"new" SNES clone

Bregalad · Post by **Bregalad** » Tue Dec 12, 2006 1:10 pm

I'm curious. Link please as to when and where this was discovered?

Well, ask Anti-Resonance, he's the one with worked with all that stuff. I did just notices that samples 0, 1, 4, 6, 7 and 8 of Chrono Trigger sounds often different from emulator to emulator and from SPC player to SPC players. The basic decompression algorithm (often noted "old ADPCM system") did just not handle one thing in overflow correctly, causing some short samples to behave different when looped on themselves. When the samples overflows in the BRR compression, it's possible to turn the decompression in a pseudo-random number generator and having resonnant noise samples. Chrono Trigger uses it to its samples 4, 6, 7 and 8. They sound all like white noise, but with different kind of resonnance. On old emulators they sound just like normal tune (no noise), and more recent emultators/SPC players used various hacks so they sound just like perfect white noise, making them sounding all the same, while on the real hardware it isn't pure white noise, but 4 variants of different noises. I think Anti-Resonance worked hard to find the exact behaviour of this, and eventually a real algorithm is out. Chrono Trigger's sound effects still sound different on the last SNES9x than on my real SNES (while sounding much closer than older versions), and AlphaII SPC player seems to have everything accurate.

EDIT : To make things clear, I uploaded some evident WAV examples. They have been ripped from Chrono Trigger, when wind is played.

CT_wind_old uses the old innacurate method that don't support noise-self-looping samples. It's weird and doesn't sound much like wind. I'm pretty sure all dated emulators will render something like that.

CT_wind_fake uses a "hack" to support noise-self-looping samples, but in a bad and inacurate way. I think there has been many variants arround, but this one is from Super Jukebox.

CT_wind_real has been recorded from latest AlphaII SPC player and sounds like the real hardware. It's close from version #2 so Joe Gamer won't see any difference, but it actually sound more sofisticated noise than the one above.

Post by **tepples** » Tue Dec 12, 2006 1:41 pm

Third sample is 404 (not found).

Bregalad · Post by **Bregalad** » Tue Dec 12, 2006 1:43 pm

Well, don't be so fast. I always have problems typing down all things correctly (stupid case sensitiveness), so it takes me a few minutes to test and fix problems. It should be okay now.

dXtr · Post by **dXtr** » Tue Dec 12, 2006 2:00 pm

there sure was some difference there

is there any documentation released on these discoverys?

cheese007 · Post by **cheese007** » Tue Dec 12, 2006 3:49 pm

Jagasian wrote:How good is the pack-in controller compared to an official SNES controller? AFAIK, there is no longer any place to get new official SNES controllers... and that is the part of the system that does wear out over the years.

Same basic feel, but seem slightly cheaper. If you want an authentic feel pop the boards inside into an SNES controller shell since the normal SNES controller does not function properly with NES games.

gannon · Post by **gannon** » Tue Dec 12, 2006 4:46 pm

cheese007 wrote:since the normal SNES controller does not function properly with NES games.

Why not? It should just be that the buttons on the right are remapped.

cheese007 · Post by **cheese007** » Tue Dec 12, 2006 7:14 pm

Haven't tested it myself, it's just what I've heard.

Post by **Memblers** » Tue Dec 12, 2006 8:34 pm

Speaking of controllers, I have a Gamestation (thanks to Lik-Sang sponsoring y2kode when they were still around and cool, fuck Sony BTW), which is probably the same clone chipset mostly, and the only thing that really gets me is that it won't work with 2 genuine SNES controllers.

I can plug the "Fighting Back" controller that came included with it into either port, and at the same time I can plug a normal SNES controller into either port. But if I plug normal SNES controllers into both ports, it goes haywire when you press any button and acts as if every button is being pressed at once.

Any idea why?

Post by **tepples** » Tue Dec 12, 2006 9:51 pm

If the B button makes all buttons go at once, the 'strobe' line isn't getting through.

cheese007 · Post by **cheese007** » Tue Dec 12, 2006 9:52 pm

That's interesting. I have no clue why it would do that, even if it's a clone.

gannon · Post by **gannon** » Tue Dec 12, 2006 9:58 pm

tepples: Yep, found that out while soldering to many NOACs when tracing the controller lines wrong

Don't see why plugging 2 official controllers in at the same time would somehow mess with the strobe lines though... unless the controllers are bridging lines that the pirate ones aren't and that's screwing it up.

Post by **tepples** » Tue Dec 12, 2006 10:02 pm

Or perhaps one controller is holding its output low/high after the transaction has completed and the other is 3-stating it. Or perhaps the other data lines (corresponding to NES D3/D4) react differently. Do the unlicensed controllers work on a Super NES brand system?

Bregalad · Post by **Bregalad** » Wed Dec 13, 2006 10:53 am

dXtr wrote:there sure was some difference there
is there any documentation released on these discoverys?

Unfortunately not, else I would have clarrified things better for a while.
There is only one BRR decoding doccument, and I cannot found it anymore unfortunately. It states the algoritms to decode in 4 different decoding modes. One decoding mode is just normal PCM, the second one is a variant of DPCM wich feedback on the previous value to find the next sample, and the two last are variants of ADPCM wich feedback the last two previous values. There is some formulas to decode samples, and those are valid in all cases. However, the problem comes from clamping the samples to 16-bits, and to samples that overflows from the 16-bit ranges at the same times they loop on themselves making the feedback behaving differently on the second loop than on the first, and this for infinity can crate weird noise samples. If feedback looping and clamping isn't emulated 100% accurately, you get very various results. Without counting that some emulators actually used hacks to detect this and create noise when this is detected artificially to prevent sound effects sounding completely wrong.

The problem for ALL SNESdev related stuff is the serious disorder in all doccumentations and the lack of a "big" site where any SNESdev stuff is odered in. Also, SNESdev people know a lot of things that have never been doccumented aside being doccumented unproprely on forums, and a lot of very old doccuments stating very wrong information as being right. It's even worse thant GBAguy's tutorial for NES, because there is even so called official doccumentation wich states wrong things, and people just think it's a reference because it's official, and are getting things all wrong.

Post by **tepples** » Wed Dec 13, 2006 11:12 am

Bregalad wrote:The problem for ALL SNESdev related stuff is the serious disorder in all doccumentations and the lack of a "big" site where any SNESdev stuff is odered in.

So would it be a good idea to make a wiki documenting the Super NES, much like Quietust's NES wiki?

Bregalad · Post by **Bregalad** » Wed Dec 13, 2006 1:22 pm

Sure, it would be great, but only if the info in it would be confirmed to be accurate.