Neo Geo Static Sprite corruption, maybe S-ROM related

Hi

First post here, having just got myself an AES system from Japan, I'm having some odd issues with it.

I've made a video to help demonstrate the issue I'm having. Essentially though the copyright symbol, insert coin, text, health bars all look wrong. I have checked how they should look by watching some longplay youtube videos. The region warning should have a black background, the insert coin text should be red I think, and the health bars should be green.

Mine either look very very bright, or completely white. I have done some research and found that my issues could be related to the S-ROM as this is the part of the system that handles static sprites (health bars, insert coin etc), which seems to be the issue I am having.

I have traced FIXD0 - FIXD7 as apparently these are related to S-ROM, I tested the continuity from the cart all the way to the NEO-B1 chip and it seems fine. But my knowledge pretty much runs out there.

Anyway here's my video (the region warning screen is white, but not as white as my video makes out, I think it was having issues with the exposure for that shot).

I have used the Neo Geo diagnostics bios and it reports no problems. I have the UniBios as well, however I cannot test it just yet as I am still waiting for my controller to be delivered. Also I have no other games to test with, so I cannot rule out that it is the game that is at fault.

The UniBIOS can only test the P ROM so I'm not sure how much help that'd be for you in this case.

This looks to me like the palette selection for the FIX layer is incorrect.

Ok, so just wanting to know what I can do to fix the issue. Is it likely to be the console as opposed to the cartridge?

Anyone got anything I could try in order to resolve this?

Well, what I would do is:
- Test with another cart or carts, to make sure the issue is the console or the cart.
- Make sure the problem lies only in the fixed layer. It seems so, but double checking does not hurt.
- If the console is definitely faulty, check all the traces responsible for the S chips of the cart.
- If traces ok, I'd worry because AFAIK the responsible chips for tha fixed layer processing are custom and Not easy to replace.

Regards.

Like mentioned, I would first try another game cart.

Ok, so I've tested my game on another AES and it does not present the same problems. The Insert Coin/Health Bar and other static text displays fine.

I have also tested a different game on my AES, Magician Lord, and this displays similar issues to Samuri Spirit 2. The text along the top of the screen looks wrong, as well as the life bars at the bottom etc.

So, it looks like my system is at fault. I think I have narrowed the problem down to the NEO-B1 https://wiki.neogeodev.org/index.php?title=NEO-B1.

These glitches also mean that I cannot see the UniBios text under the Neo Geo logo on bootup, and also I see no text when entering the UniBios menu by holding A+B+C.

I have attached some pics of my chip, and to me it looks like there is possibly some corrosion going on, what do you think?

While Samuri Spritis 2 was running I flexed the board slightly around the B1 chip, and I could get the colours to change, and if I was lucky I could get them correct.... if I then reset the console I could in fact see the UniBios text. As soon as I let go of the board it went back to its normal incorrect state.

If board flexing helps, I would reflow the B1 chip. As I stated above, the problem stems from palette selection when outputting FIX layer graphics. I recommend doing this with a soldering iron, flux, and a little solder (for surface contact) over a hot air solution, unless you have the correct tools for the job (a paint stripper gun is not the correct tool).

The bright yellow pins starting with PA are the palette address pins.

radiokid said:

Ok, so I've tested my game on another AES and it does not present the same problems. The Insert Coin/Health Bar and other static text displays fine.

I have also tested a different game on my AES, Magician Lord, and this displays similar issues to Samuri Spirit 2. The text along the top of the screen looks wrong, as well as the life bars at the bottom etc.

So, it looks like my system is at fault. I think I have narrowed the problem down to the NEO-B1 https://wiki.neogeodev.org/index.php?title=NEO-B1.

These glitches also mean that I cannot see the UniBios text under the Neo Geo logo on bootup, and also I see no text when entering the UniBios menu by holding A+B+C.

I have attached some pics of my chip, and to me it looks like there is possibly some corrosion going on, what do you think?

While Samuri Spritis 2 was running I flexed the board slightly around the B1 chip, and I could get the colours to change, and if I was lucky I could get them correct.... if I then reset the console I could in fact see the UniBios text. As soon as I let go of the board it went back to its normal incorrect state.

View attachment 38359

View attachment 38360

Click to expand...

I already see some solder bridges on the B1, reflow it.

So, reflowed solder on all pins (to the best of my ability). Tested for any bridged pins, which there are none of. Tested traces from the PA pins to the relevant pins on the two Palette Ram chips, all seemed to have continuity.

Still not working

What I did notice however is that it is not pushing or flexing the B1 chip (or its legs/solder pads) that seems to cause the colours to change (which I would describe as actually cycling through a load of colours). It seems to be when I am holding the underside of the B1 area and touching various solder points on the underside. It doesn't seem to be that consistent, for a while it would do it but now I can't seem to make it happen anymore.

Suffice to say this is confusing me a lot. Sometimes it seems I'm so close, and then it just goes back to how it was before.

Trying to get my head round this description, to see if there is something else I can try, but not sure I fully understand it.

"Both fix graphics from the S ROM and sprite graphics from the C ROM are fed to NEO-B1 for display on screen via 2 pairs of alternating line buffers. Sprite graphics come from the C ROM multiplexer (NEO-ZMC2, PRO-CT0 or NEO-CMC), while fix graphics come directly from the currently enabled fix ROM.

The chip outputs the palette RAM address to select colors for pixel output. The data output of the palette RAM is latched by a pair of 8bit registers, which in turn feed the video DAC. The NEO-B1 handles address bus switching between the 68k bus and palette RAM. Priority is always given to the 68k which results in harmless display glitches when games access palette RAM during active display."

I'd like to see a pic of the reflow.

Here's a quick pic I took. The work doesn't look that pretty, and initially the leg on the bottom right was shorting with its neighbor so I tried to fix it, at which point I realised it was suppose to. Aside from that one I have tested each leg to its neighbour with a multimeter and I detect no shorts. I'm leaning towards needing to remove the entire chip and cleaning up the legs and pads on the board, however this is no easy task.

You need to reflow it better. There's a lot of suspect pins on that. Use a lot of liquid flux.

Appreciate that I need to do a better job. What am I actually trying to achieve here? As I said I am not getting any shorts between pins, so is it just the contact between the pins and their relevant pads that I am trying to improve?

Right, so got some ChipQuik in the post today, going to desolder (carefully) the whole B1 chip. Then clean up the legs of the chip and the pads on the board. At this point I am also going to check for any bad traces around the B1 location. My plan is to then refit the chip, solder it back in using the drag solder technique and lots of flux.

This is a big job, and probably the most complex solder work I have done so far. What I would like to know however, is what exactly is going on here? (be as technical as you like). From searching everywhere on the web I haven't found a another issue precisely like mine; most are related to in game sprite corruption, or full screen garbled graphics.

Does my issue seem to anyone to be any of the following options?

a) Faulty B1 chip
b) Bad contact between pins and pads of the B1
c) Bad traces between the B1 and the two Palette RAM chips
d) Bad Palette RAM chips
e) Bad traces anywhere else

I was going to suggest reflowing the B1 again. Just looking at the first photo there I see what looks like solder partical on the top of the chip between 2 pins - is that your problem? If flexing the board fixes it then it has to be a bad connection somewhere around there. I would reflow the chip using the chipquik flux (but not the solder) as the flux is really good. Then clean it with IPA, brush gently over the connections to make sure all solder particals removed and inspect and retest. If its still glitching then I would be looking for bad solder points nearby on other chips. Inspection is really the key I think. I would certainly spent more time inspecting things before removing the B1.

EDIT: One final point - you have cleaned the cart slot as well to make sure you've not just got a bad connection from the S ROM on the cart? Or perhaps a questionable trace coming from the S ROM connections on the cart socket?

So, an update, pictures later.

Followed Gadgets advice and did not remove the chip, instead used the flux (it really is good) and reflowed all pins. Tested continuity, had a couple of shorts but a further reflow sorted them out. Switched on, but not quite there yet. I now see the UniBios text underneath the Neo Geo logo on bootup, however it is white not green. The text/health on Samuri is still white, but on Fatal 2 it is blue with the health bars being slightly off colour.

Now here is where it gets interesting. On the right of the B1 chip there are two ceramic capacitors both labelled 101, and are connected to pins PCK1 & PCK2 on the B1 chip; labelled as "Latch signals, shared with LSPC2-A2". Now, if I touch (not bend/flex) the lower of the two capacitors which goes to PCK1, the colour of the problem areas begin to cycle; almost like a flicker really, I'll take a video of this when I'm home from work.

Reading up on the PCK1 & 2 from here https://wiki.neogeodev.org/index.php?title=LSPC2-A2 it says that "LSPC2-A2 is the second generation Line SPrite Controller", "Generates SROM, CROM and LOROM addresses".

Given my "cycling colour" issue, I am wondering if my problem lies here, rather than the B1 chip sending the fix layer information to the palette ram chips.

Anyway, gonna look at this further when I get home, Got to about 12 last night and I had had enough. These Neo Geos really are on a different level compared to other retro systems. Put it this way, I'll be very happy when I sort this issue out. Got a Magic Key and MVS 161 in 1 on their way and want it running smoothly.

Ok, TLDR it was a bad trace and now it's fixed.

Here's what happened.

Firstly, here's a video of what was happening when I touched the underside of the capacitor solder point that goes to the pin labelled PCK1 on the B1 chip.


Now, I don't know exactly why my finger on that point was causing the colour cycling, I suspect I was changing the capacitance or introducing some kind of unknown factor into that particular address bus.

Apparently the PCK1 is a latch signal, it is shared with the LSPC2-A2 chip. But when you look at that chip, PK1 (which I assume is linked to PCK1, but I could be wrong) it is labelled "Clock to latch C ROM address from P bus".

So, I traced the trace from the leg of the capacitor to it's next point on the board (oh and I also replaced both the caps you see), and there was no continuity. At this point I'm thinking, surely its not this simple, I add a wire to join these two points back together, as shown in the picture below.



Then I powered her back up, and, well take a look at the video below.


So in the end it was a bad trace. Now, I've no idea whether re flowing the B1 chip helped matters, or whether it was a red herring. But whatever, it's done now. Looking back at the picture you can actually see the trace on the board does have some slight damage, but then again so does another trace near by and that doesn't seem to affect anything.

I'm just glad it's fixed! Thanks everyone for their input.

Ah, I hadn't seen that it was cycling. Yeah, that cycling looks typical of a floating input. The frequency of that cycling is often a beat frequency between the AES's VSync and the mains power frequency.

Touching around the board and seeing what changes what, or breathing on things, is a great diagnostic tool. Glad you found the bad trace and fixed it.

Nice one! Glad you fixed it!! =D

Just chiming in to say thanks for this thread. I don't have any current issues, but it is always great to watch people who know what they are doing troubleshoot. From radiokid's (OP) excellent communication of the problem, and the process he did for troubleshooting before he made a thread, to mikejmoffitt's excellent assistance and again, very clear communication. I wish that all problems could be solved this eloquently, and quickly. Kudos guys!

c0nn0r said:

Just chiming in to say thanks for this thread. I don't have any current issues, but it is always great to watch people who know what they are doing troubleshoot. From radiokid's (OP) excellent communication of the problem, and the process he did for troubleshooting before he made a thread, to mikejmoffitt's excellent assistance and again, very clear communication. I wish that all problems could be solved this eloquently, and quickly. Kudos guys!

Click to expand...

Thanks for your kind words c0nn0r My aim in posts like this is always to a) troubleshoot the problem and get input from others and b) to be as clear and detailed as possible so as to help anyone else who may experience this kind of problem. In my opinion the Neo Geo system is definitely one of the most complex home consoles out there, and compared to the number of forum posts about other more common consoles (sega, nintendo etc) there isn't as much for the Neo Geo (although there is much more than I expected). That makes it even more important to keep a record of issues like mine, through forums like this.

Thanks again everyone!