What's new
can your programmer test sram?
Dont currently have one. It’s my next planned purchase. There is some good advice in other threads about what is cost effective and functional that I’m reading now. I’m currently free-of-employment so my budget is small :-/
 
well the TL866-II+ is best for the money,
but as you dont have one it's irelevent.

for other people reading though, it's faster to test a neo by just pulling all the smd ram off and testing it with a 28pin sop adapter.
 
Even faster to assume if it’s being reported bad, that it’s bad and just replace it.

Unless you’re getting changes in the error code which could be a bad trace or something else the RAM test on Neo is a fairly reliable indicator
 
Last edited:
bad ram report may not be the ram,
also you could buy some ram and it may be bad too.

i always test ram before i fit it to a board.
 
So my new plan is as follows
1) replace bios socket with new one with forger pins, confirm that bios is in round the right way and etc - complete and confirmed, every now and then it will boot to the test bios for a moment or two before the watchdog reboots it
2) reflow LSPC2-A2 and see if i can get the sync stable - sync is now stable most of the time
3) reflow the Neo D0 a d see if I can get stable clock - no improvement
4) use the microscope and multimeter to inspect for broken traces - spent about an hour on this
5) learn how to use the logic probe and start probing
6) order new backup ram chips, and fast video and etc just in case
Will take a more disciplined approach tomorrow and update.

I gave this guy about an hour under the microscope yesterday, being massive was a bit of a pain with my tiny scope so I had to mark it off into quadrants
IMG_8545.jpeg

The bottom of the mainboard did have some damage, but nothing major and what looks like a little bit of track rot
IMG_8550.jpegIMG_8549.jpegIMG_8548.jpeg

But these were interesting - they look like very light scratches, but under heavy magnification they are actually tiny cracks. When I probe them for continuity they sound a bit varable.
IMG_8547.jpegIMG_8546.jpeg

I need the practise with trace repair anyway - so I'll go ahead and fix all of this, just with bodge wires for a start, to see if that improves the situaiton - if it does I'll try my hand at "proper" trace repair
 
Well, I spent several hours on this tonight randomly trying “stuff” but because I didn’t have a plan I did not accomplish very much.

Still watchdoging, and unstable sync

Very occasionally I could get a stable picture out of it. Normally those stable pictures complained of the same backup ram error I had before.

So my new plan is as follows
1) replace bios socket with new one with forger pins, confirm that bios is in round the right way and etc - complete and confirmed, every now and then it will boot to the test bios for a moment or two before the watchdog reboots it

2) reflow LSPC2-A2 and see if i can get the sync stable

3) reflow the Neo D0 a d see if I can get stable clock

4) use the microscope and multimeter to inspect for broken traces

5) learn how to use the logic probe and start probing

6) order new backup ram chips, and fast video and etc just in case

Will take a more disciplined approach tomorrow and update.
Good plan.
Check the pins on the NEO-B0 too. IIRC it provides the sync signal and may cause the issues you describe if it's unstable.
You can do a quick check for cold solder joints by pressing hard on the chip to check. Remember to remove the PCB legs and place it on a smooth surface though -- don't flex it!
 
i would actually suspect any crystals or oscillators first.
i'v had to replace a number of the 4pin oscillators on various boards.
 
OK, so mystery solved as to why it suddenly stopped working.

It stopped working because I cooked it with too much voltage. My supplier selection for variable pots was off and instead of my normal Bourns trimpots I fitted one power supply with an off brand. The Bourns pots have a detent in the middle which roughly coresponds to a safe +5.00v. The cheaper ones are dont have a detent so it's a bit of a guess and I guessed wrong - also being drunk at the time probably did not improve the situation

So in this particular case the board was cooked by approx 5.5v which (as I understand it) is going to cook most of the RAM chips first, but judging from the behavior has destroyed at least one of the customs too

Here is some power supply pron - now with the crappy pot replaced with a nice Bourns one.
IMG_8600.jpegIMG_8599.jpegIMG_8596.jpeg
So now I need to find ANOTHER $20 NeoGeo MVS to use as a donor for the previous MVS
 
I’ve got…. a few…. MVS boards that can spare their customs. Which ones do you need?

The fast video SRAMs are the ones most likely shot by too high voltage (have seen this a lot) - unusual for customs to die though, 5.5v isn’t that high (other than for RAM)
 
I’ve got…. a few…. MVS boards that can spare their customs. Which ones do you need?

The fast video SRAMs are the ones most likely shot by too high voltage (have seen this a lot) - unusual for customs to die though, 5.5v isn’t that high (other than for RAM)
Thank you for your kind offer! At this stage I’m not sure. I’ve got a very bad feeling about Neo D0 (clock) and LSPC 2A (sync). I haven’t actually checked the spec sheets, if they exist, yet to find out how likely 5.5v is to do and damageto those.
I’ll start with the ram because it is comparatively easy to test and cheap to replace and work my way up from there.
I may soon be begging for parts :)
 
If anyone else has a similar problem in future this might be useful

ram shopping list from here
https://wiki.neogeodev.org/index.php?title=RAM_chip
chip locations from here
https://github.com/jwestfall69/neogeo-diag-bios/blob/master/docs/ram.md

These are all readily available.
32k vram upper = CXK58257AM-10L, uPD43256AGU-12L
Backup ram = CXK58257AM-10L, uPD43256AGU-12L (fact check)
Work ram = CXK58257AM-10L or uPD43256AGU-12L
https://www.digikey.co.nz/en/produc...LI-TR/706-1310-1-ND/4733140?itemSeq=333404651

And replacing this guy seems to be highly regarded, and is easily available too
74HC32 = https://www.digikey.co.nz/en/produc...conductor-and-storage/TC74HC32AF-EL-F/6679562

The Fast Vram is not so easy to get - I'll probably borrow this off one of my MV1's for troubleshooting
2k Fast vram = CXK5814P-35L
https://www.retrosales.com.au/colle...l-neo-geo-aes-mvs-fast-vram-yellow-screen-fix

The Palette Ram is not so easy to get either - I'll probably borrow this off one of my MV1's for troubleshooting
Palette Ram = CXK5864BSP-10L or 12L
https://www.retrosales.com.au/colle...neo-geo-aes-mvs-palette-ram-colour-screen-fix

And the Z80 ram is not so easy to get either - but I'll deal with that later
Z80 ram = BR6216B-10LL or HM6116ASP-20

Sooooo, looks like I can get all the parts I need for testing without too much fuss - If anyone has a good supplier for these guys long term it would be great to know too!
2k Fast vram = CXK5814P-35L
Palette Ram = CXK5864BSP-10L or 12L
 
Fast video RAM and sound RAM are both 6116 type SRAM. Don’t worry about getting Sony specifically (I’d argue you should avoid it)

Palette RAM is 6264 type

On some boards you can use 6264 SRAM for fast video RAM as well
 
Fast video RAM and sound RAM are both 6116 type SRAM. Don’t worry about getting Sony specifically (I’d argue you should avoid it)

Palette RAM is 6264 type

On some boards you can use 6264 SRAM for fast video RAM as well

Brilliant, thanks!

So for Palette ram it looks like I can use something like this - this is "wide" but the MV4-FT2 has tracks and holes for both wide and narrow
2 x https://www.digikey.co.nz/en/products/detail/alliance-memory-inc/AS6C6264-55PCN/4234595

The video ram on this the board can only take "narrow" type so something like this would do it - this is actually on "last sale" at digikey, June 2022 is the last purchase date
2 x https://www.digikey.co.nz/en/products/detail/renesas-electronics-america-inc/6116SA15TPG/2008361

And the upper vram, backup ram and work ram are as before
These are all readily available.
32k vram upper = CXK58257AM-10L, uPD43256AGU-12L
Backup ram = CXK58257AM-10L, uPD43256AGU-12L (fact check)
Work ram = CXK58257AM-10L or uPD43256AGU-12L
6 x https://www.digikey.co.nz/en/produc...LI-TR/706-1310-1-ND/4733140?itemSeq=333404651

And replacing the backup control guy seems to be highly regarded, and is easily available too (restricted stock)
1 x 74HC32 = https://www.digikey.co.nz/en/produc...conductor-and-storage/TC74HC32AF-EL-F/6679562
 
I've fixed worse... Replace the bad Backup RAM and try it out.

Reading this is giving me frustration! Backup RAM is tied to the main CPU data and address busses. If the output is bunk you'll end up with bunk on the CPU data bus and it'll have a hard time starting up. Replace that bad Backup RAM and try the diagnostic BIOS.
 
When you find yourself in a hole, stop digging - that was the good advice that I did not take...

I needed parts for the MV4 repair so I bought this MVS-MV2FS for $20 - seller was very clear. "Does not work, parts only". In hindsight this was kind of a stupid thing for me to do, But it turned out OK.

Got it home and powered it up just for fun before I started stripping it for parts, and it worked

It didnt work very consistently; sometimes it would boot and sometimes it would not, sometimes the sound would work and sometimes it wouldnt
So today turned into MVS-MV2FS repair day instead!
IMG_9362.jpeg
It also had these strange graphical glitches. At first I thought they were "cart slot needs a clean" glitches. But they were not.

And then I opened it up and found this horror show
IMG_9374.jpeg
As @stj mentioned earlier, it's possible for this kind of damage to occur even when the battery is apparently ok. The battery was sparkling and imaculate - everything around it was an acid-ravaged wasteland.

At this point I was starting to doubt the wisdom of continuing with the repair so I didnt spend much time cleaning, but I did beep out all the traces to see where the problems were. The back of the board has some track rot that I decided to deal with later when I knew if it was repairable or not. The front had two broken traces coming out of ZM02 which does clock stuff. The graphical glitches looked very "bad clock" so I ran some bodge wires back to where they belonged on the 257
IMG_9385.jpegIMG_9405.jpegIMG_9407.jpeg

Then I fired it up, smug and certain that I had fixed the problem

Aaaaaand nothing - would not even boot. Then I left it alone for half an hour and it did boot to jailbars, then I moved the carts around and it did a slow watchdog, then I moved them around again and it worked perfectly but with bad sound.

Intermittent weirdness normally = bad physical joint somewhere

The Universe Bios was consistent in not working - but the Diagnostic Bios kept giving bad backup memory errors, probably more likley to be bad addressing for the backup memory (thanks @xodaraP!) - so I started poking around HC32 which does lots of address stuff - thanks to @Raph_friend and @Kujako for the tips in this thread https://www.arcade-projects.com/threads/mv1c-backup-ram-error.13969/
IMG_9410.jpeg

Eventually I traced the fault to the little 22k resistor that sits between the HC32 and +12v - on the outside it looked perfect, but the legs on the resistor had corroded away - so the resistor was only resisting intermittently, HC32 was only addressing intermittently, and that was my intermittent weirdness for backup ram, sound, and boot
IMG_9412.jpegIMG_9413.jpeg

So I salvaged a 22k resistor off the MV4 board (ironically) and now the MV2 works perfectly. Graphics Glitches cleared, Sound works, boots every time, all diagnostic tests passed
IMG_9414.jpeg

Now I just have to go back, wash with vinegar to arrest the corrosion, clean up all the track rot, remove the bodge wires and re-build the traces and generally tidy up. I've been practising that on old waste boards, still got a way to go...
IMG_9398.jpeg

Success! Thank you to everyone for sharing your knowledge and subject matter expertise. I was really thankful to everyone that you had taught me what I needed to be able to complete the MV2 repair myself without having to ask a single queston.

Still no progress on the MV4... I'll get to work on that again when I get another source of spare parts :)
 
When you find yourself in a hole, stop digging - that was the good advice that I did not take...

I needed parts for the MV4 repair so I bought this MVS-MV2FS for $20 - seller was very clear. "Does not work, parts only". In hindsight this was kind of a stupid thing for me to do, But it turned out OK.

Got it home and powered it up just for fun before I started stripping it for parts, and it worked

It didnt work very consistently; sometimes it would boot and sometimes it would not, sometimes the sound would work and sometimes it wouldnt
So today turned into MVS-MV2FS repair day instead!
IMG_9362.jpeg
It also had these strange graphical glitches. At first I thought they were "cart slot needs a clean" glitches. But they were not.

And then I opened it up and found this horror show
IMG_9374.jpeg
As @stj mentioned earlier, it's possible for this kind of damage to occur even when the battery is apparently ok. The battery was sparkling and imaculate - everything around it was an acid-ravaged wasteland.

At this point I was starting to doubt the wisdom of continuing with the repair so I didnt spend much time cleaning, but I did beep out all the traces to see where the problems were. The back of the board has some track rot that I decided to deal with later when I knew if it was repairable or not. The front had two broken traces coming out of ZM02 which does clock stuff. The graphical glitches looked very "bad clock" so I ran some bodge wires back to where they belonged on the 257
IMG_9385.jpegIMG_9405.jpegIMG_9407.jpeg

Then I fired it up, smug and certain that I had fixed the problem

Aaaaaand nothing - would not even boot. Then I left it alone for half an hour and it did boot to jailbars, then I moved the carts around and it did a slow watchdog, then I moved them around again and it worked perfectly but with bad sound.

Intermittent weirdness normally = bad physical joint somewhere

The Universe Bios was consistent in not working - but the Diagnostic Bios kept giving bad backup memory errors, probably more likley to be bad addressing for the backup memory (thanks @xodaraP!) - so I started poking around HC32 which does lots of address stuff - thanks to @Raph_friend and @Kujako for the tips in this thread https://www.arcade-projects.com/threads/mv1c-backup-ram-error.13969/
IMG_9410.jpeg

Eventually I traced the fault to the little 22k resistor that sits between the HC32 and +12v - on the outside it looked perfect, but the legs on the resistor had corroded away - so the resistor was only resisting intermittently, HC32 was only addressing intermittently, and that was my intermittent weirdness for backup ram, sound, and boot
IMG_9412.jpegIMG_9413.jpeg

So I salvaged a 22k resistor off the MV4 board (ironically) and now the MV2 works perfectly. Graphics Glitches cleared, Sound works, boots every time, all diagnostic tests passed
IMG_9414.jpeg

Now I just have to go back, wash with vinegar to arrest the corrosion, clean up all the track rot, remove the bodge wires and re-build the traces and generally tidy up. I've been practising that on old waste boards, still got a way to go...
IMG_9398.jpeg

Success! Thank you to everyone for sharing your knowledge and subject matter expertise. I was really thankful to everyone that you had taught me what I needed to be able to complete the MV2 repair myself without having to ask a single queston.

Still no progress on the MV4... I'll get to work on that again when I get another source of spare parts :)
That's great work. I repaired so many arcade PCBs that i know already most of issues, so i need to do just some small measureaments on most of cases.
I'm glad that my tips did help!
 
Nice! The 2 slot boards are great with everything on a single PCB. Glad to see you were able to get it up and running!

Battery damage sucks. If you decide you want a battery on the board I recommend installing a coin cell holder so the battery can be removed if the board is put into long term storage. You can get rechargeable coin cells if you want to retain the charging circuit
 
I finished up the trace repair on the MV2FS today - this was mainly just for fun and practise so I did a mix of single side repair with .5mm un-tinned copper, pad to pad through hole repair with .25mm pre-tinned copper, and bodge wires with .25mm and .13mm enamel wire.

I dont know if "track rot" is a technical term, but I call it track rot, the trace still exists and still works, but it has started rusting from the inside. For those ones I scrape off the solder mask, then scrape back all of the corrosion until I got to shiney copper. Then I tinned over the shiney copper with solder braid+flux. Then I stuck the new .25mm tinned wire onto the top.

This works well for rotten tracks and minor breaks and looks really nice - these ones are really zoomed in, but with the naked eye they are basically invisible. This is incredibly slow.
IMG_9456.jpeg

Some of the traces were just plain gone, for those ones where there is nothing to repair I used .25mm enamel wire (magnet winding wire). The best way I found to remove the enamel was to have the iron at approx 380, melt some solder onto the end, burn off the enamel with the melted solder, then tin the exposed end. This works really well but doesnt look nearly as nice, the repair is very obvious. This is much faster
IMG_9454.jpeg

For some of the tracks the rot had gone through the via and out the other side, In those cases I drilled the via, scraped back some track on the back to good copper, ran the wire all the way through the via hole and soldered to the good copper on the trace on the back, then I plugged the via was with solder for mechanical strain relief.

IMG_9454.jpeg

After these pics were taken I covered the whole lot with blue nail polish - green would have hidden the repairs much better, but blue is easy to spot and nail polish is easy to remove with acetone if I needed to rework

This is how it looked when it was done. There were probably about 30 tracks in total that needed repair, and by the time I had soaked, scraped, prepped, tinned, repaired, cleaned and re-painted that was 8 hours of work.
IMG_9451.jpeg

If it were an absurdly expensive board I'd probably do the whole lot with the "trace rebuild" method, I still need some practise but I was surprised how well this works and how good it looks.

If it were a cheaper board I would probably do the whole lot with enamel bodge wires. This is much faster but does not look nearly as good.

So the MV2FS is now "done" - back to the 4-slot
 

Attachments

  • IMG_9455.jpeg
    IMG_9455.jpeg
    262.3 KB · Views: 95
So the smart thing to do would have been to buy new and tested parts from digikey.

But the fun thing to do was to buy a non-working MV4F for $NZ20, approx $US12. Replacing untrusted and potentially broken parts with new untrusted and potentially broken parts is a reciepe for frustration, but I did it anyway.

So to recap
The existing MV4-FT2 is doing a fast watchdog, probably due to bad ram, which I cooked through over-voltage. Unfortunately I have no RAM tester so I'm swapping randomly and hoping for the best.
The new MV4F is doing a fast watchdog for reasons unknown

The whole point of this excercise was to swap the ram fron the MV4F to the MV4-FT2. So I removed the fast ram and palette RAM, added some cheap sockets, and swapped them over.

When I did this the MV4-FT2 showed no improvement at all, but the NV4F booted to jailbars

For no very good reason I soldered a 40 pin ZIF socket on to make it easy to swap bios.
IMG_9958.jpeg

Then I got "hardware test failed z80" on the stock bios
IMG_9965.jpeg

The universe bios lets you skip error tests, so with the error tests skipped games played perfectly (but with no sound)
IMG_9960.jpeg

The diagnostic bios showed all tests passed, but by default it does not test the sound system - for that you need the companion z80 test cart, which I do have. THe z80 test cart showed "Z80 Dead / Errored / Comm issue". You can boot the z80 test cart from different slots and get slightly different error messages, but all of them are pointing to a bad Z80 or a problem in the comm path
IMG_9972.jpegIMG_9971.jpegIMG_9970.jpeg

Given the way that this is failing I'm suspecting a bad z80, so clearly the sensible thing to do it swap it for another one :)

Incidently the documentation for the NeoGeo diagnostic bios is a bit sparse - this project aims to document it fully
https://github.com/jwestfall69/neogeo-diag-bios
 
z80's fail a lot compared to other cpu's -is it even the right one?
i'v seen a number of games overclocking slow ones.
cps1 and bubblebobble being 2 examples
 
Back
Top