AES to MVS (plays AES games on MVS)

[aha2940]

Unfortunately the only games I can help test with are samsho1, aof, ffs, cyber lip. And I don't know anyone else local with AES carts, either... :/

Yep, I'm in the same boat. I only got SS1, SS2, AoF2, Nam and Magician Lord. Unfortunately the latest, more complex games are not easily obtainable in terms of $.

 

[Apocalypse]

Thanks for trying to help guys.
Yep I had a look and the cheapest encrypted game must be KOF 2000. The cheapest one I found was $150...

 

[aha2940]

Thanks for trying to help guys.
Yep I had a look and the cheapest encrypted game must be KOF 2000. The cheapest one I found was $150...

Maybe you could borrow one of those carts and test from some member here in NZ. Also, keep in mind that Metal Slug X has something unique to it, because most current MVS to AES converters have issues/special switches for it, so doing tests with it may be worth it.

 

[Apocalypse]

Maybe you could borrow one of those carts and test from some member here in NZ. Also, keep in mind that Metal Slug X has something unique to it, because most current MVS to AES converters have issues/special switches for it, so doing tests with it may be worth it.

SNK stuff is rare in NZ...
MSX AES is like $2000... No way I'll find/buy another one ever.

 

[xsq]

this is awesome man! I hope this can be tested with more carts. Great stuff!

 

[HornheaDD]

This is badass.

 

[HeavyMachineGoob]

Thanks for trying to help guys.
Yep I had a look and the cheapest encrypted game must be KOF 2000. The cheapest one I found was $150...

KOF 99 should be the cheapest guaranteed encrypted game.

KOF 98 and Metal Slug X just encrypt the P ROM(s). Some KOF 98s come with standard PROGBK1 boards though.

Every game after KOF 99 use the CHAFIO board, which is almost entirely encrypted. Later versions encrypt the M1 ROM too. Very seldomly were there games with a CHAFIO and an unprotected PROGBK1 board. Some KOF 2000, some Metal Slug 3, Prehistoric Isle 2 and Strikers 1945+ are examples of that. Later Playmore and SNK Playmore games encrypt everything, P, V, C, M and S.

 

[Apocalypse]

I did work a lot on this project, mainly to minimize the number of modifications needed on the slot.
After many tries (I won't give all the details here, it would be too long) I managed to reduce the modification in the slot to 4 lifted pins and 8 wires to solder, then 3 pins and 6 wires to solder and finally 2 lifted pins and 4 wires to solder!





1) How it works?
Basically we have to deal with 2 NEO-ZMC2 chips (or 1 ZMC2 and 1 PRO-CT0 for older slots) so the idea was to use the one in the AES cartridge for decoding and make the second one on the MVS board "transparent". However in order to reduce the amount of modifications in the slot I've split functionalities between the 2 chips.
So the chip in the AES cartridge does the decoding EXCEPT the A/B pixels swapping. Horizontal mirroring via the H signal is maintained.
The chip on the MVS board does the A/B pixels swapping only.

2) The trick
Yes there's a trick. Signals GAD0 to GAD3 and GBD0 to GBD3 from the AES cartridge are wired as follow:
- GAD0: CR0
- GAD1: CR8
- GAD2: CR16
- GAD3: CR24
- GBD0: CR1
- GBD1: CR9
- GBD2: CR17
- GBD3: CR25
All the other CR signals are left unconnected. LOAD and H signals come from the MVS board. EVEN is grounded (A/B pixels swapping disabled).
The trick is to trigger CR signals 4 times on the MVS side during a normal 3Mhz cycle. The ZMC2 in the AES cartridge will just provide pixel data in a serial way always on the same CR pins on the MVS side. Therefore we need to reset the data latch on the MVS side each time new pixels are decoded.
First cycle: CR0/CR8/CR16/CR24/CR1/CR9/CR17/CR25 are decoded and available through CR0/CR8/CR16/CR24/CR1/CR9/CR17/CR25 on the MVS side
Second cycle: CR2/CR10/CR18/CR26/CR3/CR11/CR19/CR27 are decoded and available through CR0/CR8/CR16/CR24/CR1/CR9/CR17/CR25 on the MVS side
Third cycle: CR4/CR12/CR20/CR28/CR5/CR13/CR21/CR29 are decoded and available through CR0/CR8/CR16/CR24/CR1/CR9/CR17/CR25 on the MVS side
Fourth cycle: CR6/CR14/CR22/CR30/CR7/CR15/CR23/CR31 are decoded and available through CR0/CR8/CR16/CR24/CR1/CR9/CR17/CR25 on the MVS side
I've just used the 12M signal as the LOAD signals on the MVS side and it works almost perfectly (see 4)Drawbacks).

View attachment 34562

3) Schematics:
I used a 74LS157 to redirect signals either in their original configuration or modified for the converter. The selection signal I used is SLOTCS which goes low when the slot is selected. Thus you can use the adapter in any slot. When adapter is removed, simply use a termination connector (connects pads to pins in the MVS just as the original wiring).



4) Drawbacks:
Unfortunately this isn't 100% perfect. There are few graphic glitches which, I think, are due to the fact I simply used the 12M signal to feed the LOAD on the MVS. I will make more tries to see if I can fix it (delaying the signals by 1/4 of a cycle, etc.).





Still I'm happy with the result as it needs few modifications on the slot (2 pins lifted, 4 wires to solder).

 

[Apocalypse]

No success in trying to get rid of those glitches. I've tried a lot of things (delayed signals with capacitors or logic, XORing, etc.).
I'm not even sure the game itself isn't faulty (although it worked fine the first time) but don't have either an AES or a second AES game to make more tests.
I've contacted fluxcore (we live in the same city) but had no answer so far.
There's still few things I'd like to experiment (cause I'm pretty sure if the game isn't faulty it's just a small timing issue).

 

[aha2940]

Man, great advances so far. Hope you can figure out the glitches issue and solve it.

 

[playboycougar]

A stupendous achievement and a true advance for our hobby. It is dedication and perseverance like yours that truly moves Neo Geo gaming forward and creates new ways to enjoy it. I hope you continue to refine your work so it can be fully enjoyed by members of the community.

 

[xsq]

this is already pretty awesome. Now if someone could get you another cart...

 

[Apocalypse]

this is already pretty awesome. Now if someone could get you another cart...

fluxcore should come to my place on Saturday afternoon with few AES games he owns. I'll post the results here!

 

[Heinz]

fluxcore should come to my place on Saturday afternoon with few AES games he owns. I'll post the results here!

We are all eagerly awaiting the news. You have done an amazing job so far, I could almost overlook those glitches.

 

[mikejmoffitt]

It's cool to see the use of the ZMC2 as a "pasthrough". However, I think that is giving you a 1px (1/12MHz) delay that is going to be a problem. This isn't an analogue problem, but rather one from putting an extra synchronous flip flop in the way (which is how ZMC2 is configured with your setup). You may get around it by clocking the ZMC2 and LOAD signal with an inverted 24Mhz clock (or possibly non inverted, try both perhaps).

 

[Apocalypse]

It's cool to see the use of the ZMC2 as a "pasthrough". However, I think that is giving you a 1px (1/12MHz) delay that is going to be a problem. This isn't an analogue problem, but rather one from putting an extra synchronous flip flop in the way (which is how ZMC2 is configured with your setup). You may get around it by clocking the ZMC2 and LOAD signal with an inverted 24Mhz clock (or possibly non inverted, try both perhaps).

Yes that's what I thought too. I've already tried pretty much everything (12Mhz inverted, 24MHz inverted or not, 12M.24M, NOT(12M).24M, 12M.NOT(24M), NOT(12M).NOT(24M), 12M+24M, NOT(12M)+24M, 12M+NOT(24M), NOT(12M)+NOT(24M), XOR, NXOR, delay, etc.).
I might get it work with a sufficient delay on 12M and LOAD to feed the ZMC2 in the MVS. However if that works it means lifting 3 pins and soldering 6 wires in the MVS.

 

[Apocalypse]

Hey Mike,
I'm not so sure what I tried was valid at all. What I thought were 74HC03 chips are in fact 74LS03...
Not good to use TTL to revert clock signals etc. because of levels uneven between inputs and outputs.
Do you think I should redo all my tests (not really an issue, chips are on sockets) with HC serie chips?

 

[mikejmoffitt]

I don't think it should make a big difference - the HC series will be faster but at 24MHz nothing is particularly fast here anyway.

 

[mcjefe]

I would be interested in one of these since I have a good sized AES collection and just recently picked up a big red cab.

 

[Apocalypse]

fluxcore should come to my place on Saturday afternoon with few AES games he owns. I'll post the results here!

Unfortunately he couldn't come last week.
I'll try to reschedule it to this Sunday.

 

[Apocalypse]

So fluxcore came to my place yesterday.
We've tried the converter with another AES game and the result was the same : few graphic glitches.
We've tested my AES game too in his console and everything was fine so I know now that I need to improve the converter.

 

[fluxcore]

Issue seems to be the edge column of pixels on some but not all sprites (not quite sure if it's only the left or sometimes rightmost column as well), a bit weird really. Apocalypse has a few other things to try, so there's still hope.

Pretty cool seeing it working though, absolutely playable, just the minor graphical glitches. Great work

 

[Apocalypse]

I got it running almost perfectly by feeding the 12M AES cart with NOT(12M) and using NOT(12M) OR 24M to feed the LOAD signal on the MVS board:

[video]http://fr.tinypic.com/r/i20ody/9[/video]

View My Video

If you pay attention you can notice small graphic glitches when sprites are mirrored (not always visible):

 

[fluxcore]

A big improvement on the last attempt... the fragile cart loom still scares me a bit though

 

[Apocalypse]

A big improvement on the last attempt...

Yeah but still not perfect grrrr...

the fragile cart loom still scares me a bit though

You know I like to do things on the cheap