Sunday 22 August 2021

V-Five repair log and Toaplan 'RC8528' reproduction

Received for repair a V-Five PCB, a  vertical shooter arcade video game originally developed and published by Toaplan and also know as  'V-V / Grind Stormer'.The board was in very good condition :

But it booted all the time to a 'PALLET RAM ERROR' message on screen (I think 'pallet' is a japanese misspelling of 'palette')


The palette RAMs are two 2K x 8-bit devices @U4-U5 (Sharp LH5116 the exact part used)

Probing them with a logic probe revealed some stuck outputs of the device @U5 :

I promptly pulled the chip and tested it out circuit.It failed :

With a good RAM chip fitted the board finally booted into game:

I wanted to play some games for testing further the board but I could not coin up with player 1.Pin 16 parts side of JAMMA connector was stuck low :

I traced the input back to a custom resistor/capacitor array marked 'RC8528' :

Measuring the resistance of the internal resistors (which act as pull-up) I found one with a value lower (588 Ohm) than others (4.7K Ohm)

The part was bad internally but I had no spares of it.I remembered some years ago I reproduced a similar custom resistor/capacitor array used on NMK PCBs (marked as 'NMK-006') so, after figured out the correct resistors values, I assembled one and installed it :

But I was still unable to coin up.Still going backwards there is a 74LS240 @U29, input pin 8 was almost shorted to GROUND :

 Pulled the device and tested it out-of-circuit, it failed in the exact logic gate unit :

Fitted a good chip and I was finally able to coin up.All seemed working fine but going futher into the game I noticed some sprites were glitched :

All the graphics are generated by the custom ASIC ' GP9001' which produces the addresses for six Fujitsu MB81C4256 (256K x 4-bit) dynamic RAMs and read back their data :

I found the datasheet of this DRAM and looked at pinout :

Probing the data outputs revealed unhealthy signals of the device @U17, you can see them on the right of the below picture (compared with good signals on left)  

The device failed when tested out-of-circuit:


 A fresh RAM chip was installed on machine tooled socket to finish the job :

Board 100% fixed and another successful repair.



Sunday 8 August 2021

Bubble Bobble repair log and 'PS4' custom MCU reproduction

Who among you does not know and has never played Bubble Bobble in his arcade life? Personally I love this game and I was looking for an original board since ages but never been lucky to get one so I always ended up settling for bootlegs.But recently I spotted on eBay a legit PCB from Taito.The description said the board was in good condition but not working due to missing security chip.Despite this I made some bids and won the auction (final price was not really cheap though...).When I received the board there were no surprises to me, the chip @IC17 on top board was missing hence the PCB was dead :

 This chip is, indeed, the custom MCU with Bub silkscreening, here's a picture of it found on the net:

Some years ago this chip has been decapped and internal ROM extracted improving the MAME emulation that used a simulation of the chip behaviour until that moment.You can read more about on the blog of Nicola Salmoria (the original developer of MAME)

http://mamelife.blogspot.com/2006/08/good-news.html

http://mamelife.blogspot.com/2006/08/completed-at-last.html

Thanks to this brilliant work we know now that the Taito 'PS4' custom IC is a Motorola 6801U4 MCU in disguise with some additional features.Since the internal ROM was dumped the only chance to repair my board was to get a chip and program it with MAME dump.Sadly I found no MC6801U4 around but only MC68701U4 which is pin to pin compatible.The chip arrived and thanks to one of my programmers I was able to program it with the file in the MAME ROM set :

With MCU installed the PCB came back to life, the game was fully playable with sound but some parts of graphics were wrong or missing :

 


 


All the graphics are generated by the bottom video board :

 Data are stored in 12 27256 EPROMs :

 

I dumped all of them and did a compare with MAME, two files were unknown as they didn't match any existing ROM sets :

 


 Trying the bad dumps in MAME exactly reproduced the board issue :

Reprogramming two blank EPROMs fixed the graphics.Playing some games I noticed the sound was quiet also on highest volume setting , I fixed this issue by fully recapping the sound section :

The repair was now complete, board was 100% working but I thought that solution to replace the 'PS4' custom MCU with a MC6801U/MC68701U4 was not accessible to everyone because very few programmer supports this obsolete device.So, I looked for another solution and found that some bootlegs use a little sub-board with a different MCU and some glue logics as replacement of the 'PS4' custom IC.I managed to find such bootleg and did a modern reproduction of this replacement, it worked fine when I tested it :

 

 

Another successful repair and another custom IC preserved!

Saturday 7 August 2021

1943 : Midway Kaisen repair log

Got for repair a 1943 : Midway Kaisen PCB (the japanese release of 1943 - The Battle of Midway).The board, a two stack one, was in good condition :


But according to the owner the PCB lost its sound after he connected it backwards.I had confirm of the issue as soon as I powered the board up :

 

Using my audio probe I was able to figure out that the lack of sound had a digital nature.Indeed, probing the Z80 audio CPU revealed the /INT line (pin 16) was asserted all the time :

This means a maskable interrupt to the CPU was triggered by an external I/O device.This causes execution to jump to a specific interrupt vector (which is some code at a fixed location) so the CPU was not properly running.

The Z80 CPU was socketed by factory so this was a good chance to fire up my Fluke 9010A troubleshooter and perform some tests.Looking at the sound memory map in MAME source the Z80 RAM lies from 0xc000 to 0xc7ff of the address space of the CPU :

On the board the device is, indeed, a 2K x 8-bit static RAM, specifically a Toshiba TMM2015 which is well known prone to failure :

 

 

I set up the Fluke 9010A for a RAM long test which failed at initial address :

 

 

This meant the RAM chip was likely bad so I removed it :


The pulled chip failed, indeed, the out-of-circuit testing :


 I fitted a socket and a good RAM chip :

This restored the sound and fixed the board completely.Another successful repair.