The game arrived at the show OK and was initially running OK after setup.
On Friday shortly before the show opened I noticed that the game had lost all 3D vectors. Checking the self-test flagged matrix errors with all matrix tests returning 0000. The spare board set was swapped in to get it running again.
At some point on Saturday the game went down again with no picture and spot killer on. The show medics reported that the game was playing blind and a replacement game board set had not brought the picture back. A bad deflection board was suspected. With no spare deflection board to swap in the game was withdrawn to the back room and wrapped for return.
Since I didn't have a means to run the deflection PCB on the bench I started out with static measurements. After confirming all the fuses were good I moved on to an in-circuit back-to-back diode check of all the XY deflection transistors, also cross checked between X & Y. No anomalies were found on the MPS-U07/U57 or 2N3716/2N3792. Moving back to the small signal input transistors I found a difference between Q2 & Q12 (2N3904) where the C-to-B forward voltage was 0.69 on Q2 and 1.20 on Q12. Removing & testing both found Q12 Hfe=136 and Q2 Hfe=183. There can be a wide variation in Hfe and any reasonable Hfe value at all indicates "basically working" but I still replaced both with a new batch-matched pair.
Moving on to the spot killer circuit and transistors Q8, Q9, Q10 found a back-to-back diode fail on Q10 (2N3904). Removing Q10 and testing found Hfe=0 confirming it was bad. After replacing it the board was ready for re-test back in the cabinet.
Setting up the main PCB on the bench with the Arduino ICT and running the basic matrix tests hit an "E: MX 5555 0000" at "MX Test 17". Tests 18,19,20 also returned 0000 consistent with the in-game self-test results. With matrix test 17 set to repeat, inspecting IC 8B (LS385) found pin 12 output active, pin 15 input active (0x5555) and pin 14 input idle low (0x0000). Checking IC 6C (LS384) found pin 6 always low with inputs showing 0x5555. Pin 7 (CLK) was active as was Pin 1 (CLR) and pin 17 (Y). I suspected IC 6C was bad and replacing it fixed all the matrix tests.
With the deflection PCB back in the cabinet there was still no picture and spot killer on. After a few minutes it became clear resistor R6 was burning hot and TO-3 transistor Q7 was also hot. Probing the XY input transistors with a pocket scope found:
Pulling and Hfe testing Q1 & Q3 (2N3904) found them both OK. Pulling Q4 (MPS-U57) and Hfe testing was also OK. With those out, back-to-back diode testing of Q7 (2N3792) found a suspicious B-to-E diode voltage of 1.616V. Pulling Q7 and back-to-back testing out of circuit still found the unexpected B-to-E voltage and the transistor was declared bad.
I didn't have any spare 2N3792 but decades ago prior to wide scale internet commerce I'd bought a couple of sets of compatible replacement pairs of MJE15003 & MJE15004 that were available from the local electronics parts supplier (Farnell in the UK). From the data sheet they were superior in almost all parameters to the original 2N3xxx. These were fitted as a matched pair (leaving a suspected good 2N3716 spare for the remaining axis). The board was ready for re-test back in the cabinet again.
Back in the cabinet there was still no picture, spot killer on and R6 & Q7 getting hot. To allow the circuit to be investigated without burning up R6 & Q7 again I opened the fuses F1 and F2 to disconnect the yoke drive. Looking on the scope the input to Q1 (2N3904) was bad with the actual PCB output good. Checking R3 (390R) found voltage missing on both sides compared with the other axis. Checking R8 (3.3R) found voltage missing on one side and a closer examination found a small hole burned in the side of it. Powering off and checking with the meter confirmed that resistor R8 was open. Replacing R8 with a reclaimed one from a spare deflection PCB and powering on resulted in a flash from capacitor C1 (0.1uF). Replacing C1 and powering on with F1 & F2 in, still there was no picture, spot killer on and R6 & Q7 getting hot again. Metering R8 found it open again, presumably blown by C1 shorting out. Replacing R8 again and powering on this time found all the voltages and signals matching between X & Y axis. With F1 & F2 back in and powering on, the spot killer was off, R6 & Q7 were normal temperature and there was finally a picture :)
The picture looked good and the game ran for a few hours with no issue, also confirming that the matrix repair had completely fixed the game PCB. I couldn't see any obvious problem related to the MJE1500x substitution - the picture was sharp & stable.