Caius

Oct 072017
 

A quick double repair log of Toki, a funny and addictive platform game released by TAD Corporation in 1989.

The first PCB:

On power up it was stuck on a blank static screen:

Doing a visual inspection I noticed some scratches on solder side:

Using my multimeter in continuity check I found a couple of severed traces in the 68000 main CPU area:

After patched them the game successfuly booted:

But sound was absent, this was due missing YM3812 OPL chip and custom SIL ‘HB-41’ module:

I finished the repair installing the missing components, this restored full sound.Board 100% working again.

The second PCB:

Board booted but backgrounds graphics were corrupted and flashing :

Also here I found a broken trace on solder side:

This restored graphics:

But, while I was testing the board, suddenly I lose sound and I could no longer coin up.The two issues are related each other since on this hardware the Z80  CPU commands the sound system but also handles coins.Indeed, probing it revealed a stuck signal on CLOCK pin 6 where a 3.579545 MHz one should be:

Looking at hardware, the clock is derived by a 14.318 MHz quartz then divided by the custom ‘SEI80BU’ (which also does interface between main and sub CPU)  and lastly divided again by a 74HC74 (14.318/4= 3.5795)

I replaced the quartz  with no luck , the 74HCT74 was not receiving any 7.159MHz signal so, for exclusion, the ‘SEI180BU’ had to be the culprit.I removed it and took the spare from a scrap Raiden PCB:

This restored sound hence the custom was really bad:

Double repair accomplished.

 

 Posted by at 11:06 pm
Sep 212017
 

Another board from the box of the “death”: The Simpsons

It immediately appeared to be a “desperate case”, board was heavily “harvested”, many components were missing:

A smashed TEST switch too:

I tried to power up the board and obviously it didn’t work, it was  in an endless loop:

Summing up, these were the missing/broken components : 

  • 6264 RAM @16G
  • 3.579545MHz oscillator @4G
  • ER5911 serial EEPROM @12C
  • MB3722 amplifer plus electrolytic capacitors @C21 (1000uf 16V), C16 (1000uF 16V), C114 (330uF 16V)
  • momentary switch @16A  

I reinstalled the RAM, the oscillator and the EEPROM (temporarily omitting components in audio circuit)  and board showed the self-test reporting bad devices @12C (ER5911 serial EEPROM), 6G (sound Z80 code 27C010 EPROM) and 7G (2k x 8-bit SRAM of sound system)

I cleared the first error by re-initializing the EEPROM (holding down the TEST switch while powering up the board) 

As for the other two errors, I dumped the EPROM @6G and code didn’t match any existing MAME set so I reprogrammed the device with correct code.This cleaared the error but the one regarding the device @7G was still present.Probing the device (a Sony CXK5814, a 2k x 8-bit static RAM compatilbe with 6116) revealed unhealthy signals on data lines:

Once removed, the chip failed the out-of-circuit testing:

Finally the self-test no longer complained so board booted into game but colors of some backgrounds were wrong and sprites were mssing parts:

The MASK ROMs check reported a bad device @16L, this is a 8Mbit one containing sprites data:

 

I replaced it with a 27C800 EPROM but with no luck so I focused on the background colors issue.The hardware uses an ASIC marked ‘053251’ which is responsible of palette and priority:

The chip outputs 11 bit of palette index called ‘CO0-CO10’ on schematics:

Probing the outputs with a scope revealed bad signal on pin14 ‘CO4’ (healthy signal on the left for comparison)

I removed che custom and replaced it with a spare take on a donor board (Premiere Soccer)

This fixed the color issue but sprites were always wrong :

and device @16L was still reported as bad:

Schematics don’t cover this part of circuit but I knew the design.The two ASICs marked ‘053246’ and ‘053247’ work in tandem, the first generates the address lines to the sprites MASK ROMs and the latter read and process their data:

I checked connections between the ASICs and MASK ROMs, no problem found.Actually the data lines from MASK ROMs are not directly tied to the ‘053247’ ASIC but there are some 74LS374 in the middle (two for each ROM) which latch data bits.Piggybacking the ones @12K and 10K restored partially the sprites.I desoldered the first and it failed the out-of-circuit test:

 

This improved sprites but not yet perfect :

So I removed also the other 74LS374 @10K, it failed too!

This fixed the sprites issue completely.As for lack of sound it was only matter to fit the missing MB3722 amplifer and electrolytic capacitors  @C114, C21 and C16.

Another board torn from the darkness, long live to it!

 Posted by at 7:36 pm
Sep 182017
 

We all know that Konami manufactured wonderful arcade games but, you know, beautiful things are often complicated too.And surely their hardware is! This mainly because of the use of many custom chips with the most disparate functions and shape.

The ‘052535’ is one of them, used on countless PCBs of ’80-’90 :

The ‘052535’ is basically a 5-bit video DAC (one for each R,G,B color) in SIL package used to convert the digital signals of the palette circuit into analog, we can see its pinout and implentation in this snippet from Lethal Enforcers schematics:

During my repairs sometimes I had to replace faulty ones so why not reproduce this part too?Perhaps someone else did it already but i did it my way.

First of all I removed the black epoxy to expose the circuit and scan it :

The black squares are printed film resistors, the part on the left marked ‘LF’ is a NPN transistor in SOT323 package, the four SMD parts are zero Ohm resistors.I metered the printed resistors (but first I removed the soldered parts to avoid interactions), this was the result :

As you can see the Konami ‘052535’ is nothing more than a R-2R resistor ladder with the resistors values tipically doubled (starting from 2.5KOhm up to 43.50KOhm).The NPN transistor (I marked it as a BC848 but it’s a NPN general purpose one)  is used in final stage to amplify the analog signal adapting it to RGB arcade standard.The circuit is very simple so it took few time to draw schematics of it and route them to a PCB which ended up with more or less the same dimensions of original part:

Sent it to manufacturer and after some time got the bare PCBs:

Here is the assembled reproduction, I used the legs from thru-hole components as pins which fit well in a female header: 

Testing on board was successul, reproduction validated!

See you all to next project!

 Posted by at 9:22 pm
Sep 172017
 

Another board from the box of faulty ones, an Irem M92 motherboard (with Undercover Cops license seal)

For my testing I used the ROM board from a working Undercover Cops (but I could use a whatever one).On boot I was greeted by this screen, self-test failed  reporting errors related to some RAMs 

There are two  6116 SRAMs @IC43 and IC44 whose data bus is tied to the two graphics custom ASIC ‘GA22’ and ‘GA21’  :

When I went to piggyback the RAMs, error changed in this:

Probing them with a scope revealed the outputs were dead on both despite the address lines were properly toggling:

Both chips failed when tested out-of-circuit:

With good RAMs the board successfully passed the self-test and entered in game:

But sound was missing at all, I could hear only a slight buzzing noise:

Using my audio probe I was able to verify that both FM sound and PCM samples were properly generated and then pre-amplified/mixed by OP-AMPs but signal got lost somewhere not reaching the input (pin 1) of the MB3730 amplifier.The whole audio section was previously recapped by someone so I ruled out electrolytic capacitors.But a closer inspection revealed heavy corrosion on many 1/8W resistors (probably due leakage of original ELNA capacitors), here are pictures of them taken with a microscope:

Measuring them gave me infinite resistance sign that they were open:

Here in details the bad resistors I replaced:

  • R201 –  1KOhm
  • R203 –  100KOhm
  • R205 –  4.7KOhm
  • R208 – 4.7KOhm
  • R209 – 2.7KOhm
  • R211 – 10KOhm
  • R213 – 10KOhm

This restored full sound, board 100% fixed.End of job.

 Posted by at 11:58 am
Sep 132017
 

Received some days ago a box of faulty boards for repair.There was a clean Sunset Riders PCB:

Game played fine except for sound since samples were only noise (music was fine instead):

Sound samples are stored in a 8Mbit MASK ROM @1D:

MASK ROMs check reported it as bad:

I was sure device was really bad (Konami ones are well known to be prone to failure) so I removed it :

But it turned out to be good instead since reading it ended up in a good dump and also replacing it with a programmed 27C800 EPROM had no effect .

Sound samples circuit is very simple, the ASIC ‘053260’ (a custom PCM/ADPCM sound chip) addresses the MASK ROM @1D and plays its data:

No other component is direclty involved so most likely the ASIC was the culprit.I removed it :

Took the spare one from a scrap Asterix and soldered it back on my board:

Success!MASK ROMs check reported now a good device @1D hence board 100% working and fixed :

 Posted by at 6:33 pm