MERCS B-Board #1: No video. Actually a fix on the C-board, simply one of the inductors had a broken leg. Reattached, works fine.
MERCS B-Board #2: No video. Also a fix on the C-board, this time by reflowing the solder on the big CPS B-12 chip.
Renegade bootleg: Not much sound. It would scream at me when I first turned it on, then nothing. The audio CPU would work at startup for a moment then go silent, albeit without /HALT or /RESET. Checking the audio cpu ROM, it gave me no matches on the mame database. I pulled a few romsets and found that by inspecting other ones.. (to be continued, forget which board I was working on..)
Cook Race: No audio. The 6502 that controls the audio seemed dead and hot to the touch. Replaced with a spare, audio returns! One of the sound effects was super loud compared to the others. The sound effects all get to the final amplifier by way of a simple resistor-based mixer, and one of the resistors was 47k instead of 4.7k, giving that sound effect a much larger amplitude. Judging from the flux around this resistor, it was replaced at some point with this incorrect value.
Nibbler: Wrong colors and missing half the sound effects. Made an adapter to read the color PROMs and they are "Pioneer Balloon" PROMs, so someone did a conversion and didn't quite finish! For the missing sound effects, I started checking the schematics and quickly found that one of the '161s (IC58) wasn't outputting the carry bit, which is needed.
Spiders: What a neat game. This should be a classic. Anyway, this game only had sound issues. I spent a few days looking at the terribly scanned schematics and combing through the MAME drivers (which were super helpful). Basically, after verifying everything should be working, I happened to notice that one of the main analog switches (CD4066) had a floating GND pin. I actually still don't know where the break is, but it wasn't connected to the rest of the GND net. I soldered a little jumper to a nearby ground and all the beautiful analog sound returned. Seriously, the sound on this thing is all generated from 555 timers.
Galaxian (Midway): Missing the stars and the explosion noise when the ships get hit. This article on Marco's Arcade gave me some quick tips on fixing this. Neatly (word?) enough, the starfield and explosion noise are both generated by a pseudo-random generator consisting of some '164 shift registers. I checked the '164s and one was bad. Replacing the '164 brought back the starfield and sound but the stars appeared to be streaking beyond the border. Marco's article also says the '366 gate at 2B controls this, and while that chip seemed to be OK, one of the enable pins was stuck low. Traced back one chip to a '10 (3-input positive NAND) which was borked. Replaced it and we're good.
Mania Challenge (bootleg): Distorted audio, replaced one of the 4558 audio op amps, fixed.
Nebulous Bee (Galaga bootleg): Booted but no sprites. 3 of the PROMs were disintegrating, replacing those brought back the sprites and got sound going. All the music had the wrong pitches, replaced a few Fujitsu TTLs on top and got it back to normal. The culprit was a '157 at F8. After that, everything was working except the tractor beam, which would flash white. This is on the "4L" ROM. Unfortunately the video board is littered with Fujitsu TTLs, so I started replacing things around that ROM. The Galaga proper schematics are useful if you can get your bearings since the layout is completely different.
NOTE TO SELF: Never try to shotgun a board by pulling like 10 chips at a time. If you do this, and you put the chips back, and your board now appears dead, it's going to take you a LONG time to figure out the mistake you made.
Still working on this final issue with the tractor beam.
Police 911/2: Apparently this game is was a kit offered to upgrade Police 911, which makes sense since it's such a unique cabinet with its body sensors. These boards will often have a dead timekeeper, which is a M48T58Y-70PC1 from ST Microelectronics. This is a bit of SRAM with a battery and realtime clock. When turning on a Police 911 that has been converted to 911/2, a dead timekeeper will show "CONVERSION FAILED", whereas a missing timekeeper will show "RTC MEMORY FAIL". I removed the original timekeeper and put in a socket. Since there is no dump of the Police 911/2 NVRAM contents, I attempted to program the new timekeeper with the NVRAM contents of Police 911 (the first one) from the MAME romset, using my GQ-4X EPROM burner. There is a note that pin 26 must be shorted to pin 28 during writing. I did this, and while it would program properly, the verification would fail, so I'm not actually sure if this part of the process worked. Nevertheless, I put the timekeeper back in the socket and booted the game (with Police 911/2 security dongle and CF card in place). The game booted, this time I got a "CONVERSION SUCCESS". Next it showed "RTC MEMORY FAIL" (I think) but this is to be expected. It sent me to the test mode where I exited, and (barring some warnings about sensors missing), the board booted properly.
Midway / Atari Seattle Boards: These are proving to be tough. I have seven boards and only two are working? Haven't been able to "fix" any of them yet. I had some theories about interchangeability of the ROMs (between MACE, Blitz 99 and Blitz 2000) but swapping things around I can't seem to find any incompatibilities despite board differences. For example the -05 revision board is missing U3 and U65 (which are present on the -08 revision) which seem to be sound FIFO buffers, but on the three aforementioned games it doesn't seem to make any difference.
#1: Revision -06. Works, syncs, but no video.
#2: Revision -08. Appears to boot, gets as far as blinking LED4 which is "normal operation". But no activity on HDD, game doesn't appear to actually be playing at all. No sync or video.
#3: Revision -06. LEDs stick on indicating no activity at all. Tried changing CPU, no change.
#4: Revision -08. LEDs stick on indicating no activity at all. Tried changing CPU, no change.
#5: Revision -08. LEDs stick on indicating no activity at all. Tried changing CPU, no change.
Empire City 1931: Game was mostly working but only showing blue on the video output. The color RAM on this board is a 6116 for red/green and a 2114 for blue. Replaced the 6116 to bring all colors back
MEC-M1 (Robocop): Looked working except the blue video was streaking and red was missing. The color RAM on the motherboard consists of three 6116 (45ns) SRAM chips. Decided to replace all three, working great now.
MEC-M1 (Bad Dudes / Dragon Ninja): Background tiles were being repeated vertically, sync was not working unless board was flexed. Close inspection of the quad-flat-pack chips showed some legs were bent. Straightened the legs and resoldered a few, this brought back sync and fixed the repeated tiles. Sprite issue remained (one in four lines missing basically), but pressing on one of the QFP chips fixed it. Held the chip down with a screwdriver and heated around the chip with a heatgun until solder reflowed. Problem fixed!
Paddle Mania: Every other video line was messed up. Boards of this era (1988) often have 2 line buffers RAM chips, where an entire video line is held in RAM while the next line is built for a second RAM. It was fairly obvious which RAM chips were the line buffers due to their proximity to the resistor DACs. Piggybacked a working 6116 onto each and verified which one was bad, then replaced it.
Double Dragon II (bootleg): Sprites and overall video issues. Overall video issues seemed to be due to iffy ribbon cables. After getting most of the video back, the sprites were missing lines or they would be shifted over by several dozen pixels. I figured out where the sprite RAM was by shorting data lines. Next to the RAM was a pair of '374 ICs which are flip-flops. I noticed one pin was floating and deduced it should be connected to a RAM data line. Found a broken trace on the bottom of the board. Repaired and fixed!