AES black screen sync issues (click of death?)

[Kujako]

Trying to bring an AES back to life, symptom is a black screen regardless of cart etc. Running it through my Medusa, I see it struggling to figure out what the video signal is.

Video shows the Medusa. Here is what it looks like with an identical but working AES system.


Can't spot any visible issues on the main board, everything is super clean. Not even any dust in the case.


Any thoughts as to where I should start? Could just recap everything, but that's sort of a shotgun approach.

Edit: possible "click of death"...

 

[Ryoandr]

Does game actually launch ? Do you have sound, through either AV Out or headphones jack ? Did you try on composite cable, on a crt ?
I have found that some MVS and pretty much all AES I've tested tend to have sync signals that throw off modern hardware, displays, scalers, etc.

 

[Kujako]

Does game actually launch ? Do you have sound, through either AV Out or headphones jack ? Did you try on composite cable, on a crt ?
I have found that some MVS and pretty much all AES I've tested tend to have sync signals that throw off modern hardware, displays, scalers, etc.

I have not tried composite but have with RGB, this is a known faulty unit and the Medusa is very good at figuring out the correct input information. But as the video shows it can't lock onto the signal.

RGB to CRT also gives a black screen... BUT, and I hadn't noticed this before. There is a clicking akin to the "click of death" on a MVS system.

 

[Kujako]

Down to tracing pins from the BIOS/SP1 under the assumption that it's the watch dog / "click of death". 15 down, 25 to go...

Brain is failing me now, so calling it a night. Hopefully someone can narrow this down for me a bit. Otherwise I'll keep tracing and if I fail to find a fault I'll look at pulling the BIOS and replacing it with the UniBIOS/NeoDIAG.

 

[Kujako]

Finished with the traces, can't find a fault (don't think I missed any). So, kinda stumped. May be time to try socketing the BIOS and seeing if that provides more information.

 

[Kujako]

UniBIOS installed, has the same issue. So the failure is elsewhere. Kind of out of ideas...

Still black screen with clicking...

 

[Ryoandr]

That fast click means something early is bad. Do you have something to check activity, at least a logic probe, or an oscilloscope.
Main crystal (if faulty usually results in literally nothing, not even click of death) goes into D0 that divides for almost all clocks.
Some ICs might stick an address or data line up/down, and there are many on the lines :/ outside of SNK chips and memory, HC259 is a fairly common failure too
No video at all is kinda weird tho... LSPC2 failure ? Or the associated 74s

 

[Kujako]

That fast click means something early is bad. Do you have something to check activity, at least a logic probe, or an oscilloscope.
Main crystal (if faulty usually results in literally nothing, not even click of death) goes into D0 that divides for almost all clocks.
Some ICs might stick an address or data line up/down, and there are many on the lines :/ outside of SNK chips and memory, HC259 is a fairly common failure too
No video at all is kinda weird tho... LSPC2 failure ? Or the associated 74s

I've got all the tools (oscilloscope, logic probe, etc) but not sure what to check and where on an AES. Any direction would be greatly appreciated.

 

[Ryoandr]

https://wiki.neogeodev.org/index.php?title=NEO-D0
check all the purple pins, Xin / out is the master crystal, all the others are outgoing clocks. If all are pulsing check if each clock actually go to each chip needing it.
check 68k address / data lines activity, each (I think) should be pulsing.

 

[Kujako]

https://wiki.neogeodev.org/index.php?title=NEO-D0
check all the purple pins, Xin / out is the master crystal, all the others are outgoing clocks. If all are pulsing check if each clock actually go to each chip needing it.
check 68k address / data lines activity, each (I think) should be pulsing.

So, probing around the NEO-D0 I find the following. XIN and XOUT seem right, but I also see the reset pin pulsing in time with the boot loop.

Checking temperatures, not seeing anything too weird but then I'm not sure off hand what normal should be. Guess I can open my working system and compare. The Yamaha chip seems like it's warmer than it should be however (105F). No hot spots on the CPU or D0. Z80, Yamaha, LSPC and VRAM seem to be the only hot spots.

Based on the ~8hz reset cycle, I'm thinking bad work RAM, CPU or NEO-B1. But not sure how to confirm functionality of each.

 

[ShootTheCore]

Based on the ~8hz reset cycle, I'm thinking bad work RAM, CPU or NEO-B1. But not sure how to confirm functionality of each.

If you have an EPROM programmer, it'll probably test RAM chips as well. Most EPROM programs can test TTL and RAM chips in DIP packaging - but you'll have to desolder the RAM from the board.

The CPU is also DIP packaged and can also be desoldered and socketed for testing - a replacement CPU can be inexpensively ordered from Ebay.

Before you go any further on the RAM or CPU though, try swapping the UniBIOS EPROM out for the Neo Diagnostic BIOS - it'll run some in-depth tests and tell you what's going on.
http://smkdan.eludevisibility.org/neo/diag/

 

[Kujako]

If you have an EPROM programmer, it'll probably test RAM chips as well. Most EPROM programs can test TTL and RAM chips in DIP packaging - but you'll have to desolder the RAM from the board.

The CPU is also DIP packaged and can also be desoldered and socketed for testing - a replacement CPU can be inexpensively ordered from Ebay.

Before you go any further on the RAM or CPU though, try swapping the UniBIOS EPROM out for the Neo Diagnostic BIOS - it'll run some in-depth tests and tell you what's going on.
http://smkdan.eludevisibility.org/neo/diag/

Unable to find a 12mhz CPU anywhere, but I have some donor boards around. Testing the SRAM for me is more trouble than it's worth, would just replace the lo work RAM and see what's up with it. Can try the diagnostics BIOS, but not expecting it to work. System would seem to be boot looping before the BIOS can kick in.

 

[ShootTheCore]

Can try the diagnostics BIOS, but not expecting it to work. System would seem to be boot looping before the BIOS can kick in.

Try it and see - you might be surprised. Based on your symptoms, I think bad Work RAM is very likely.

 

[Kujako]

Try it and see - you might be surprised. Based on your symptoms, I think bad Work RAM is very likely.

Tried it once I found my big box o' BIOS's and same symptoms, black screen and ~8hrz clicking. Swapping out the LO work RAM will be a pain, there's hot glue on it from the adjacent coil and a lot of components to shield from the hot air, but will consider doing it this weekend. Somewhere around here I have some replacements, but otherwise may have to pull it from a donor board and hope it's good.

 

[Kujako]

Try it and see - you might be surprised. Based on your symptoms, I think bad Work RAM is very likely.

Swapped out the lower work RAM, no change. My understanding is that it's not likely to be the upper preventing the system from booting.

Granted, I don't know for sure that the replacement RAM module is good. But I would at least expect the issues to change. Not my best work, but all the pins seem secure.

Pulled the CPU from the donor board as well as a couple 43256 RAM chips. But that's a pain to desolder... not going to try swapping the CPU on a whim, at least not yet.

 

[kazuo]

You can remove the hot glue pretty easily by soaking it with IPA.

I'd lean toward component (IC) failure since you didn't find anything wrong buzzing out the traces. These boards are super fragile and often experience trace corrosion, especially around the BIOS and the LSPC2. VRAM failure is also common from what I've read... does bad VRAM trigger watchdog though?

 

[Kujako]

You can remove the hot glue pretty easily by soaking it with IPA.

I'd lean toward component (IC) failure since you didn't find anything wrong buzzing out the traces. These boards are super fragile and often experience trace corrosion, especially around the BIOS and the LSPC2. VRAM failure is also common from what I've read... does bad VRAM trigger watchdog though?

My understanding is that it would only be lower work RAM prior to the BIOS engaging. If it were VRAM etc I should be able to get into the diagnostic BIOS. But I am often wrong about such things... from what I've read, if it's not lower work RAM (which I replaced) it may be a bad CPU. I'm trying to see if I can find any tests to confirm CPU functionality in this case, since desoldering the 64 pin 68k is a PITA. I did already pull the CPU from the donor board, so if I can find a 64 pin socket I'll probably try that next (guess it doesn't really need a socket, but I try to socket anything I remove these days).

 

[Kujako]

Swapped CPUs, and the issue persists. Kind of out of ideas...

 

[ShootTheCore]

What revision is the AES board? If it's at least a Rev 3-4, the watchdog is driven by the NEO-B1 chip and you can disable it by tying Pin 94 to Ground.
https://wiki.neogeodev.org/index.php?title=NEO-B1

Try disabling the watchdog and see if the Diagnostic BIOS runs long enough to show you something. The NEO-B1 will trigger the watchdog (and cause Click of Death symptoms) if it can't be written to by the CPU, so give it a good inspection.

 

[Kujako]

What revision is the AES board? If it's at least a Rev 3-4, the watchdog is driven by the NEO-B1 chip and you can disable it by tying Pin 94 to Ground.
https://wiki.neogeodev.org/index.php?title=NEO-B1

Try disabling the watchdog and see if the Diagnostic BIOS runs long enough to show you something. The NEO-B1 will trigger the watchdog (and cause Click of Death symptoms) if it can't be written to by the CPU, so give it a good inspection.

It's a 3-6, so I'll give this a try.

Edit: so... I swear I inspected everything, but look at the mess I found on the NEO-B1. Not sure how I missed this, but I think we have a culprit.

After cleaning things up a bit, it looks like those pads are just gone.

Pins 115 and 116. A20 and A21 respectively on the 68k address bus. Not sure where the traces should go to... possibly direct to the 68k pins?

Will trace on a good board and see where they go. Annoyed with myself since the 68k address lines are specifically listed as causing the click of death. Ah well, it’s a learning experience.