Following on from the repair log @qjuk did for Namco Pacman, I've recently been working on its older sibling, a Namco Galaxian so thought I'd do a repair log for it. I've had this pcb sat in the loft for years but thought it's about time I had a look at it, and It ended up being a bit of a basket-case.
I know my way round a Galaxian board quite well so I've also done some commentary on how bits of the circuit work in case in case it helps someone in the future.
Here's the board, it's had some work done on it previously and also been converted to DC. Powered it up and got the following static screen
There was activity on the CPU address and data bus but the watchdog was barking constantly. Time to dig out the test board (this allows memory tests to be done on various boards, but also has status LEDs to display the results of tests so it is still useful if you're getting rubbish displayed on screen)
Powered up again and same as before, watchdog still barking, which is odd because the test board directly accesses the CPU address and data buses, and there's nothing much on the main board that will make it watchdog. Looking at the schematic the CPU connects to the shared address bus through a couple of 74367 buffers at 7D & 7E, and the watchdog reset is generated by a 74138 decoder at 8N so one of these is my guess.
Checking the inputs and outputs of 7D found 2 outputs stuck low (despite the inputs toggling) so I swapped it out, ran the test board again and this time it ran, but indicated faulty work RAM. I was seeing RAM chip select and write-enable signals working so it suggested the RAM itself was faulty, so rather than mess about I just swapped both 2114's at 7N & 7P.
The test board now shows work RAM as good, but also object RAM (the 2 x 2101's at 4F & 5F) as good, which is positive news, but then it fails on the video RAM test. These had already been swapped by the previous repairer, note the state of the turned pin sockets below
I swapped both 2114's at 3F & 3H (into the existing sockets, with the intention of replacing the sockets at a later date). Re-ran the test board and this time all memory tests pass (but still the same mess on the screen)
As the object RAM is working, that proves the circuit as far as the Horiz. Pos. Bus so the next bit to look at is how the video RAM is addressed for displaying on the screen.
There are 2 74283 adders at 4N & 5N that generate the screen position signals. Probing 4N showed 2 of the outputs stuck low (despite the relevant inputs pulsing) so swapped it out and now we get something semi-sensible on the screen. Unfortunately it's showing only half a character repeated, instead of a full character. Moving to the next chip, a 74273 latch at 2M had output pin 6 feeding E2 stuck low meaning the lowest 4 bits of each character would get repeated. 2M was swapped and finally we get proper text being displayed during the tests (but with a solid yellow bar across the screen)
More repairs and commentary in the next part...
I know my way round a Galaxian board quite well so I've also done some commentary on how bits of the circuit work in case in case it helps someone in the future.
Here's the board, it's had some work done on it previously and also been converted to DC. Powered it up and got the following static screen
There was activity on the CPU address and data bus but the watchdog was barking constantly. Time to dig out the test board (this allows memory tests to be done on various boards, but also has status LEDs to display the results of tests so it is still useful if you're getting rubbish displayed on screen)
Powered up again and same as before, watchdog still barking, which is odd because the test board directly accesses the CPU address and data buses, and there's nothing much on the main board that will make it watchdog. Looking at the schematic the CPU connects to the shared address bus through a couple of 74367 buffers at 7D & 7E, and the watchdog reset is generated by a 74138 decoder at 8N so one of these is my guess.
Checking the inputs and outputs of 7D found 2 outputs stuck low (despite the inputs toggling) so I swapped it out, ran the test board again and this time it ran, but indicated faulty work RAM. I was seeing RAM chip select and write-enable signals working so it suggested the RAM itself was faulty, so rather than mess about I just swapped both 2114's at 7N & 7P.
The test board now shows work RAM as good, but also object RAM (the 2 x 2101's at 4F & 5F) as good, which is positive news, but then it fails on the video RAM test. These had already been swapped by the previous repairer, note the state of the turned pin sockets below
I swapped both 2114's at 3F & 3H (into the existing sockets, with the intention of replacing the sockets at a later date). Re-ran the test board and this time all memory tests pass (but still the same mess on the screen)
As the object RAM is working, that proves the circuit as far as the Horiz. Pos. Bus so the next bit to look at is how the video RAM is addressed for displaying on the screen.
There are 2 74283 adders at 4N & 5N that generate the screen position signals. Probing 4N showed 2 of the outputs stuck low (despite the relevant inputs pulsing) so swapped it out and now we get something semi-sensible on the screen. Unfortunately it's showing only half a character repeated, instead of a full character. Moving to the next chip, a 74273 latch at 2M had output pin 6 feeding E2 stuck low meaning the lowest 4 bits of each character would get repeated. 2M was swapped and finally we get proper text being displayed during the tests (but with a solid yellow bar across the screen)
More repairs and commentary in the next part...
