The game arrived at the show OK and was working fine during setup on Thursday.
On Friday morning the game was still working OK but shortly before the show opened it died. Initial inspection found the +5V missing, traced back to a blown 20A fuse F3 on the power brick. Replacing fuse F3 with 25A (apparently an Atari service bulletin upgraded F3 from 20A to 25A) brought the game back to life, but it was clear the fuse holder and power connectors were burned. One of the show medics had a new fuse block and the tools to replace the connector so we were able to properly fix it before the show opened.
The game ran OK for the duration of the show :)
The monitor had always had a flare in the center of the screen after the LOPT had been replaced with the Cinelabs reproduction. The Cinelabs LOPT generates 6.3VAC heater voltage by winding the heater supply wire around the ferrite core. As it was set with three coils the heater voltage was a little bit high causing excess emission in the gun. With two coils it was a little bit low and the picture was a bit dim as a result until the cab warmed up. To preserve the CRT during its long operating hours at the show I decided to both splice in a connector to allow the coils to be adjusted and also take off a coil leaving two at a slightly low heater voltage.
With the lower heater voltage the colour balance and intensity needed to be adjusted using the built-in test screen. The adjusted picture looked OK and the centre flare was gone.
I'd also hoped that the reduced heater voltage might improve the high brightness scatter on the green gun but alas it still remained :(
After about an hour the game PCB developed a fault with the vectors that I set aside :(
With the show rapidly approaching it was time to fix the game PCB. On the bench the same vector jitters were present as they were in the cabinet. In test mode 7 beeps indicated RAM 4L (Am9128) was bad. To confirm the sockets I swapped IC 4L with IC 4M and ran test mode again that rang out 8 beeps confirming the RAM was bad. Replacing 4L cleared the self-test however the vector scale test still showed a problem.
Checking the B1 to B8 inputs to IC 6D (DAC-08) found them all active except B1. Looking carefully again found that B1 was active only at a particular point in the scale test. The DVY inputs to IC 5E (LS273) were also all active. IC 6L (LS157) pins 7,2,15 were all active. I couldn't find any obvious missing signals and began to suspect that the DAC-08 was bad. Grounding the data inputs on the DAC had no significant effect on the scale test. Replacing the DAC-08 with the equivalent NEC uPC624 fixed the scale test and the picture looked good in game mode.
Fitting the second board set into the cabinet found it also had vector problems and no star field. The self-test flagged divider errors and the board was removed for repair :(
On the bench the PCB displayed the same divider errors in test mode. Setting up for the Arduino ICT and running through the ICT equivalent divider test 21 flagged "E: DV 4000 7FFF". Setting the test to repeat the test for 20s and ignore errors I probed with the scope. IC 5N (LS83) pin 14 (C4) was always high. Using the ICT signal capture I verified the ~STP and ~REN signals were operating. I captured a few signatures from IC 8P (LS109) pins 9 & 10, the 3MHz clocks, that all seemed OK. Checking the adder inputs on ICs 5N,4N,6N,6M found them to match the tests 0x4000. I'd already replaced IC 7M (LS299) in a prior repair so I speculatively tried a piggy back of IC 3N (LS299) that paid off - with the piggy back the test passed and replacing IC 3N fixed the divider test and the game played OK on the bench. With the game working I captured a few more signatures around the divider circuit for future use.
Back in the cabinet the repaired game PCB ran OK for several hours :)
I'd already noted from prior work on the cabinet that the ARII +5V was set to it's maximum and the voltage at the game PCB was only slightly above 5V. Whilst still within the good range there was a risk at some point that the power pins would burn out. In preparation for such an event to occur at an inconvenient time (i.e. during the show) I decided to prepare a new set of power cables comprising new wiring, brick plug pins and edge connector pins. To fit it would need only a jewelers screwdriver to pop out the existing pins from the connector housings.