PCB Repair Logs DoDonPachi
Forum Thread: DoDonPachi PCB Repair
The last board from a box that Bwodie sent me back about 6 months ago was a Do Don Pachi PCB, similar fault symptoms to the ESP De Rade board in that it would run but the gfx colours were a mess..
...the audio was also a bit strange, rather metallic sounding.
Gave the board a quick visual but didn't notice anything obvious, due to the similarity of the fault and the mint condition of the board I suspected a problem with the mask roms. This board doesn't have a RAM/ROM test and causing a fault by bridging a mask roms data lines together didn't give any error messages on boot-up either so the only way to rule out the masks was to desolder them all and test them one by one.
Going round each one with the scope didn't show anything unhealthy looking or conspicuous by its absence. Initially I thought I had found the fault as every mask rom read as unknown, the silk screening under the bank of 4 masks claimed the chips were LH5332A00 chips which are equivalent to 27c322 eproms. Every single mask gave an unknown dump, which could explain the fault if something nasty had happened to a common address line. It turned out that the silk screening is wrong, the MAME roms are 2MB in size yet 27C322s are 4MB eproms, when the masks were read as 27c160s they all gave perfect reads that were recognised as Do Don Pachi ROMs. The only bonus to this wild goose chase was that I found a smoked track...
under one of the audio mask roms, something had shorted it to 5V judging by the look of it. The ROM itself was fine and as this track goes to a set of link pads its probably to do with configuring the address lines for a couple of chip options.
Fixing this track sorted out the audio, from how its wired it seems that the smoked track sets a high address line so with it blasted it left the high ROM of the bus, so 1 byte in every 2 read was "empty" which would explain the raspy sound.
With all the mask roms ruled out I was left with the task of going round the surface mount RAM chips with the scope, again this showed nothing obviously wrong. The next step was to start poking the data lines, momentarily shorting them together to see the effect ( this is quite risky if you are not 100% sure of the pinout of the chip, its not uncommon for some lines to be tied high or low on purpose and its possible to destroy the chip if you end up connecting the wrong things together). Mostly I got varying degrees of graphics corruption and board crashes, but on one pair of SRAMs I got a slight darkening of the screen colours.
..then the strip of gfx at the bottom of the screen would suddenly come good.
Powered the board off and buzzed through the logic controlling the address lines, they all went to a bank of 74LS157s, these are quad dual-input multiplexers and the output of each of the 4 mux's went to one of the address lines on the bank of SRAMs. On all of the 157s each of the dual inputs per mux were busy, except the one on pin 10 of U31 which was held low, this looked suspicious. The track traced down the board and through a number of vias to the underside of one of the program EPROMs at U27, by pulling the EPROM I could see where the track went and also the fault. On the very edge of of the eprom socket there was some board damage, something had clearly hit the board on the very edge of the EPROM at some point and fractured the surface. As soon as I poked it the board surface flaked away, taking about 0.5mm of track with it. I could buzz through the fractured end back to the L157, and the other end to the custom chip UC1. A small length of hookup wire connected to the via under the EPROM and to a section of track I exposed by scratching the lacquer away bridged the gap.