Multicarts & Board / Slot Compatibility regards PCM sound

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
I wanted to ask this of the wider community since I've got limited access to different MVS hardware. I've been wondering about some sound issues on MVS multicarts for a while now and in the first instance I put it down to dodgy NVPROM chips or the odd incorrect byte in the ROM images etc. I've started to question that thought due to the fact that it always seems to be related to PCM sound and nothing else. ie. I have a 161 in 1 which has very good sound, with only noticable glitches in Pulstar. I recently got a couple of 138 in 1s and for one or two games the sound can occasionally glitch, but not always. You can play say the title music of Ninja Masters and it can play perfectly. You can then power the system off and on and listen to it again and it can glitch on one note and be OK all the rest of the way through.

I've been told that some of the older slots are less reliable in terms of connection, and ultimately the board design - buffering etc. I am now wondering if some of the 1 slots have more of an issue related to sound vs some of the larger 2, 4 and 6 slot boards.

Has anyone experimented with these kinds of PCM issues on multicarts vs different MVS boards to see (or hear) if there's any improvement?

Part of me thinks its a glitch in the implementation - track layout, lack of bypass caps, FPGA / CPLD implementation perhaps inaccurate or some minor timing glitch etc.

But just to be clear, it is always sound, never P ROMs, S, C, or M. Seemingly just V ROMs only.

I've tested in both of my MV-1FZ boards and they behave identically.


All of that said, I can't discount that maybe there is the odd bit flakey in the ROM images, or maybe the odd lost bit in dodgy NVPROM chips, and possibly in C ROM it might not be visable, yet with PCM audio that 1 bit is enough to raise the volume.

It's also worth pointing out that the problem I hear is the volume goes loud on its own for that one note. It doesn't sound garbled etc, it increases in volume - almost double.

Starting to wonder if my board needs a recap or something as well - but then why does it only do this in specific games, and only from multicarts?
 

ggallegos1

Cholecystectomy Required.,
10 Year Member
Joined
Oct 1, 2013
Posts
5,153
I have an MV-1A and a 161. I've played most of the games extensively, with the exception of the hacks and some of the fighters. I know on mine Pulstar will have some sound cut off from time to time and Blue's Journey has no sound at all.
 

Ninjainspandex

Sakura's Bank Manager
Joined
Jan 20, 2013
Posts
732
That is interesting, Blues Journey has flawless sound on my 161 and I have never had the sound go out in Pulstar. Maybe yours has a factory defect? These things are not made in the highest quality.
 

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
I am sure there are some flakey ones out there - My Pulstar works OK for the most part but has the rare issue where the volume of one sample just goes loud for a second. Hence why I am wondering if it's an issue with the MV-1FZ itself. It behaves the same way on both MV-1FZ systems, so it's not a loose connection etc. Seems strange to me that we only ever seem to get sound glitches on these carts, or at least I do =/ I've got 2 x 120 in 1, 1 x 161 in 1, 1 x 60 in 1, and 2 x 138 in 1 and none have graphics or code problems, yet I do hear the odd sound glitch periodically in most of them. The manufacturers in China have said that the older slots (like mine) are more prone to sound glitches and newer MVS revisions should not have the issue. This is news to me but it is news from the actually factory that make them. Whether they are talking crap remains to be seen...
 

aha2940

AH, A, COLUMBIAN!,
Joined
Dec 15, 2013
Posts
2,528
I am sure there are some flakey ones out there - My Pulstar works OK for the most part but has the rare issue where the volume of one sample just goes loud for a second. Hence why I am wondering if it's an issue with the MV-1FZ itself.

This same thing happens on the Pulstar on my 161, using an MV-1C board. Also, the attract mode music on KoF95 has missing instruments on my 161.
 

chris1

POCKETBIKE NUT,
Joined
Oct 18, 2002
Posts
10,830
I have a 161-1 in my 4 slot...with three original MVS carts in the motherboard..
So far everything seems to play fine...what I do notice is some of the games on the multicart are lower in volume then others on the cart,this noticed especially when the cab is in cycle-demo ..The low volume multicart game if selected /in slot one will play low,then when the next game=original MVS cart in the cab starts it's demo it's volume is louder,..the original MVS carts in slot 3,4 will cycle-demo with the same volume as the original cart in slot 2...once the low volume game in the multicart stats it's demo it's low again...that only goes for a few games in the multicart,I think Pulstar is on of them..

Multicarts King Of Monsters...It does nothing via playing a attract demo ..The title screen just sits there KING OF MONSTERS..is there a wait time to a demo starts or no demo at all..?





http://youtu.be/Ja3Cjm2C4nI


http://youtu.be/7neHVEeZxy8

http://youtu.be/E4ve5rQ0s84
 

ne7

Armored Scrum Object
15 Year Member
Joined
Aug 12, 2008
Posts
268
as long as you use it on a 4slot with nothing else plugged in - the cart works just fine for me, though its never had any real use as I have 98% of the games on real carts anyway... tried on both 4 (the big and mini version) slot revs too... :)
 

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
Thanks for the info so far! I've been testing some of my carts some more and I am convinced that there's an issue which relates to the PCM audio on these crappy multicarts, certainly affecting the MV-1FZ. The nature of the glitches is very infrequent and tends to be when there's a lot of access to the VROM at the same time - ie. in really complex music. I've even been able to cause the glitch to occur on one cart just by moving the character selector in order to cause the system to access on a different address vs the address(es) it's currently hitting for the background music. This leads me to believe there's some noise or something somewhere, perhaps a dip in current (lack of bypass cap). Looking at the board in question here (138 in 1), I see there are quite a few missing bypass caps.

I am going to replace the 'random' value caps that are positioned on the board and ensure every chip on there has a bypass cap, and swap out the diodes for a 3.3v regulator like I did on my 120 in 1 (that regulator sorted all the wierd stuff out on that cart). I guess the clue here is like I said - graphics and code always 100% rock solid, sound on some titles, particular those using PCM tend to very infrequently glitch, but only on certain titles. The other possibility here is the implementation of the PCM audio chip(s) which they embed within the FPGA / CPLD aren't 100% accurate. Or perhaps its an issue due to the 3.3v logic and or bad track layout, lack of sufficient pull ups or something.
 
Last edited:

ggallegos1

Cholecystectomy Required.,
10 Year Member
Joined
Oct 1, 2013
Posts
5,153
Thanks for the info so far! I've been testing some of my carts some more and I am convinced that there's an issue which relates to the PCM audio on these crappy multicarts, certainly affecting the MV-1FZ. The nature of the glitches is very infrequent and tends to be when there's a lot of access to the VROM at the same time - ie. in really complex music. I've even been able to cause the glitch to occur on one cart just by moving the character selector in order to cause the system to access on a different address vs the address(es) it's currently hitting for the background music. This leads me to believe there's some noise or something somewhere, perhaps a dip in current (lack of bypass cap). Looking at the board in question here (138 in 1), I see there are quite a few missing bypass caps.

I am going to replace the 'random' value caps that are positioned on the board and ensure every chip on there has a bypass cap, and swap out the diodes for a 3.3v regulator like I did on my 120 in 1 (that regulator sorted all the wierd stuff out on that cart). I guess the clue here is like I said - graphics and code always 100% rock solid, sound on some titles, particular those using PCM tend to very infrequently glitch, but only on certain titles. The other possibility here is the implementation of the PCM audio chip(s) which they embed within the FPGA / CPLD aren't 100% accurate. Or perhaps its an issue due to the 3.3v logic and or bad track layout, lack of sufficient pull ups or something.

Let us know how this goes for you. I've never heard of anyone modding a 161, so I'm intrigued
 

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
Break through!!! I was right! These are so crappy assembled... OK, so I've fitted around 10 x 100nF ceramic caps in all the spots where the bypass caps are missing, but you need to go further... I would personally remove all of the other bypass caps and replace those with 100nF as well. The caps on there seemingly are random values, and get this - I found a 220 ohm resistor in the spot for a bypass cap!!! Now that resistor wasn't the cause, I tested at every stage to work out how much each change improved the situation and all had an effect but all were required before it was 100%. The 2 diodes on my 138 in 1 have been swapped out for a 1117 3.3v regulator, and the final step here which made the most improvement was me adding a reservoir cap to the 3.3v line in the SMD spot next to the regulator. I started with 33uF and it was a lot better but still glitching, more so after I removed the diodes on the CHA board and joined a wire to the regulator board. At that point I came to the conclusion that since it had worsened after powering both boards from the 3.3v regulator it must be a current issue. I swapped the 33uF reservoir cap for 220uF and bingo - no more random volume issues.

I will now go back and do similar mods to my 120 in 1, and if I have the patience work out the best way to get the 161 in 1 working with a regulator (You cannot easily just swap out with a 1117 due to current required, and I think there might be some 5v logic somewhere that depends on how some pads are linked.

I've done the 100nF caps using standard ceramics at the moment but already ordered some SMD caps. Once those arrive and I've cleaned the board up I will upload the video.

Please note this fixes a very specific issue here that may not exist on other MVS models, although I am sure it will help a lot. If you've got corrupt or glitchy V ROMs then it won't fix issues related. For example, on the 138 in 1 I've been working on I already came to the conclussion that whilst I had the random PCM issue on Metal Slug 5, there are also some glitches in the related NVPROM. ie. For example certain sounds will always sound screwy on that game, in certain music tracks - always at exactly the same time, always repeatable. But where I was getting random volume issues which were not the same every time the issue has gone. Ninja Masters and Art of Fighting 3 had this to a minor degree, now flawless.
 
Last edited:

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
Here's the video - https://www.youtube.com/watch?v=Dai7QoxobVU

I managed to get both of my 138 in 1's working perfectly! (almost) There's still the odd game with a sound glitch on certain carts, but that's clearly (I think) NVPROM corruptions etc. For example on one of the 138 in 1 carts Metal Slug 5 has a couple of corrupt samples, but every other game seems perfect on both carts, and MS5 on the other cart is perfect. It's amazing how bad these carts are from the factory. If you've recently purchased one of these (or even in the past), it's well worth doing the mods I did, and certainly no matter what you should remove that 220ohm resistor that they seem to be installing in one of the bypass cap positions. I did wonder if someone had filled the tape machine with the wrong component spool but they seem hand soldered?!? Having said that, maybe they are using a pick and place to put the components on there and then soldering by hand. That would explain why there's a resistor where a cap should be. Also worth pointing out that a single change I made on its own was not responsible for getting the cart to a proper working state regards PCM. I tested at every stage and only by the time I had done all the mods was it 100%.

EDIT: Looked at my 120 in 1's today as well. Same issues all over again - tonnes of bypass caps missing, more than on the 138 in 1. There are some underneath the FPGAs which seemed to make the largest difference for me. Also note that there was a 220ohm resistor in a bypass cap spot again lol. Needless to say its had the full clean up work. I seem to have gone through 50+ 100nF bypass caps in 2 days due to these crappy carts lol. I've 3 more carts to look at yet but I fully expect they will need the same stuff doing.
 
Last edited:

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
Just a quick update on this. Whilst its a lot better its still not 100% resolved. I've been looking at the AES and MVS schematics and come to the conclusion that the 3.3v logic levels on the YM2610 bus are the problem. I am going to try some pullups on the data bus and will report back. It's all related to the YM2610 bus only as far as I can tell. These carts probably work better where there's some buffering going on between the YM and the cart - possibly multi slot systems - no idea, not seen schematics from those.
 

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
Anyone done a teardown of a ngdevteam cart? I am wondering if they used level shifters to deal with this. I see lots of small ICs near the cart edge in Fast Striker photos, I guess this is what multicarts are missing.
 

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
Just another update here, not that anyone seems to care lol - It's here anyway. I tested with 5v pull ups and pull downs on the YM2610 data bus (on the cart) and it made no difference. So the logic level wasn't the final part of the problem. Tonight I've made more progress and found that it really needs a cap on the 68K clock input to the CPLD. I finally checked the P/N and they are CPLDs, not FPGA. In any case I thought it would be useful to note here that they are 5v tolerant on IO pins (I checked datasheet first), so there's no risk to using pullups there. In any case the YM2610 data bus seems multiplexed and serves as an address bus as well as far as I can tell, which meant the YM2610 would already have been outputting 5v TTL levels to the cart (therefore it was logical 5v pull ups wouldnt damage the CPLD).

Tonight I added a 470pF bypass cap to the 68K clock on the cart and that made a massive difference. It pretty much sorted my remaining sound issues with these carts (120 in 1, and 138 in 1). I am going to try a 220pF cap instead and move it nearer to the CPLD to see if that completely fixes it. I noticed later on the 161 in 1 they've done exactly that - used a 220pF cap on the clock. For some reason they put bypass caps on the upper prog address lines as well, I figure because they run in parallel to the clock? On the 138 in 1 there are 4 caps on A18, A17, A16, and A15, but not the clock (as default). On the 161 in 1 there are caps only on A18 and the clock. Since the 161 in 1 works perfectly I figure this was done for a reason and I am speculating its to do with the 2 connections being next to each other, and ultimately maybe the CPLD doesn't like the raw clock signal. Not sure how it relates to PCM but the 470pF cap I tested with made a noticeable difference - I tested for well over 1 hour and had only a couple of sound glitches. Without that cap when testing certain games the sound issues were very common. For example on Strikers 1942 - very easy on MV-1FZ to get dodgy sound on that game, but with all the stuff I've done to the cart now its hard to get it to glitch.
 
Last edited:

aha2940

AH, A, COLUMBIAN!,
Joined
Dec 15, 2013
Posts
2,528
Just another update here, not that anyone seems to care lol - It's here anyway.

Man, it's not that we don't care. I, for one, follow this thread and read it whenever you make any update, no matter that the technical level is way over mine. I think many people here care the advances you have made so far to improve the quality of these multicarts, however I'd appreciate (if possible, of course) if you can release some pics or schematics or something graphical that might help understanding the mods you made to the carts, so other people may help you testing and even improving the mods until these carts work as good as they can. Too bad that the problem with aerofighters 3 and Garou on the 161 is not fixable, though but nothing's perfect.

Finally, thanks for your efforts GadgetUK, I am sure they are very appreciated.

Regards.
 

KalessinDB

Mr. Big's Thug
Joined
Dec 17, 2013
Posts
202
Ditto. I enjoy reading threads like these, and knowing that there are much much smarter minds than mine working to improve the systems I love and their accessories, but holy crap you're so far over my head technically that it's like "Mhm.. oh yes... yeah I know some of these words!" level for me.
 

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
Thanks for the nice comments! I am not worthy (Really not! There's a lot I still don't know, and I am learning a lot from people on here!). What I do know is that after testing for a week or two for quite a few hours here and there, I can definitely say that the changes I made are a MUST for anyone with a 120 in 1, or 138 in 1 (or any other multicart that lacks so many basic components) - certainly on my 2 x MV-1FZ at least. The difference is amazing, I've sorted all sound problems I had on these carts except for Striker 1942, which is better than it was but still not 100%. I cannot quite work out what it is about that game that is different from other games. It's easy to assume the VROM images for Strikers was corrupted, but that cannot be the case because now I can play the music in game and have no sound issues for large amounts of time, then the odd glitch, sometimes a lot of glitches all at once then nothing for a while. It's also strange how the same game on a different cart (eg. 120 in 1) has the same exact problems with the same intermittent nature. I still think there's a minor noise issue somewhere there - I really need to get the scope onto it at some point. The same thing happens with Pulstar on the 161 in 1 and 120 in 1. Is it corrupt VROM - How can it be when it's intermittent? It's clearly (well I think) not the Fujitsu NVPROM chips as all other games sound fine on all those carts. I guess there's a small chance there is a timing glitch in the CPLD related to how those games make use of the PCM bus - no idea, I don't understand enough about how the PCM stuff works. I did wonder if the original PCM audio tracks had perhaps been further compressed (from the originals) and as a result there's some artifacting or something in the compression, which means a timing related chance of distortion? ie. Audio is out of spec with what the YM2610 can handle under normal operation, and if too many sounds occur at similar times it can glitch it out? No idea, I guess we would need to dump the VROMs and throw them to someone who's an expert with the YM2610.
 

fluxcore

Another Striker
Joined
Nov 4, 2013
Posts
324
So do you think the bypass cap on the clock is the 'one true' fix to these issues, or are the others (replacing the resistors with capacitors etc) necessary as well (obviously they are a good idea)? You've changed quite a lot of things on these carts by now, just wondering what the 'mod summary' is.
 

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
I've not quite finished with mods and testing yet, I will summarise later! I want to try and add some more bypass caps later. I need to look closely at where PCM data is regards the CPLD - I think a nearby connection might be helping generate noise on the PCM bus. It can't be the CPLD logic or many games would have the PCM volume change issue I am hearing in Strikers. It would really help if I knew other peoples experiences with these games / carts. Anyone else get random volume on only PCM channel for a Strikers 1945?
 

The Webmiester

Pvt. Picklestein,
Joined
Aug 4, 2001
Posts
425
Hi!

I'm working on consolizing my 4-slot and I have a 161-in-1 as well, so I thought I'd start following this too. I have a logic analyzer and scope so if we can figure out where the problems are I might be able to help solve them too.

How do I identify PCM vs analog audio? Is PCM just the sampled stuff?
 

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
Yes, PCM is the samples basically. In Strikers 1945+, what I notice is very randomly the PCM part of the music increases volume (almost double, if not more), yet the other audio tracks remain at the same level. The question is, is it randomly letting bit states slip on PCM data bus, and the YM2610 results in shifting the volume, is it a problem with the way they are replicating the latches in the original PCM chip on official SNK boards - ie. a bug in the CPLD. Since its random and PCM audio works properly in many other games I am having a hard time understanding the cause. It cannot be the V ROM images because of the intermitant nature, and in my case always on specific games like Strikers 1945+. Adding all the bypass caps and voltage regulators has made a big improvement but clearly not fixed it completely. It really needs someone with more experience than me! Ive only got an analogue scope and a pretty crappy logic analyzer, so its difficult for me to make sense of whats going on. I guess the intermittant nature of the problem doesn't help either. Its definitely not a bad connection or dry solder etc, my boards have had a good overhaul and I guess we could even rule out the NVROM chips since they dont glitch for P, M, S, or C stuff, only samples. I did wonder if I was looking in the wrong place, and perhaps the issue relates to the Z80 bus? I know the Z80 controls the YM2610, but does it have any control over volume of samples? Also wondering if there's some parasitic capacitance issue here related to the YM2610 data bus and it running too close to other traces. I did have a go at adding some 47pF bypass caps to the data bus yesterday but it made no difference.

On the 161 in 1 I find the sound very very stable - I think there's only Pulstar on there that can have the volume issue with PCM and I am not sure if on the 161 in 1 it relates to bad V ROM data. Whereas on the 120 in 1 and 138 in 1, those games can one minute sound perfect, then randomly have the loud volume issue on PCM, then a second later be normal again - even playing the same music over again, the issue often does not repeat at the same point.

Since this appears in certain games I am also speculating that if it is a noise / parasitic capacitance issue, it could be happening more with games that are using the PCM bus in a certain way - ie. contstant high level of activity as samples are loaded into the YM2610 (is that even how it works???) due to complex music, or rather music that uses lots of samples. Lots of activity would aggrevate a parasitic noise issue I think.
 

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
I think I've sorted it - yay!!!

SDPMPX - Needs a 47pF bypass cap. This apparently is related to the YM2610 - anyone know how specifically? Something to do with the multiplexed address / data bus or something? In any case, I added 47pF cap on my first 120 in 1 - sound glitches in Strikers 1945+, Pulstar and slight glitches fixed in Neo Turf Masters - all working 100% OK. Repeated for the 2nd 120 in 1, and voila - problems solved in that 120 in 1 as well. It was a little harder in the 138 in 1 as I had to scratch the solder mask off on a trace to fit the cap, but it's sorted on there as well =) I am so happy lol! These carts now work really really well. Straight from the factory they are s***. If you've got one a 120 in 1, or a 138 in 1 you MUST do the changes if you want to get the best out of it.

Fingers crossed it doesn't re-occur after everything has cooled down over night. Going to do my 2nd 138 in 1 now to make sure the same fix works on that.
 

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
Tested thoroughly this evening on 2 x 120 in 1, and 2 x 138 in 1.

It's FIXED =)

Now, the question is, was this just a problem on the MV-1FZ or does anyone else get glitchy sound on the 120 in 1 or 138 in 1?

I will upload a video later and perhaps create a new thread with the end solution (overall).

EDIT: I found this link which explains the use, and now all makes sense to me regards the behaviour.

https://wiki.neogeodev.org/index.php?title=YM2610_bus_timing

EDIT2: Whilst its now perfect, I think its worth adding a 47pF cap to SDRMPX as well. I will do that tomorrow, but it's working perfectly already on all 4 carts - even after cold test. This is brilliant news!
 
Last edited:

GadgetUK

Ace Ghost Pilot
Joined
Sep 27, 2013
Posts
1,323
Here's the link to some crappy pics that might help if you want to fix the sound on your 120 in 1 or 138 in 1. The same fixes (albeit in different component numbers / positiions on the boards) will probably apply to other multicarts as well.

If you are unsure about something just ask. Note that this only fixes general stability and random sound glitches (in particular the PCM volume level issue in games like Pulstar, Strikers 1945+, Raging Dragons etc).

138 in 1, and 120 in 1
https://www.dropbox.com/s/nx0fhk0eimacrpi/multicartfix.zip?dl=0

This also fixes Pulstar on the 161 in 1 for me as well =D The 161 in 1 needs traces scratching (like the 138 in 1 does). The 120 in 1 seems to have 2 pads for the caps already.
161 in 1 (150 in 1 also?)
https://www.dropbox.com/s/kdf1bd2fqow6xkv/161 in 1.jpg?dl=0

I will upload a video later today.

Wow, I've noticed it has fixed a lot of little glitches in games like Puzzle Bobble, Blazing Star, Pulstar etc. Its worth checking on all multicarts to see if the SDRMPX and SDPMPX pins have a 47pF bypass cap on them, if not add it. That change on its own will give much better sound. I do recommend adding missing bypass caps though, and swapping out the 220 ohm resistor in bypass spot in 120 in 1 and 138 in 1. 3.3v regulator is a good idea on 120 in 1 and 138 in 1. I've not fitted a regulator to the 161 in 1 yet.

I am going to create a new thread for this - I think it needs it or people wont see the relevance from the current title.
 
Last edited:
Top