This game PCB was from a dedicated Astro Wars upright in France. The reported problem was that the game display had a band of white noise across the screen and the owner provided a video of the problem that showed it had missing shells. I suspected that the two issues were one or more faults in the shell circuit.
The owner shipped the PCB to me in a wooden frame to protect it and it was worth it - the game PCB was in pristine condition (I discovered later that the whole machine was in excellent fully original condition also, the best Astro Wars I've ever seen).
Since I didn't want to snap off the top of the original power connector on this board to be able to plug in my old cobbled together Quasar adaptor I added the correct AMP MODU I power connector (I kept the old type of power connector as well for use with Midway Lazarian that uses that style).
On the bench the game reproduced pretty much the same problem as the owner reported. Looking in detail confirmed that the white band of snow was infact a white band of random shells across the screen.
The first step was to test out the board with the Arduino ICT to eliminate bad shell RAM as a cause. The ICT passed all the RAM on the board thus the problem with the shells had to be after the shell RAM in the display generation circuit.
The shell circuit is very simple - the line counter feeds the address of the shell RAM and the shell RAM data is the row position to display the shell. The row data is latched into a counter that counts up from that value on each row clock. When that counter overflows the shell is displayed and the counter stopped until the next reload at the start of the next line.
Using the ICT to write the shell RAM all low and poking around with a scope confirmed that the ~SHELL signal looked crappy with bursts of activity. IC 4G (LS161) pins 15 & 11 looked incorrect, as did IC 5G (LS161) pins 15 & 11. The inputs to these counters from the shell RAM data looked OK. Writing the shell RAM all high or all low didn't change the ~SHELL signal all that much. Breaking out the comparator showed IC 4G as OK but I couldn't attach the comparator to IC 5G due to resistor R74 being too close and fouling the comparator IC clip :(
Further poking around with the scope found the outputs of 4G all unstable. 5G outputs also all looked unstable except pin 15 that looked OK. The breakthrough came looking at the 5G outputs QA to QD that didn't look like counter outputs (they all looked the same). I suspected IC5 5G (LS161) was bad and replacing it fixed the game :)
The ICT flagged all the ROMS as bad against all of the known sets and I noticed that
the ROMs on this board were all labelled '02', something that I'd not seen before. I
dumped the ROMs and found them to be a very different revision from any sets previously
known that were all minor variations on the same revisions. I dumped this set for the
Zaccaria Astro Wars '02' ROM set (10.9KB).
I kept a picture of the back of the board that had a few straps on it for future reference in the unlikely case there were hardware modifications that went with the ROM revision.
A few hours of burn-in testing on the bench and a few more hours in my Astro Wars cabinet didn't encounter any issues - the board was running fine and was returned to it's owner.