably (aside from IC 12H) but the game still didn't boot with the 2650 CPU installed. With a replacement 2650 CPU the game booted and played through without issue. Repeating a couple of times confirmed that the original 2650 CPU appeared to be bad.
Moving on to the colour RAM 12H, checking output pin 12 found it mostly floating with an occasional high after a "RAM Write All AD" (writing data == address). Replacing IC 12H (2102) fixed the RAM test failure such that all ICT tests were now passing and the game appeared to be running without issue.
Pre-show testing found the game running OK but with almost all of the sounds missing. The Quasar sound PCB has a test mode, activated by DIP switch 1 ON, that cycles through all the sounds. On this PCB there was no audio in test mode. Scoping the 8035 CPU found the oscillator was running and the address & data lines were active. The DAC inputs were all idle high, consistent with the lack of audio. There was also no access to the RAM (6810) since pin 10 (CS0) was idle. Swapping the 8035 with a compatible 8049H brought the sounds back in test mode. Rechecking with the original 8035 confirmed that it was bad.