Discrete Vs. Integrated YM3438: Sega Genesis music blind test results.

very cool work! i would like such a test for cd-audio reproduction of psx, saturn, 3do, cd-i... I'm sure there would be some really interesting results

BTW have you checked the recordings on audacity or any other audio editor to check for visible differences in output volume, stereo sepratio and such?

 

Science bitches!

 

i have a model 3, but haven't opened it yet to see which version it is. there are ways to restore back 32x and sms compatibility but haven't seen a hack for mcd yet. even if it was done, plugging a mega cd would be very hard physically. genesis 3 with a 32x plugged looks very sad (i feel like genesis 3 is in pain), i can't imagine a plugged mega cd to that poor little genesis 3.

unfortunatley genesis 3 in stock audio is mono and the hack to get stereo sound isn't easy. so ccam looks like the best way. wonder what the hell sega was thinking while making this beauty mono.

 

Um...

Sega didn't make it. They licensed it out to Tec Toy for making. That should explain all...

 

Wow, I gotta say those are some interesting results. I would agree with the sentiment that for the most part it's indiscernible. I only came to my conclusions by listening MANY times.

 

Ok, for those who still maybe wondering, and because I did not really talk about it in the post;

Considering that the acoustic properties of the different version of YM3438 are very similar, the main criteria of selection should be background noise.
There are notable differences at this level.

My rundown goes as follow.

(best) M2VA4 or early Model 3 > Discrete Ym3438 on model 1 > Early model 2 > Model 1 VA7 (worst)

I will get you recording samples and waveform pictures of background noise when I get back to home later today.

 

Ok because you guys obviously LOVE audio comparisons, I have some more quality data for you.

Stock Sega Genesis Model 1 recordings​

Firstly, many of you expressed the wish to have recordings of the Original Stock Model 1 High definition graphic Sega Genesis to compare with the CCAM recordings in the blind test.

So here it is, minus Turrican Intro, which I do not have.
Those have been recorded using the same exact methodology used for the blind test.
Contra and D&D are from a VA6.
Sonic 2 and ThunderForce are from a VA3 and the console was connected to a CRT because it is an older recording I made.
For a while now I have stopped making recordings of consoles when connected to a CRT tv. I now know this will induce a low volume high pitch feedback in the sound.
The effect is relatively minor though. See if you can detect it in the sound

DnD-Valley1-M1VA6.flac
TFIV-levela1-VA3(postrecap).flac
CHC-Gamestart-M1VA6.flac
Sonic2-Emerald1-VA3(postrecap).flac

*I will let you guys discuss observations you make before commenting on my own

*More of my recordings can be found here : http://www.sega-16.com/forum/showth...ck-consoles-mods-and-more&p=591943#post591943

Comparative Study of the Background noise in different YM3438 equipped Sega Genesis consoles.​

In this test we have 4 consoles, which are representative of most YM3438 setups.

-Model 1 VA7 : FC1004 Asic model 315-5487 ( Afaik this Asic is exclusive to the VA7)
-Model 2 VA1 : FC1004 Asic model 315-5660 ( Asic also found in M2Va0,Va1,Va1.8,Va3 and X'eye)
-Model 1 VA3 : Equipped with discrete YM3438 ( same machine used in the blind test )
-Model 2 VA4 : FJ3002 Asic model 315-5960 (M2Va4, M3Va1)

Please download this file containing a short sample of background noise for each of these. Total download is about 2mb
https://docs.google.com/uc?export=download&id=0BzsnL20-4a37MEwyRElFVU1wX3M

These samples are all of the same exact snippet of the Mega Turrican Intro. I left parts with actual sound so you can estimate how loud is the noise compared to the audio. Sound levels are normalized so the audio level is consistent from one sample to the next.
Mega turrican is quite representative of your average Megadrive game regarding sound level. Some games are louder, some are quieter. Naturally, the quieter the actual audio is, the more the noise will contrast.

Those are all recorded with the same CCAM PCB. However please note that they were all recorded with the Stock RetroGamer CCAM amp, which comes with a lowpass filter ( 0.28uF ceramic caps to ground ) to make it sound closer to a Stock Model 1. I do not have all these samples recorded with the unfiltered CCAM used in the blind test sadly. However this should do perfectly well to compare the relative magnitude of the signal to noise ratio across those different consoles.

As far as i know I am missing at least the following 2 ASIC YM3438 but I do not have any machines containing them so I cannot test.

-FF1004 Asic model 315-5700 : Possibly the rarest ASIC, Found in the Sega Nomad and in some model2 VA3
-FQ8007 : found in the late Genesis model 3 (M3Va2)

Now lets look at the waveforms ​

I am sure you can tell the difference simply by listening to the samples, but its cool to also check the waveforms.

These are the very waveforms of the Flac samples for download in this thread.
For each file there is a full view of the waveform in stereo as well as a super zoom of the Left channel.
In each case the Super zoom consists of 500ms of audio from the same region between 2.5-3s of the audio file.
The vertical scale is also the same. Maximum displayed amplitude is 1 db in normal view and 0.1 db in super zoom.

I took a measure to partly fool proof these results and made recordings of a Model 2 VA3, which also uses the 315-5660 and the SNR is consistant with the M2VA1.

Model 1 VA7 : FC1004 Asic model 315-5487




Model 2 VA1 : FC1004 Asic model 315-5660




Model 1 VA3 : Equipped with discrete YM3438




Model 2 VA4 : FJ3002 Asic model 315-5960




60hz multiples ?