Feb 212017

Some days ago I had on the bench for a repair this pretty rare Air Raid PCB (manufactured by Seibu Kaihatsu) :

As you can see from picture above hardware use some big SIL modules which contains all tiles/sprites data (currently undumped but MAME team is working on).My board was faulty, it has sever graphics issues :

Tiles was filled with garbage :

And sprites had blank lines through:

PCB was in almost mint state, only two 2k x 8-bit static RAMs replaced.Besides, it was fully populated with Hitachi TTLs which  are very reliable.So, my suspicions fell on the remaining RAMs (more than on SIL modules) since they were all TMM2015 (like the already replaced ones, all 6116 pin to pin compatible) which means high chance of failure

Probing the two @6D and 14D revelaed unhealthy signals on data lines:

They failed when tested out-of-circuit:

Replacing them restored graphics completely.End of job.

 Posted by at 11:36 pm
Feb 202017

Received for repair this original Track & Field (manufactured by Konami but actually it mounted the Centuri ROM set)

On power up it showed almost nothing on screen, most of graphics were missing, only some sprites were barely visible:

On a closer ispection board was corroded especially in the backgrounds generation circuitry, both sockets and inserted ICs ( the ‘082’ custom and the three EPROMS) were affected:

I replaced the sockets, rebuilt the ‘eaten’ legs and cleaned the oxidized ones but this didn’t improve anything.All the graphics is generated on bottom VIDEO board and most of the logics were from Fujitsu (which means an almost certain failure) so I went to probe TTLs with my logic comparator and found a bad 74LS157 @9H (its outputs generate some address lines for the two 2018 tiles RAMs)

This improved a lot, more graphics were displayed :

Always probing in-circuit the Fujitsu TTLs, I found a 74LS139 @14D with all floating outputs:

It obviously failed when tested out-of-circuit:

Now graphics were mostly present but gameplay was accelerated and sound missing too:

This is issue must have caused by some problem with main CPU,  all the circuitry was on top board and also here Fujitsu TTLs were massively used.Probing around I found a 74LS244 @1C just near the Konami-1 CPU with some stuck outputs:

Replacing it restored completely graphics, speed and sound:

But I couldn’t coin up and start a game with both players, inputs were unresponsive.Tracing them back from edge connector they are tied to some 74LS253, from Fujitsu of course:

Without thinking twice I removed the two ones @3E-3F involved in P1 and P2 controls (the other two are for P3 and P4)

They both failed when tested out-of-circuit:

Now board played perfectly, the last issue I had to solve was that highscores and records were not saved, I got only some random characters instead:

This obviously had to do with the +3Volt CR2032 battery installed on CPU board , it was really low (only few millivolts measured)

I installed a fresh battery with its holder to finish the job :

A great classic preserved!

 Posted by at 10:29 pm
Feb 082017

I had on the bench this Rastan PCB:

Board suffered from severe graphical issues, sprites were absent and jailbars all over the screen.Sound was corrupted too:

I had the feeling that the backgrounds were actually good although the jailbars since the lines were changing as sprites moved so I went to check this part of circuit.Objects are generated by the custom ‘PC090’  which takes data from some 1Mbit MASK ROM and write them to four 2018 static RAMs:

Probing the four RAM revelead weak signals on data lines of the one @IC5:

Piggybacking a good RAM restored all graphics:

The chip failed when tested out of circuit:

But after I installed a new RAM on board, sprites were not good as when I made piggybacking, just blank with no colors :

Probing around I found a 74LS245 @IC85 which didn’t do its job :

This is used to exchange data between two bus called ‘IO BUS’ and ‘MD BUS’ on schematics:

Chip failed the out-of-circuit test:

Graphics were OK now so I moved on to trobleshoot the sound issue.I found three missing electrolytic capacitors in sound section : @C23 (1000uF 16V), C26 (10uF 16V), C28 (100uF 16V):

and replaced a bad one with increased ESR @C27 (100uF 16V) :

But this was not enough to restore a good sound.So I decided to use my new audio probe:

An audio probe is essentially a portable amplifier useful to test parts of an analog sound circuit by ‘hearing’ them and hence figure out what is actually faulty.Mine was designed by Phil Murr from UKVAC forums, credits to him:

Using it revealed that sound was still good on input pin 6 of the TL074 OP-AMP @IC106 (used on final stage to mix FM music with PCM samples) but then was outputted distorted from pin 7 :

Replacing the OP-AMP cured the issue.Mission accomplished.


 Posted by at 11:21 pm
Feb 042017

Received a Lightning Fighters for a repair.Board was in bad shape, some ICs were missing and solderside had a lot of scratches:


In the details, it missed two 4Mbit MASK ROMs used for sprites (but rounded machine-tooled strips were installed in place of them), the YM251 and MB3722 amplifier and its heatsink.Besides, the YM3012 DAC was partially desoldered and damaged:

At power up I was greeted by the typical flashing blank screen, board was constantly resetting due active watchdog circuit:

This meant that main CPU program code couldn’t be properly executed due a probable problem in  68000 busses.With the help of schematics I went to check this circuit and found a missing connection between data line D4 (pin 18) of the program ROM @E15 and corresponding data line (pin 1) of 68000:

Tracing back both pins I found a broken trace on solderside :

I restored the connection and board successfully booted with obviously no sprites and sound due missing components.I decided to fix the graphics for first so I burned two Macronix MX27C4100 EPROMs (same pinout of 27C400) to replace the two missing 4Mbit MASK ROMs @K2 and K8.In this way sprites reappeared but they were glitchy:

MASK ROMs check reported a bad device @K2 (and another one @C5 but this was due the missing YM2151 as I discovered later)

Again I checked connections of the device @K2 against schematics and found intermittent contact between pin 21-31 and VCC (pin 31 of a 27C400 EPROM must tied HIGH to enable the Word-wide organization of data).Who put hands on this board removing the two MASK ROMs and installing round machine-tooled managed to rip off the rivets which are inside the holes, this explains the poor contact of the two pins.I reinforced the connections with some AWG30 wiring them between the solderside of the pins and alternative VCC points :

Sprites were good now but still not perfect, some of them (like explosions) had jailbars:

I reflowed the ‘053244’ custom ASIC sprites generator:

Jailabars were gone:

All graphics were pefect now, confirmed also by the MASK ROMs check:

So only the sound needed to be restored.l fitted all the missing componens (the YM2151 and MB3722 amplifier with its heasink), removed the partially desoldered YM3012 replacing it and at same patched some broken traces on its solderside:

MASK ROMs check reported now success also on the device @C5 (which contains PCM samples)

Once entered in game I had confirm that sound was fully working.Another PCB saved from certain death!

 Posted by at 3:26 pm
Feb 032017

Found this Sunset Riders PCB in a lot bought from a former arcade operator:

The board worked fine except for the audio, sometimes wrong samples were played and often music was not present too:

The MASK ROM check reported a bad device @1D which is the one containing samples:

Obviously I removed this MASK ROM and dumped it but it turned out to be good.This device is addressed by the near ‘053260’ ASIC which processes its data as well.Tracing connections I could figured out that this custom receives serial data from the YM2151 then mixes them with PCM data and output the whole packet to the YM3012 DAC so it’s something more than a samples player, this would explain why sometimes music was missing too.Before replacing the ‘053260’ ASIC I analyzed it with a logic probe and found its pin 10 totally silent, this had to be connected to pin 16 of two 74LS245 @13F and 14F which allow 8-bit of data exchanging between the ASIC and main 68000 CPU data busses.

A simple jumper wire cured the issue.End of job.

 Posted by at 11:26 pm