Arcade PCB Repair Logs

Arcade PCB Repair Logs #4: Capcom Ghosts 'N Goblins (Bootleg)

Hello faithful readers,

Back again, this time with some new tools to play with.

I have to start off by raving about the NeoLoch LLC Inquisitor board and SRAM blade. The Inquisitor board is a platform for IC testing and more, and the blades add compatibility to specific families. The SRAM blade does all the common SRAM I've come across, such as 2114, 6264, 62256, 2147, 2148.. nicely this includes the 4-bit SRAM that most EPROM writers don't bother to include compatibility for. In general, I've found that my Topmax 3100 EPROM writer doesn't do a good job of testing SRAM at all, since all it allows you to do is write a specific set of data to the SRAM and read it back. Anyway, I bought the unassembled combo, and had it up and running in about half an hour.

Let's take a look at this bootleg Ghosts 'N Goblins.

Thankfully, schematics exist for the official Capcom board, and this bootleg generally follows the same layout. A few things have been shifted so a chip marked "5M" on the schematics might be "6M" or "5N" on the bootleg. Click here to link to the manual and schematics page of the wonderful International Arcade Museum website.

The board uses a Capcom pinout, which looks deceivingly like a JAMMA pinout to the untrained eye. This can cause a lot of trouble since the 12V power connections will be in the wrong spot. 

Upon powering up, most of it starting coming to life. Unfortunately, I forgot to take a picture of the initial power up, and I forgot to document exactly what the first few problems were. As I recall the screen was upside down, there were obvious graphical errors, and input problems. In fact I'm pretty sure when I first booted it up it gave me an error code and wouldn't make it to the attract screen.

Rifling through the schematics, I found that the "FLIP" signal was used all over the place to ensure the graphics were the right side up, but the DIP switch that generates this signal wasn't having an effect. In fact, it was looking like several things on the DB0-7 bus weren't working right. First, I inspected the 74LS138 which is an address decoder, used to activate different chips to allow them to use the bus when they're supposed to.

Note how "IN1" and "IN2" are tied to the "/EN" (enable) on the '367s, which in turn connect to the inputs like coin switches and controls. The outputs looked unstable so I pulled it. Popped it into the ole IC tester and it shows as "Not found!".

Tried another '138 which tested "GOOD" so I popped it in.

This let the video flip properly so it was right side up and I could get into the game.

Next, I noticed the inputs were still messed up. Pressing the "right" button would cause the character to stutter forward erratically, and it was obvious it was related to input (as opposed to video). I probed around the 74LS367 (buffer with 3-state outputs), and was finding confusing results. I didn't have a logic comparator at this point, so I decided to hook up the Kraken, aka the logic analyzer. Forgot to take a picture of the analyzer in action, sorry.

This lovely doohicky can sample up to 8 channels of logic for short periods of time, and fewer channels for longer periods of time. I put a logic clip on to the relevant '367, and hooked up three channels: one for the '367 chip enable (/CE), one for the button input, and one for the data output to the bus.

What's expected? When the "Left button /EN" goes low, the input should be copied to the output. As you can see from the screenshot, this was not always the case. In the screenshot, the button is held down (which is why the input is low). You can see that when the /EN line goes low, the output line jumps up and down. This is bad. The '367 was removed, confirmed as broken and replaced. Inputs working now. 

Now we come to the sprite issues. Horizontal lines through all of the sprites revealing the background behind it.

I probed around the sprite generation section on the video board and everything looked OK to me. I replaced one or two of the 2114 SRAM, mostly just to see if the old RAM was still good. The "Shiny TMM/TMS" RAM, which has more of a polished surface than most ICs, are often bad. My SRAM tester got its first workout, but they showed "Good". So I moved my attention to the line buffers. This made sense as a place to look since the issues were in horizontal lines. Probing around, it appeared that on one of the SRAM data pins, the signal was stuck high. I pulled out the pair of TMM2018 SRAM chips, and they tested "Bad"! I put in some sockets just in case, and put in some tested good 2018 SRAM, and all graphical issues disappeared.


Welcome back GnG!