Testing sound of different MSX sound chips and their FPGA implementations

Page 6/9
1 | 2 | 3 | 4 | 5 | | 7 | 8 | 9

By Pencioner

Hero (634)

Pencioner's picture

26-05-2018, 23:20

Alexey wrote:

Manuel, what tunes did you use to make those mp3s? I would like to record the sound with the new Carnivore's firmware. We also optimized the filtering for PSG and SCC.

@Alexey, the tunes are the same i used for tests, there's references in my posts with tests

By Pencioner

Hero (634)

Pencioner's picture

13-06-2018, 00:09

I have an update for PSG tests - added Zemmix Neo recording, updated Carnivore 2 and MFR recordings (those both were flashed with newer firmwares)

By Parn

Champion (304)

Parn's picture

13-06-2018, 14:49

Pencioner, why did you made the new recordings at 44.1kHz (as opposed to the previous recordings at 48kHz)? Just curious, I think it won't be of much difference.

My personal ranking:

  1. Yamaha YIS-503: Truly a benchmark. Sounds unbeatable.
  2. Zemmix Neo: Surprisingly good, despite loud noise.
  3. MegaFlashROM SCC+SD (New): Volume seems a bit unstable, perhaps it isn't getting enough power? A bit weak on the bass but pleasant nonetheless.
  4. Panasonic A1WX: Sounds muffled due to excessive filtering, and hollow, which is weird and I have no idea why.
  5. MegaFlashROM SCC+SD (Old): Weak on the bass and some distortion. If not for the distortion I think it would sound better than the Panasonic.
  6. Carnivore 2 (New): Sounds good but too muffled for my tastes. Still, much improved IMHO.
  7. Carnivore 2 (Old): Distorted and muffled, I really didn't like it TBH.

Thank you very much for the recordings! They were most informative. Smile

By Pencioner

Hero (634)

Pencioner's picture

13-06-2018, 16:01

Parn wrote:

Pencioner, why did you made the new recordings at 44.1kHz (as opposed to the previous recordings at 48kHz)? Just curious, I think it won't be of much difference. Thank you very much for the recordings! They were most informative. Smile

Uhm, first recordings done on old laptop while my Intel NUC was in service and after i got it back i had to reinstall OS, and i forgot to change settings, but i agree there wouldn't be much of difference.

My personal oinion about new recordings:
1) C2 misses some high frequencies, so sounding muffled, but when i add 1-1.5 dB on high band on mixer's EQ it sounds much better than other implementations (including Zemmix Neo)

2) MFR sound improved, but when i looked again on the wave (i was curious about envelope) - surprisingly it was the same straight line. Which makes me think that file at this comment contains the output of internal Panasonic's chip by mistake

3) I wonder if somebody can make PCB layout for Zemmix Neo in respect of audio quality (i mean having much less noise). I'd definitely buy one even if it costs more

P.S. I've performed hearing experiment, routing the sound of real chip to left channel and output of C2 (and then MFR, with patched vgmplay to send commands to both chips at same time) to other channel, and listen tunes in headphones. With this setup you don't notice lack of high freqs, and i was suprised how similar C2 and real PSG are sounding together. Almost no glitches. So i think that if guys improve audio part of schematics (to get more high freqs), it will rock.

By Guillian

Prophet (3139)

Guillian's picture

13-06-2018, 16:57

Pencioner wrote:

2) MFR sound improved, but when i looked again on the wave (i was curious about envelope) - surprisingly it was the same straight line. Which makes me think that file at this comment contains the output of internal Panasonic's chip by mistake

It was recorded using a modified version of VGMPLAY. The PSG is accessed using ports #10-#11, so the sound of that file was produced by the MFR SD.

Anyway, I have cheched again and without the MFR SD I get no sound. With the MFR SD I get this wave:

So it seems that wave form depends on the computer mixer. Bear in mind that MFR SD is designed to output its sound through the computer.

It could be nice to get a raw sound output from original PSG before it passes the computer's sound mixer.

By Pencioner

Hero (634)

Pencioner's picture

13-06-2018, 17:52

Guillian wrote:
Pencioner wrote:

2) MFR sound improved, but when i looked again on the wave (i was curious about envelope) - surprisingly it was the same straight line. Which makes me think that file at this comment contains the output of internal Panasonic's chip by mistake

It was recorded using a modified version of VGMPLAY. The PSG is accessed using ports #10-#11, so the sound of that file was produced by the MFR SD.

Anyway, I have cheched again and without the MFR SD I get no sound. With the MFR SD I get this wave:

So it seems that wave form depends on the computer mixer. Bear in mind that MFR SD is designed to output its sound through the computer.

It could be nice to get a raw sound output from original PSG before it passes the computer's sound mixer.

This time it didn't pass computer mixer at all, as i soldered the connector and cut of the pin 49 - on photo MFR is on top (i just didn't glue over new sticker yet, and many thanks for the new case and sticker!), then MMM, then flashrom on real SCC, then Carnivore 2:

So i just can't explain why i get no envelope again... (are there any hardware differences between the cartridge from 2013 and newer ones?)

By Guillian

Prophet (3139)

Guillian's picture

14-06-2018, 08:54

Pencioner wrote:

So i just can't explain why i get no envelope again... (are there any hardware differences between the cartridge from 2013 and newer ones?)

The FPGA is an older revision, but that should not affect the sound output.
IIRC, your cartridge has the "sound fix" for the Philips 8250/55/80. Do you get the sound before the fix or after the fix?

By Pencioner

Hero (634)

Pencioner's picture

14-06-2018, 10:22

Guillian wrote:
Pencioner wrote:

So i just can't explain why i get no envelope again... (are there any hardware differences between the cartridge from 2013 and newer ones?)

The FPGA is an older revision, but that should not affect the sound output.
IIRC, your cartridge has the "sound fix" for the Philips 8250/55/80. Do you get the sound before the fix or after the fix?

I pick sound right before it reaches the MSX - pin 49 was cut off, and there is good soldering point just before the pin. Soundfix could affect it, though. Maybe you can guide me with some photos where should i solder to get it before? I might try to carefully open the cartridge (i think it is possible without breaking anything if i am patient enough) and solder to another point, because there's no reason for any sound fix now.

UPD: i have file from other MFR (thanks @sdsnatcher73 for your help!) and i can confirm that this issue with envelope presents only on my cartridge (mostly because of that Philips fix). So i think it is worth trying to open cartridge and resolder that connector before it is being "fixed" Smile I'd like to get some guidance, @Guillian

By Guillian

Prophet (3139)

Guillian's picture

14-06-2018, 13:34

It is difficult to open the case and not break it. It depends on how much tight are the internal locks. Probably they will break.

The sound output trace from the FPGA goes to two pads, labeled as "SND". In those pads there are a resistor and a capacitor (that is the sound fix). And the "output" trace from the soundfix goes to pin 49.
So you have to take the sound from the first SND pad.

By Pencioner

Hero (634)

Pencioner's picture

14-06-2018, 15:32

Guillian wrote:

It is difficult to open the case and not break it. It depends on how much tight are the internal locks. Probably they will break.

The sound output trace from the FPGA goes to two pads, labeled as "SND". In those pads there are a resistor and a capacitor (that is the sound fix). And the "output" trace from the soundfix goes to pin 49.
So you have to take the sound from the first SND pad.

Wohoo! It works! And envelope is finally back! I will update test with no-fix recording sooner.
Btw, i was lucky to open case without breaking it - one of internal locks were more soft, to lift up other i helped with screwdriver, and was doing very slow, and the missing sticker was of help - because i could free that top latch so both sides of case can move in parallel, with no tilt. Some photos later. And I made a mistake of first removing the soldered cable, i could do it with less soldering effort - solder out the capacitor+resistor and make a bridge with solder drop between both sound pads so it goes through. Missed that oportunity, but nevermind, now it sounds much better (doesn't have that specific distortion which was ugly to my hearing)

Really, the "fix" (very intentional quotes here) might fix the Philips volume but it screws up sound...

Page 6/9
1 | 2 | 3 | 4 | 5 | | 7 | 8 | 9
My MSX profile