Got this Sunset Riders PCB always in a trade with my friend Ifog:

Board was really in mint condition but, as my friend warned me, it was faulty.

Actually when I first powered it up it played perfectly but after some minutes it showed an issue where some parts of graphics had wrong colors (pinkish, I’d say) :

Since I have repaired other similar Konami boards with same issue, I already knew where start to look at.Palette RAMs are two 2018 devices @13D and 14D:

I probed them with my oscilloscope but didn’t notice anything abnormal (though they were hot at touch), also piggybacking them didn’t change anything.So, I went more ahead in the palette circuitry following the path of the data bits of these RAMs.Just before the generation of the three RGB colors there are some open-collector buffers (74LS07):

Probing the one @8C with my scope, I found discrepancies between input pin 3 and output pin 4:

Once desoldered, the IC failed on pin 4 indeed:

Comparing on a tracer the bad output with a good one, you can see how the internal junction was altered (good one on the left of the below picture):

Fitted a good IC fixed the colors.

Success!

Got this faulty Knuckle Bash PCB in a trade with my friend Ifog:

For the uninitiated, it’s a side-scrolling beat ’em up developed and published by Toaplan in Japan in the 1993.

As expected, on the power up I was greeted by this scenario:

Game was playing almost blind, all graphics were totally wrong replaced by colored blocks.All the GFX DATA are stored in four 42 PIN 16Mbit MASK ROMs:

These devices are addressed by the near custom ASIC ‘GP9001’ @U13 which is the graphics processor unit of the system:

Also schematics confirmed this:

When I went to probe address lines of the MASK EPROMs, I found that most of them were silent.A closer inspection revealed the many pins were lifted.Pressing the custom chip with a clamp (a rude method perhaps but always effective) restored all the GFX :

So, I did a reflow of the custom chip passing the iron tip on its pins (with the clamp still in place) and this fix GFX stably.

Last thing to do was replacing some leaking 10mF axial electrolytic capacitors in the audio section (luckily corrosion didn’t involve any traces ) :

Another great game preserved to eternity (well, hopefully…)!

P.S.

This board turned to be an undumped Korean revision.I sent dumps to MAME developers.

Got this original Data East Rocobop for cheap:

Board was marked as not working but when I fired it up all was fine except for sound what was completely missing.Swapping boards with another working set I had , I could pinpoint fault in the top CPU board.Analog section was fine so fault was of digital nature.Sound system is made of a main 6502 CPU, a YM3812 and YM2203 sound synthesis chips:

Probing the 6502 CPU I found that it was undergone an hardware interrupt since its NMI line (pin 6) was low.Technically speaking an interrupt is an asynchronous signal generated to CPU from external peripherals indicating an event that needs immediate attention.

This lead me to think that the YM3812 or the YM2203 were bad (the IRQ line – pin 4-  of the 6502 is connected to both these devices) so I desoldered them.But I was wrong since they were good.Analyzyng with a logic probe  the TMM2015 SRAM @2F connected to the 6502 CPU revealed nothing abnormal but this didn’t convince me so I used my analog oscilloscope and found weak signal on some data lines compared to healthy ones:

I piggybacked the RAM and sound was fully restored.Time to desoder and test the IC out of circuit in where it failed:

Fitted a new RAM and archivied also this board as fixed.