Konami Cobra Hardware Troubleshooting and Discussion (Fighting Bujutsu: Wu-Shu)

[twistedsymphony]

Since there is seems to be a few people who own this hardware and it's rarely functional, I figured I'd create a thread where we can all collaborate and discuss progress and ideas on getting these boards running again.

Some great disassembly photos here: https://www.arcade-projects.com/threads/konami-cobra-system-disassembly-for-archive.32988/
Scan of the Operators Manual: https://www.arcade-museum.com/manuals-videogames/F/Fighting-WuShu--Fighting-Bujutsu.pdf
Alternative Scan of Manual: https://mega.nz/file/4xYkHLwD#sDc244CdhZs90d_x3iC8DX4Nq9Jz-9WJmyTWcg6PfLg
MAME Driver: https://github.com/mamedev/mame/blob/master/src/mame/konami/cobra.cpp
Listing on System 16: https://www.system16.com/hardware.php?id=580
[Any other good resources we can add?]

Known Owners:
708089 - @awbacon1
986015 - @chacal231077
@ekorz
Phil Bennett
980115 - @twistedsymphony
[Any other known owners out there?]

 

[ARIKAHistorian]

So it’s a bunch of boards and sub boards stacked on top of each other right? 1 small misalignment can trigger the E1 system error. Thermal pads might help the system is cooking itself and the temperature check is wrong hmm… nylon nuts and bolts might address the leaf springs problem.

 

[awbacon1]

Since there is seems to be a few people who own this hardware and it's rarely functional, I figured I'd create a thread where we can all collaborate and discuss progress and ideas on getting these boards running again.

@ekorz
@awbacon1
@chacal231077
Any other owners out there?

They are all sadly just large doorstops

I am dealing with E1 error which seems to mean "shit could be anything. Good luck kid"

Attempts so far: cleaned the ever loving hell out of every connector, rebuild the PCB from parts back to its stack and checked with a scope to see if there was any meaningful activity that would read as "its trying to post" on power on...nothing but garbage

Took out all socketed bios / roms and checked against known good MAME set and they checksum / data match to the good dumps so there is no issue on the data side

Switched SCSI HDD to solid state and flashed image

it remains a door stop

 

[awbacon1]

Since there is seems to be a few people who own this hardware and it's rarely functional, I figured I'd create a thread where we can all collaborate and discuss progress and ideas on getting these boards running again.

@ekorz
@awbacon1
@chacal231077
Any other owners out there?

the lid on your PCB looks like the one I have and its the most discolored over the IBM boards. Heat + metal = rainbow uh ohs lol

 

[twistedsymphony]

So it’s a bunch of boards and sub boards stacked on top of each other right?

No, most of it is inter-board connectors and harnesses.

1 small misalignment can trigger the E1 system error.

also no, we don't know what triggers the E1 error but the assumption is that the main CPU board is dead and the root cause.

Thermal pads might help the system is cooking itself and the temperature check is wrong hmm…

I wouldn't recommend thermal pads for this. traditional thermal paste would work much better for such a high temp system., pads are ok if you're just looking to advance passive cooling but you really need to use paste for systems generating a lot of heat like this one.

nylon nuts and bolts might address the leaf springs problem.

the spring tentioners clip in on brass studs mounted to the chips, they're also about 4mm wide to fit between the slats on the heat syncs.
I'm not sure where or how nylon nuts and bolts fit into this picture

the lid on your PCB looks like the one I have and its the most discolored over the IBM boards. Heat + metal = rainbow uh ohs lol

yup. I'm fairly confident that the reason none of these boards work is the IBM chips have overheated to the point of destruction.

 

[twistedsymphony]

Here's a video I shot when I first got my board in ... you can see what I'm talking about with the loose heatyncs:

View: https://youtu.be/6qOzgN0wyNg


the little silver bars are the spring clamps meant to keep pressure between the heatsync and the chip and they've all lost their ability to keep pressure.

 

[ekorz]

between parts I had and @Phil Bennett had, Phil did have one boardset working. He also wrote a custom bios to help diagnose some of the ram IIRC. I will have to check but I mailed most of what was left (and a second system I found) to @awbacon1 and I'm not sure I have anything useful left at this point!

 

[ARIKAHistorian]

No, most of it is inter-board connectors and harnesses.

also no, we don't know what triggers the E1 error but the assumption is that the main CPU board is dead and the root cause.

I wouldn't recommend thermal pads for this. traditional thermal paste would work much better for such a high temp system., pads are ok if you're just looking to advance passive cooling but you really need to use paste for systems generating a lot of heat like this one.


the spring tentioners clip in on brass studs mounted to the chips, they're also about 4mm wide to fit between the slats on the heat syncs.
I'm not sure where or how nylon nuts and bolts fit into this picture


yup. I'm fairly confident that the reason none of these boards work is the IBM chips have overheated to the point of destruction.

Can’t you dial in the exact amount of tension you need to make contact with the chip?

 

[twistedsymphony]

Can’t you dial in the exact amount of tension you need to make contact with the chip?

no, attempting to add some additional bend to the spring clamp causes it to fold like a piece of aluminum foil. They're probably heat stressed to the point of the metal fatiguing.

 

[ARIKAHistorian]

I see. And the heat causes everything to expand when in use and contract when not in use.

 

[twistedsymphony]

And the heat causes everything to expand when in use and contract when not in use.

technically yes but but practically no.
spring steel expands at 6.5x10^-6 inches per deg F
that means if we've got a 0.5mm thick spring clamp raising the temp by 100 deg would make it 0.00065 in thicker which is smaller than you can even measure with a set of calipers.

So yes Physics says it will expand, but for our "practical" purposes there is no expansion.

 

[awbacon1]

Here's a video I shot when I first got my board in ... you can see what I'm talking about with the loose heatyncs:

View: https://youtu.be/6qOzgN0wyNg


the little silver bars are the spring clamps meant to keep pressure between the heatsync and the chip and they've all lost their ability to keep pressure.

The board set ekorz sent does not have this issue. Heat sinks are all making solid mechanical contact to the IBM chips. That being said you can see the board flex. I have a feeling some internal traces are severed on the GPU board

Ive gone over every board closely looking for physical damage, scuffs, severed traces, etc etc etc and have found nothing. Granted its a needle in a haystack search considering how many boards there are. Even one shot component could cause a non-boot situation

 

[awbacon1]

I never did get that bios if you still have it

 

[ekorz]

I never did get that bios if you still have it

i don't, but it wasn't a general purpose tool that anyone could use. it was targeted at whatever issue(s) our specific boards were having at the time. maybe Phil can clarify but I always got the impression he was basically writing code for X purpose, burning it to a new bios chip, and running it on the hardware to diagnose a specific chipset on that specific boot. Then the LED display would be used as a means of communication for the issue he was testing.

 

[awbacon1]

i don't, but it wasn't a general purpose tool that anyone could use. it was targeted at whatever issue(s) our specific boards were having at the time. maybe Phil can clarify but I always got the impression he was basically writing code for X purpose, burning it to a new bios chip, and running it on the hardware to diagnose a specific chipset on that specific boot. Then the LED display would be used as a means of communication for the issue he was testing.

I was working with a dude who was RE'ing the boot sequencing in MAME debugger to try and get it to spit out some meaningful information about the error messages. More on an FPGA dev but ill touch base again

I hate Konami error codes. They truly mean nothing unless you MADE the board in Konami in the 90s

 

[ARIKAHistorian]

Just curious are you all using the original arcade power supply?

 

[awbacon1]

Just curious are you all using the original arcade power supply?

yep. Checked all power rails too and voltage in. Its the GPU board on the board set I have here

 

[ARIKAHistorian]

Hmm I’m trying to peruse Phil’s list of pcb repairs on his site but cannot find documents for fighting wu-shu and you said he got it working? When exactly was that I just ask because I’m trying to look into his entries by date. I mean I found the YouTube video he made on it.

 

[chacal231077]

I completely missed the creation of this post… My God!

I’m reposting the conversation I had with TwistedSymphony almost a year ago.

I think this idea of a rigid frame is essential. Unfortunately, I don’t have the 3D modeling skills required to design such a project.
If someone with experience in this field is interested, I would really like to see what could be achieved.

My idea may be a little crazy, but we need to find a solution to overcome the problem… Unless we design a completely new graphics card with a modern architecture and ask TSMC to manufacture the chips!

I also had another, even crazier idea: creating an exact imprint of the entire surface of the chips in order to design a custom monoblock heatsink, specifically engineered for optimal cooling in a push-pull configuration, using the two case fans.

I’m not saying these ideas are necessarily practical or economically viable, but they might be worth exploring. After all, some of the most unlikely ideas are sometimes where the best innovations come from.

 

[ARIKAHistorian]

Ok so a couple thoughts occur to me. If the architecture utilizes some form of IBM tech are there similar tech that can be found that have a similar type of technology to it? If so you could look at the internals of those potentially? Is the actual ide spinning when powered up? Essentially this issue is because Konami wanted to compete with Sega model 3 that they over complicated the design or?