NEOSD AES Support thread !

[ProGEO]

Got my NeoSD AES the other day and everything has been working great.

I do have one question though, is Crouching Tiger Hidden Dragon not working? I thought I had seen a post where Raz confirmed it worked, but I cant get any of the 3 versions in MAME to work for me. After it flashes the system just constantly resets until I turn it off and back on again to get into the menu.

I've tried the roms in .183 rom set, and the ones provided by Rot.

I'm on v1.05 of the firmware, and I used the v1.03 neobuilder.

How did you get the menu to load!? I am having the same problem, I attempted to load Crouching Tiger and am caught in the boot loop, but am unable to get out of it. I have tried holding player 1 start while I turn on the AES console, but it still just loads up the repeating reset issue! I really want to enjoy some NEO GEO gaming right now and this is not fun...

I will say that this cart is pretty amazing and I am sure I will be able to get it working again, even if it has to be with the help of the neosd team.

Thank you.

[Sceptre_JLRB]

How did you get the menu to load!? I am having the same problem, I attempted to load Crouching Tiger and am caught in the boot loop, but am unable to get out of it. I have tried holding player 1 start while I turn on the AES console, but it still just loads up the repeating reset issue! I really want to enjoy some NEO GEO gaming right now and this is not fun...

I will say that this cart is pretty amazing and I am sure I will be able to get it working again, even if it has to be with the help of the neosd team.

Thank you.

http://www.neo-geo.com/forums/showth...=1#post4172992

http://www.neo-geo.com/forums/showth...=1#post4173000

http://www.neo-geo.com/forums/showth...=1#post4177410

[ProGEO]

http://www.neo-geo.com/forums/showth...=1#post4172992

http://www.neo-geo.com/forums/showth...=1#post4173000

http://www.neo-geo.com/forums/showth...=1#post4177410

Best... info... evar. YOu saved my night. Thank you.

[ckchan]

Hello,

This may sound stupid, but it happened to me, it took me some time to realize why i was getting resets with a non official psu.

I was using one of those that have multiple connectors of different sizes, i just choosen one of those connectors and the aes was having random resets, some time later i got the original psu that came with another AES and realized the connector had no looseness once attached on the AES, as the one i was using.


Check that the PSU you are using has the right diameter on the connector. There is an easy test, just move the connector and see if you get resets while moving it.

I am talking about interior diameter and exterior diameter aswel.

Attachment 43449



Thanks

Finally I tested the cart with an original PSU. The reset problem is solved. Thanks!
However, it seems to me that the PSU gets warm after flashing a few games and play less than 1 hour. Is it normal?

[GadgetUK]

Finally I tested the cart with an original PSU. The reset problem is solved. Thanks!
However, it seems to me that the PSU gets warm after flashing a few games and play less than 1 hour. Is it normal?

Yes (unless the PSU needs recapping - if you haven't done that, and its an original PSU, I would consider it as they all need recapping at some point).

The Neo SD doesn't pull much if any more power than a typical cart really. I need to check exact current drawn on the MVS version (maybe I will get an AES version later in the year and check that, but probably the same as the MVS version anyway), but the voltage indicator (drop) on my MVS is a good reflection of current drawn, and it's not dissimilar to say an original Sam Sho V.

[GadgetUK]

It might be worth someone adding a list of FAQ problem resolution check points to the Neo SD website or something, because we've had a few instances now of people getting resets or CRC issues when its not the fault of the cart. And I imagine these problems taking time to investigate etc, time that could be better spent on shipping and firmware updates etc.

eg.

1) Clean your cart slot and inspect to make sure pins aren't damaged. And after repeated use, clean the Neo SD PCB edges too.

2) Check your PSU is not the problem (ie. needing recapping). Ideally, measure the 5v on the connector, measure the 5v whilst the system is powered on (perhaps by measuring between pin 2 and pin 4 of the video output DIN? - for those that don't fancy removing the lid). Also note - 5v measured does not neccessarily indicate a good PSU, the PSU might need recapping still. You could measure using AC mode on your mulitmeter to see how much AC ripple there is on the 5v line. I've done that before to determine when C64 PSUs need recapping.

3) For alternative AES PSU (not original SNK manufacture) - are you sure the DC jack is the correct type (ie. there are two commonly accidentally mixed - 5.5 x 2.5 and 5.5 x 2.1. If your AES uses 5.5 x 2.1 (which I think they all do - unless the socket has been replaced on the AES previously, then a 5.5 x 2.5 will result in random resets and power issues.))


Additional for the MVS:-
4) Does the MVS need recapping? These systems often have thousands of hours of usage previously and whilst original carts might seem OK, the Neo SD might be more sensitive to a bad 5v line.

5) Does the 24Mhz crystal need grounding? (So far seemingly only affects some 1FZ).

6) Does the MVS have a slot riser board (where the slot fits on a PCB that then connects to the main MVS PCB? Because I've seen two examples now where replacing that riser board or cleaning the connections between the riser and the MVS can make a huge difference to how the Neo SD runs on those models.

7) Try reconnecting the 2 Neo SD PCB boards - ie. carefully disconnect them from each other and reconnect (but that might not apply to AES carts - no idea, not sure if the screws can be easily removed from the shell until I see a shell myself).

EDIT: Also, I vaguely remember seeing a problem where people get stuck in a reboot loop and some file can be deleted from the SD cart to resolve that? That could be added into that list too.

[Nostromo]

I've had this weird thing happening today... I had my CMVS running by itself for a few minutes (with NeoSD) - the game was Mutation Nation. All of a sudden, the system started resetting itself. So I went into the NeoSD menu and it kept resetting itself (so it went back to the game). I eventually switch it off, back on, flashed another game and now it seems ok. Any ideas?

[GadgetUK]

I've had this weird thing happening today... I had my CMVS running by itself for a few minutes (with NeoSD) - the game was Mutation Nation. All of a sudden, the system started resetting itself. So I went into the NeoSD menu and it kept resetting itself (so it went back to the game). I eventually switch it off, back on, flashed another game and now it seems ok. Any ideas?

It could just be a consequence of a slightly bad connection - are you using the boards without a shell (like I do atm), because I've had similar issues to that but they are very rare and usually after reseating the boards and a power cycle its usually OK.

[Nostromo]

Well, yes, no shell, still waiting for one. Having said that, the boards have been sitting in the slot since day one I got it. I never removed them. Switching the CMVS off and on seems to have fixed it. I didn't re-seat anything.

[GadgetUK]

Well, yes, no shell, still waiting for one. Having said that, the boards have been sitting in the slot since day one I got it. I never removed them. Switching the CMVS off and on seems to have fixed it. I didn't re-seat anything.

What MVS are you using?

[Nostromo]

What MVS are you using?

MV1C (a JNX CMVS). Never had one issue. I left another game running for an hour now, didn't happen again. I wonder if it was a PSU issue.

[GadgetUK]

MV1C (a JNX CMVS). Never had one issue. I left another game running for an hour now, didn't happen again. I wonder if it was a PSU issue.

Could be some noise through the PSU or something. If it doesn't start occuring frequently I wouldn't worry about it.

[wyndcrosser]

I had the random resets with an unofficial psu from ebay (to ckchan's post).

[Nostromo]

Apologies for posting in the wrong thread. Just noticed this is the AES thread.

[Terabyte]

Sorry for not answering the email. We keep investigating the issue, it's quite strange. Can you properly play the games without resets or unibios exception screen?

What do you mean resets? unibios ex screen?

I can play the game perfectly fine it plays and looks fine. Its just when I do the CRC check it fails. When I CRC check an original game cartridge it checks off good. It only fails with NEOSD.

[Gyrian]

I've encountered a small graphical issue that presents itself inconsistently. Almost looks like static, with single pixels flashing randomly. This is unscientific, but I think it's more prevalent with bigger games. I've recorded a brief video showing this glitch:



I've never spotted before this with my cartridges, so I'm thinking it's either from the NeoSD or the combo of my setup + NeoSD. This is what I'm running:

Neo Geo AES 3-5
SEGA SA-160A power supply (Mega Drive)
NeoSD 1.05

Any thoughts? Apologies if it's been addressed before, I wasn't able to find anything here other than a passing mention of a similar sounding issue by member FikNaldo in the official news thread (he described a small issue with 'speckles').

[phoenixdownita]

Got my NeoSD AES and was wondering what's to be expected on the new firmware aside ironing out the last kinks for those few that still have some issues.

Support for CTHD on AES would be something but I am not sure how feasible/realistic that is, anyway what else is brewing? Can we get some hint? Please?

[neodev]

Got my NeoSD AES and was wondering what's to be expected on the new firmware aside ironing out the last kinks for those few that still have some issues.

Support for CTHD on AES would be something but I am not sure how feasible/realistic that is, anyway what else is brewing? Can we get some hint? Please?

There are a lot of big and small changes in the new firmware. One of the (announced) big ones is the Jukebox feature, that will allow you to send sound codes to the sound hardware so you can play songs from the neosd menu. Similar to the unibios jukebox feature, but neosd allows you to name the sound codes, and use preexisting music lists so you can choose songs and effects by name.
Also included is the improved saveram & load that doesn't depend on the MVS battery to work properly, so you can have your softdips and highscores saved even if you battery has died or is not present.

Some of the minor fixes are support for the latest encrypted kof99k dump and of course CTHD (and its variations) working on AES.

[phoenixdownita]

There are a lot of big and small changes in the new firmware. One of the (announced) big ones is the Jukebox feature, that will allow you to send sound codes to the sound hardware so you can play songs from the neosd menu. Similar to the unibios jukebox feature, but neosd allows you to name the sound codes, and use preexisting music lists so you can choose songs and effects by name.
Also included is the improved saveram & load that doesn't depend on the MVS battery to work properly, so you can have your softdips and highscores saved even if you battery has died or is not present.

Some of the minor fixes are support for the latest encrypted kof99k dump and of course CTHD (and its variations) working on AES.

Nice, thanks for sharing ... wrt CTHD does it mean that in general BRAM will be supported/simulated on AES so when playing in MVS mode maybe we can save softdip settings and what not also on AES (also in other games)?

[Lone-Wolf]

I've encountered a small graphical issue that presents itself inconsistently. Almost looks like static, with single pixels flashing randomly. This is unscientific, but I think it's more prevalent with bigger games. I've recorded a brief video showing this glitch:



I've never spotted before this with my cartridges, so I'm thinking it's either from the NeoSD or the combo of my setup + NeoSD. This is what I'm running:

Neo Geo AES 3-5
SEGA SA-160A power supply (Mega Drive)
NeoSD 1.05

Any thoughts? Apologies if it's been addressed before, I wasn't able to find anything here other than a passing mention of a similar sounding issue by member FikNaldo in the official news thread (he described a small issue with 'speckles').

I have an identical setup (with official UK Mega Drive PSU) and I'm not getting the glitches. Have you checked the PSU or tried an alternative?

[Razoola]

That pixel glitch looks simular to those obtained when work was done trying to get AES carts working on MVS systems. It suggests the cause may be the result of a slight timing issue in the PRO-CT0 / NEO-ZMC2 chip / NEO-CMC emulation on the AES NeoSD.

That is unless the game in the video is that SSV xbox hack, that has pixel glitching but it is related to paletteRAM access.

[Terabyte]

Just wanted to report back it turned out to be a bad power supply connection. The dc connector on my original power supply and neogeo was dirty. I had to scrape the connectors a bit on the neo geo side and power supply side then use some alchohol and stuffed some thin paper into the connector holes and move it around. That finally fixed the CRC issue. I'm guessing the NeoSD is a little sensitive with amps. I noticed because of the connections the amps were below 1.2-1.3 afterwards they bounced to 1.5-1.6 once at that level the CRC checks were fine. My guess is the original carts use a little less power than the NeoSD. Either way its fixed. Its a flawless cart, thanks NeoSD team. I hope this info can help others.

What do you mean resets? unibios ex screen?

I can play the game perfectly fine it plays and looks fine. Its just when I do the CRC check it fails. When I CRC check an original game cartridge it checks off good. It only fails with NEOSD.

[phoenixdownita]

I've got to ask to NeoSD team if these chips where ever considered (winbond W968D6DA 256Mbits [16M x 16 at 70ns] or 32 MBytes):
https://www.digikey.com/product-deta...-TR-ND/5155090

Datasheet here:
http://www.winbond.com/resource-file...3_20130627.pdf

4.5US$ for 32MB of PSRAM at 70ns access time .... 4 chips is 128MB ..... that is for <20US$ in material PSRAM could theoretically have been used on the NeoSD cart .... I know I'm probably beating a dead horse ... just curiosity.
[I understand digikey asks for 5K pieces minimum order but that's beside the point here].

Even if 8 chips would have been needed (say to implement 5 busses and have on one a complete 128MB just because) it would have been less than 35US$ of extra material, even just writing 1 byte at a time at 10Mhz would have allowed "flashing" at 10MB a sec (the only guarantee of a class 10 SDHC anyway) .... the biggest games likely would have taken < 10 sec .... less than 30 sec to account for the vagaries of life etc... etc.... reducing the throughput to 3MB a sec because shit happens .... [on the NeoSD I personally experience around 450 KB/s for a 20MB game that I tested with and so extrapolating a bit for a 96MB games it would put it around 3m30sec, I haven't tried the big games yet but that's in line with the reported times]

As I said pure curiosity at this point, maybe they felt the extra 40/50EUR would have pushed the envelope or maybe the fact that at every "reboot" 30 sec (in my 3MB sec hypothesis above) would have been needed over and over again (for a 30MB game just 10sec though) .... not sure that's why I ask.

[neodev]

Yes, that was considered. The "Flash" or "RAM" decision was a very difficult one, and that took a lot of thinking, as each one has its pros and cons.
Let me elaborate on it (Warning, long message ahead). I know the people from the "other forum" will read this and use it to throw some shit at us, hello pals .

First some numbers (feel free to correct me if some are wrong):

The largest neogeo game is kof2003, with 89MBytes (712MEGA in SNK size count), that is 1(P1)+7(P2)+64(C)+16(V)+0.5(S)+0.5(M)

The SD card interface of the STM32 runs at a maximum of 25MHz, when using 4 bit SD protocol that means 12.5MBytes per sec, so it takes around 7 secs to read the biggest rom SD, assumming zero CPU and filesystem overhead, and having the data contiguously stored in the SD so there is no SD commands overhead. That's pretty unrealistic as adding a filesystem (FAT32 for example) would cause this time to probably double as you must go back to the FAT each cluster, and this adds CPU, reading and SD commands overhead.

Writting 90MBytes in a 70ns PSRAM, assumming the write speed is also 70ns (it's usually lower) it's about 14MB/s, so it takes about 7 seconds to write

Writting the game to Flash memory is a value that we know from the NeoSD, and it's about 3:00 (+/- 10%) minus the time it takes to read from SD, but it's the total time what matters. Don't try to compute this value by adding the time it takes to erase + write each flash, as you'll get a much higher number, the NeoSD uses the large flash wait times to program another one, so the CPU is waiting for the flashing to end as little as possible, by issuing several flash operations in parallel.

So there are 2 main choices (the times there are for the largest games):

Use Flash roms (NeoSD):
Pros:
Instant boot time once flashed (just check a NeoSD on an AES with stock bios to see what instant means )
The BIOS can't tell it apart from a real cart, there is no extra code being run or trickery involved.
You can boot straight to the BIOS test menu and it will see the cart as a real one.
Once a game is loaded, you don't need the SD card in anymore, so you can have (expensive) single game carts, especially when using the "arcade mode" in NeoSD.
Cons:
Slow game change, as it needs to erase and rewrite the flash roms.
Limited graphics for the frontend, as they must be stored in a reserved area in the flash.

So this version gives a more "real" cart experience, and favors people that don't switch games often. Booting to the same game has 0 delay.


Use RAM (PSRAM):
Pros:
Very fast game change, as with the values computed above, and giving some room for overhead, it would take 20/25 seconds to rewrite the largest game (we will use this time as a base from now on)
More graphics for the frontend menus, as you can rewrite them every time you enter the menu
Cons:
You must wait up to 20/25 seconds every time the Neogeo boots. So boot to kof2003 10 times and you already waited the same time than flashing it.
You must write the game (or at least the P data) before the bios boots, otherwise it will show cross hatch so you have 2 options: either keep the board reset while you do that (having a ugly random colored squares screen, and it also won't work on all boards) or always boot to a small bootstrap that shows a "loading" screen (this will cause its own kind of issues with the BIOS though, especially with the test mode).

This version has a less "real" experience as you either wait, or see an interim screen before the game boots. This favors people that switchs games a lot. And booting to the same game several times in a row quickly approaches to the time taken to flash it. Also it might have some issues with BIOS as you aren't booting straight to the game code.

Then you have the hybrid version, mixing flash and RAM (for example P, S and M in flash, and V and C in RAM):
Pros:
As the P roms are in flash, the BIOS can't tell it apart from a real cart.
You can boot straight to the BIOS test menu and it will see the cart as a real one.
Flash times, although present, are shorter than the full flash version.
More graphics for the frontend menus, as you can rewrite them every time you enter the menu.
Cons:
When switching games, you must wait up to around 1 minute till the P, M and S flashes are written.
Every time the machine boots, you must wait for the C and V data to be written, that's 15/20s
Unless you want to show glitches, you must ensure C and V data is written before the game is actually ran so a small bootstrap and a "loading" screen will be needed.

You get the worst from the previous solutions, have to wait for the data to be flashed when switching games, and then again when booting the machine, and you can't boot straight to the game without a bootstrap screen.

So you can improve that mode a bit by making it "multislot", it's a bit misleading calling it that way, as it's very different on how the actual MVS multislot works so we may call it "multicart" better (or x-in-1), and I think this is the way DS & co cart works (I think, as noone has seen it apart from prototype images):

Pros:
As the P roms are in flash, the BIOS can't tell it apart from a real cart.
You can boot straight to the BIOS test menu and it will see the cart as a real one.
If you switch between the same set of games, you get zero flashing time.
You can quickly switch games in loaded set.
More graphics for the frontend menus, as you can rewrite them every time you enter the menu.
Cons:
You need space to hold all the games (at least the flash), adding more chips increases price and space and complicate the design.
When switching to games not in the loaded set, you must wait up to around 1 minute till the P, M and S flashes are written.
Switching to a different game in the loaded set is not the same experience than a real multislot board (probably reset when switching)
Every time the machine boots, you must wait for the C and V data to be written, that's 15/20s (more on this below)
Unless you want to show glitches, you must ensure C and V data is written before the game is actually ran so a small bootstrap and a "loading" screen will be needed (or hold the neogeo in reset)

So, this version is better than the plain hybrid version as long as you play the same x games, that is, it's good for somebody that likes to play the same set of games. Still it's not the real cart experience, it doesn't have instant loading as an actual cart does. It needs either reset or bootstrap loading (which could mess some BIOS routines). Switching games is not the real multislot thing either.
Also this adds more questions to the design:

It's clear that you need at least x times the flash size for the data that is stored in flash, but hoy much RAM should you add?

Enough for a game and then rewrite it when switching games? that would cause a game switch delay if you can't fit all the loaded games in RAM as the new C and V data load (say you have kof2003, samsho5, svc and mslug5 loaded, that would have a 20s delay when switching games).

Big enough to store the x games? say, if you have a 4-in-1 cart, you'll need a 2GBit RAM to store the C roms,and a 512MBit one to store V roms, then you can't use single chip PSRAMs, as they aren't that big, you'll need to either separate it in several chips (adding cost and size to the board) or use plain DRAM (that adds complexity to the fpga design, as it will need to handle DRAM quirks. All fpga manufacturers provide DRAM control blocks though).
Also, if you have RAM big enough for all games, do you load all games when booting? That would be up to 20 secs per loaded game, 1 minute if you have 4 big games loaded. If you decide to load just the first game, and then another one when switching to it the first time, this will not be better than the "only space for a game" version unless you decide to switch to a new game and then back to a previously played one.


tl;dr; The flash version was chosen because it gives the closer to original cart experience, with minor code interferences, zero boot time, although it has a higher game switch cost (that you can use to go to the fridge for a beer, so that's not bad ). If you enable Arcade mode, and remove the SD card so NeoSD no longer runs (you can't enter the menu that way), there is really no way to tell that game apart from an original running in ROM carts software wise.

[Rot]

xROTx