25 Jan 2017

World Rally -Gaelco 1993 (repair log)

 The game booted but had 2 issues:
- RAM checksum error at address XXX (changes every time you power the board)
- Coprocessor not ready (indicates the game has committed suicide)

For the RAM error I first thought it was bad but discovered a cut trace. I patched it and the problem was fixed.

Then I tried to un-suicide the game. I first replaced the battery. Then I needed a RS232 to TTL converter and as my computer doesn't have a serial port anymore (I had it through the docking station but I got rid of it...) I went the USB to TTL converter route:

The pinout on the World Rally board is as follow:
- 1 (closest to the printed 0) = VCC
- 2 = TXD
- 3 = RXD
- 4 = /PROG
- 5 = GND
You must bridge /PROG to GND on the board and connect only 3 wires to the TTL adapter (GND/TXD/RXD). Pinout may vary according to the adapter so just check yours.
The port settings are 2400b/s, 8 data bits, no parity, 1 stop bit and XON/XOFF flux control.
With hyperterminal (or any other serial communication software) send a carrier return. The game should answer "DS5002 LOADER VERSION 0.1..."
Then send U, the game should answer "UNLOCKED".
Send L, the game waits for a file to load.
Copy and paste the following code:

Gaelco wrote:File too big

Can be found here:
(Or you can do as I did: download the romset on Gaelco website and convert the wrdallas.bin file to hex).

Transfer is quite long (between 5 and 10mn?).
When finished type Z to lock the DS5002.

Unfortunately it didn't work for me on first try. The cause was another broken trace. I fixed it and started the download procedure again and it worked!
Another game saved.

18 Jan 2017

Super Pang - Mitchell/Capcom 1990 (repair log)

This game is equipped with a suicide battery that holds the encryption keys of the Kabuki CPU (a modified version of the Z80). The latter was still the original one and was completely flat (0.0V).
FYI when this games commits suicide it only displays a static blue screen.

So, I followed the procedure described on "The Dead Battery Society" webpage:

Long story short, this mod allows the Kabuki to work as a standard Z80 CPU with decrypted ROMs.
3 ROMs have to be replaced and a simple wiring hack has to be done (as the decrypted ROMs are twice as big as the OG ones).

This brought the game back to life. Only to notice that sprites were severely corrupted:

This is due to a bad custom chip noted 86S105. Unfortunately I don't have any scrap board that could be a good donor... If you have a spare chip just leave me a comment.

12 Jan 2017

Sega Mega CD I (JAP) - Sega (repair)

Today's patient: a Japanese Mega CD model 1, a beautiful machine, but having a video bug on the Sega logo which is supposed to rotate on the attract screen. Also the machine hangs 1/2 or 1s after entering the CD menu: 

There are 2 possible causes:
- one or both word RAMs are dead
- one or several traces are broken

One of the most famous disease of the Mega CD 1 being the capacitors leaking with age and spreading their nasty fluid all over the board eating traces and vias, I will start to probe each trace between the RAMs and the associated Sega custom chip.

Opening the beast:

I had to remove the small daughterboard soldered to the main board (a last minute fix by Sega to correct a design flaw, that's why this board is absent from the newer revision motherboards) in order to have access to the Sega custom chip sitting underneath:

I found a broken trace on IC11 (data line D0). Capacitor C23 has leaked and eaten a trace. I patched it with a piece of kynar wire (the red one, green ones being factory soldered). I soldered one end to the RAM and the other end on a via after testing it was good (impossible to solder directly to the Sega custom chip due to the ultra fine pitch):

Reassembling and test:


As I don't want to waste my work I'll do a recap of the board.

4 Jan 2017

Garou Mark Of The Wolves (MVS) - SNK 1999 (repair log)

Graphics were severely corrupted (blocky):

I checked all the traces on the CHAFIO board but didn't find any broken one.
So I removed all the C ROM chips (8 of them, 42 pins each!):

I dumped them all and they were all identical to the MAME romset v1.
I finally transplanted them (and the M1 ROM) on a KOF99 board as I was pretty sure the CMC custom chip died (pretty common unfortunately).

Now I have a dead KOF99...