15 Nov 2018

Tumblepop - Data East 1991 (repair log)

Board had apparent corrosion damage.


Surprisingly on boot up it worked mostly fine, only sprites were a diagonal garbled bar:



I probed the board and found two data lines on one of the sprite RAM were stuck low. I pulled it but it tested good. I continued probing the board and finally found a missing input on the LS157 @D4 (pin 3, signal I1a). I tried to followed the trace but it then disappeard under the custom chip '52' used for sprite generation. By deduction I found it should be connected to pin 13 of LS157 @11A. Restoring the connection made sprites appear at the right place but still garbled:


I went back to the stuck data bits on one of the sprite RAMs, it lead me trough bus transceivers, etc. And  finally to the '52' custom chip. On the other "side" of the chip were mask ROMs used for sprites so it was double or quits. Without much hope it checked connections between the custom chip and the mask ROMs and found two data bits signals weren't reaching the '52' chip. By exposing copper just before the pads of that chip I discovered connections were lost on the last half of a millimeter before the pads. If restored connections with a bit of solder and this time sprites were fully back:




Game fixed.

7 Nov 2018

Raiden II 'new' - Seibu 1996 (repair log)

This is the 'new' version of the game with audio of lesser quality amongst other differences.
Board was really clean, no visible damage or severed trace. However graphics were heavily corrupted (sorry for the pictures quality):



Some sprites were filled with garbage and background was messed up too.
Unfortunately this game doesn't have any kind of diagnostic menu and components being SMD it wasn't possible to do piggybacking. Also most of the chips were customs... In other words problem could be caused by either RAM chips and in this cause could be repaired, or the custom chips and the board would end up in the scrap pile.

I quickly identified the many RAM chips used:
- U021/U022/U027/U028: work RAM
- U0726/U0727/U0728: sprite RAM
- U0527: background gfx
- U0528 : texts
- U0824/U0825: colour RAM

By probing sprite RAMs I found U0726 had weak data signals: replacing it fixed the sprite issue.
Then background problem: again replacing U0527 fixed it:



As I didn't have any 6264 type RAM in my stock I used 62256s, they are 4 times bigger (32kb instead of 8k) but share the same pinout appart from pin 1 (NC on 6264, A14 on 62256) and pin 26 (CE2 on 6264, A13 on 62256):



Here's a simple trick I use in this case, pin 26 isn't a problem as CE2 must be connected for the 6264 chip to work, however, pin 1 being not used it's not rare it's simply not connected to anything on the PCB, and this was the case here. When using a 62256 instead of a 6264 some run a wire from pin 1 to Vcc or ground, but I prefer to simply bridge pin 1 & 2 together, RAM chips being simple one dimension arrays, as long as you read data back at the same place where you wrote them it's fine:


Game fixed.


P.S.: On last hint, the Raiden II "new" and Raiden DX "new" boards are identical and in fact 2-in-1 boards. To swap game, on boot up maintain player 1 controls as follow:
- all 4 directions (needs a joystick hack) + button 1 to swap to Raiden II "new"
- all 4 directions (needs a joystick hack) + button 2 to swap to Raiden DX "new"
Also please note swapping game erases settings.





31 Oct 2018

Hyper Olympic (Track & Field) - Konami 1983 (repair log)

When I first powered this board is was mostly working: title screen was corrupted but sprites were OK and I could see runners during the intro.
However after few seconds it started to fail hard: background disappeared, then texts and finally sprites...

As always a quick inspection of the board revealed many Fujitsu TTL chips:


And as usual I probed them all and found 8 bad chips (far less than expected given my previous repairs of Fujitsu plagued games):
- 2 * LS74
- 1 * LS138
- 1 * LS157
- 2 * LS244
- 2 * LS245


After replacement I powered the board again fearing for more damage, but:



Board was perfectly working with sound and controls!

Game fixed.


17 Oct 2018

Sega 837-6443 - System 24 FD controller reproduction

Floppy based games on System 24 are becoming incredibly hard to find, not that rom based games are easy... but at least way cheaper, probably cause they are crappy Mah-jong or Quiz games in Japanese.

I designed a reproduction of the floppy drive controller so anyone with a rom based game can enjoy floppy based games:



It's a 1:1 replacement, same dimensions, same layout and fits in place of the OG one. And it works just great :






10 Oct 2018

Act Fancer - Data East 1989 (repair log)

This board was given to me as faulty in a deal.


Upon power up game was stuck on a garbage screen:


I quickly noticed work RAMs were shiny ones (known to have a high failure rate) but I dumped the 3 program ROMs first: they matched the Japanese rev.1 romset in MAME. Probing the RAMs didn't revealed anything weird but by using the oscilloscope I could "see" main CPU was sending data cyclically. Cleary no valid code was executed and CPU was stuck in a loop. I pulled both work RAMs and both tested bad on my programmer. After replacement game booted mostly fine, with sound, texts and background but sprites were replaced by big misplaced blocks of garbage:



Again I pulled and dumped the 8 sprites ROMs but they all matched. RAMs didn't show anything weird either so I started probing the TTL around and noticed one of the inputs of the LS10 @J4 was stuck low. Being a triple NAND gate it meant the associated output was also stuck (high this time = NAND). I tracked the signal to the Q output of one of the latches of the LS74 @J1. All inputs were correct with activity where it should be but outputs Q and /Q were just stuck. Piggybacking a new LS74 chip on top of the suspected faulty one cleared the issue. I replaced it and sprites were fully back:



Game fixed.

3 Oct 2018

Sega 834-6510 I/O board reproduction - part 2

This time I received the blank PCBs.
As you can see layout is identical to the OG one:




Side by side comparison once assembled:









26 Sep 2018

Mutant Fighter - Data East 1991 (repair log)

Game took very long time to boot, then would play extremely slow, sometimes hanged, sometimes played overspeed for a second, and also had no sound. Otherwise board was in very good condition like it's never been operated:


I started by troubleshooting sound issue. First obvious thing I noticed was sound RAM @ J24 was a shiny one. Piggybacking a known good RAM on top of it restored sound. I replaced it and that was all that was needed in the sound section.

Then the second issue regarding speed of play, hanging, overspeed. Quite weird. I probed main CPU (68000 CPU in a custom QFP package marked '59') for a dodgy RESET or CLOCK signal: nothing to mention here. Then I found by touching a specific area of the board on the solder side I could get the game to play at correct speed for a second, and by doing so every second I could have the game playing normally. I thoroughly inspected the area and found a factory defect: one of the leg of the LS74 @ K3 wasn't soldered. I had the impression that by applying pressure on it game would play fine.


I resoldered it but nothing changed. Thinking the chip could be bad and not having the possibility to test it on board with my logic comparator because of it being SMD, I took the decision to replace it: no dice.
To me there was clearly a floating signal somewhere and by touching the concerned trace/pin with my hand I could restore the connection or at least drive the signal to a state where the board would be happy with it. I quickly found a floating trace, that lead me to an other LS74 @ J5. This time chips seemed soldered properly but by using my magnifying lamp I could see one of its legs was covered in solder, so as the associated pad but there was a really tiny crack in between. I tried to reflow the leg but solder was stubborn and didn't want to adhere to it. I had to add quite a big blob of solder for it to adhere on both the leg and the pad:


This time game played normally:



Game fixed.