YM3438 in a Model 1 Genesis

So, the YM3438 has been quite an interesting topic for use on the Model 1 Genesis. Basically it's the same chip used in the model 2 Genesis, however the real 3438 isn't "as good".

There are some differences compared to the 2612 though. On some games like Adventures of Batman and Robin, the difference can be very much heard on the beginning of Tea Time's soundtrack.

Videos of the experiment.

Just dropping it in:
https://www.youtube.com/watch?v=sSWMr4rjfd0

The easy method for a solution (better cap values have been revised):
https://www.youtube.com/watch?v=lu9OgXuzA6E

Better method (better resistor values have also been used since this video was filmed):
https://www.youtube.com/watch?v=POPzgIzVSpg

However, there's some bugs with the 3438's timers that cause glitches in games:
https://www.youtube.com/watch?v=8hqVWYwbZH8
https://www.youtube.com/watch?v=VBEkqSqgOL8

So, after all this is said and done, this looked like a lost cause.

However, there may be hope, and it is basically by fooling the Genesis. This was suggested by a friend that used to be from the vintage computer forum. He reccomends to disconnect the 2612's MOL and MOR pins from the Genesis (obviously), and piggy back the 3438 on top of it, and attach it's MOL and MOR pins through the caps and resistors then to the Genesis motherboard, along with taking the 3438's /WR pin to VCC and /IRQ pin leaving connected to nothing. This basically makes the Genesis use the 2612's timers, but the 3438 just sits there and listens to the music data it should be playing.

Will it work? I have no idea, but I'm willing to try!

Awesome idea, looking forward to this!

Ok so what was originally suggested didn't work. So I consulted with TmEE, and the pin that actually needs disconnecting on the YM3438 is the TEST pin, and /RD pulled to VCC.

I actually found however that simply just also leaving /RD disconnected worked too, and kept the mod cleaner.

Oh, and the results? It works! YM3438 fixed DAC and capabilities, using YM2612 timers. And Sonic Spinball music works, so it was the timers like I had originally suggested was the problem all along.

Pictures:
https://img.photobucket.com/albums/v395/Evilweredragon/IMG_1940_zpsbc97d7bd.jpg
https://img.photobucket.com/albums/v395/Evilweredragon/IMG_1941_zps3c1bafa2.jpg

Sound:
http://www.mediafire.com/?jvy26f26bui7ghf
http://www.mediafire.com/?r347t55yhal35kh
http://www.mediafire.com/?xk8hm64w0rszyv4

I really like the bass response it has in the first sound clip.

Really awesome! Going to eventually post a schematic?

Also I hope you can eventually upload the audio samples to youtube.

Drakon wrote:

Really awesome! Going to eventually post a schematic?

Yes, but it'll be difficult. Since it's piggy backed, and it's rather hard to show piggy backed chips in a 2D schematic.

I was thinking just show which pins aren't piggy backed and what to do, the rest you don't have to draw.

As posted on Sega 16.. As I feel this is really the proof that shows the 3438 does have a difference and it's good, but I have to admit, not the "biggest" change unlike a model 2's ASIC version of the 3438. But the stereo separation and the slight DAC improvements are a nice change for a model 1.

First though, here's a rough AB comparison between the YM2612 and YM3438, with no music, select sound effects only. Ignore the noise levels please, I used a VA4 Japanese Mega Drive for the YM2612 test, and it has better filtering from VDP noise, as the one that's customized right now is a VA3, which isn't as good in VDP noise filtering. Also, since this VA3 Genesis is also overclocked, I turned off the overclock and had it running at stock speed.

YM2612: http://www.mediafire.com/?9ioimgqx88rrgo0
YM3438: http://www.mediafire.com/?y19jpqfbdjazl4u

YM3438 playing Mystic Cave Zone: http://www.mediafire.com/?ikoosno8gzr7npk

The MP3's are constant bit rate, 320KBps and true stereo.

Oh, and that stereo separation? If you use Audacity, you'll find I was telling the truth, as it's right in the wave form. YM3438 is better at stereo.

YM2612:



YM3438:


Enjoy!

Wow! Have you run into any bugs or other weirdness in any games ever since disconnecting the test pin?

How strange that the YM2612 has that fuzzy stereo bleed going on and the YM3438 is so much better at reproducing stereo sound.

evildragon wrote:

Yes, but it'll be difficult. Since it's piggy backed, and it's rather hard to show piggy backed chips in a 2D schematic.
I was thinking just show which pins aren't piggy backed and what to do, the rest you don't have to draw.

Looking forward to this

Well it's my understanding that the test pin is simply for timer information. Since we don't want to use the YM3438's timers, that's the theory for removing, and keeping the YM2612's test pin connected, since we are in fact using that.

I stillhaven't done a schematic yet, I really don't know how to draw this schematic, since it's piggy backed, but I'll see what I can do. I may need help with this though.

Just take a picture of the chip from the datasheet and trace where the re-routed pins go.

But it's not really that hard. The YM2612 has all of it's pins attached normally, except MOR and MOL removed (or lifted).

YM3438 is soldered directly on top, with it's TEST pin lifted or removed, and RD pin attached to VCC. Then the YM3438's MOR and MOL pins through a cap and resistor, then to the genesis (where it normally goes).

evildragon wrote:

But it's not really that hard. The YM2612 has all of it's pins attached normally, except MOR and MOL removed (or lifted).

YM3438 is soldered directly on top, with it's TEST pin lifted or removed, and RD pin attached to VCC. Then the YM3438's MOR and MOL pins through a cap and resistor, then to the genesis (where it normally goes).

Yeah exactly, it's not hard to draw a schematic haha. Makes it easier so people don't have to hunt down which pin is which. Also people will want to know the mixing circuit.

There's so many crossing pins I am sorry, I am not a schematics artist.

Here's the pins.



/NC number 10 is the /TEST pin.
And here's the mixing circuit I used.

Code:

MOL ----------| (-------/\/\/---------
MOR ----------| (-------/\/\/---------

Capacitor is 10uF, resistor is 240 ohms.

That's good enough, much appreciated!

Here, I spent the past couple hours drawing this out. I had the ohms wrong in the previous post.

Awesome, didn't know you connect rd to vcc.

Drakon wrote:

Awesome, didn't know you connect rd to vcc.

ahem,

Quote :

and RD pin attached to VCC

I said that earlier. lol

evildragon wrote:

Drakon wrote:

Awesome, didn't know you connect rd to vcc.

ahem,

Quote :

and RD pin attached to VCC

I said that earlier. lol

Yeah I never read it earlier. Just more proof that a schematic makes it easier to see things.

Kinda lazy though don't you think, to not read it? I mean, I worked with TmEE completely through words and even I figured it out.

Anyway.

Here's a better audible sample that shows how much clearer the Sonic 2 logo's twinkle is.

STOCK YM2612: http://www.mediafire.com/?qyana9z2tt6y6ld

MOD YM3438: http://www.mediafire.com/?arnancoja9jnknw

I've ordered a YM3438 Your schematic and samples have been quite helpful, evildragon, thanks again.
I'm going to throw a DPST switch in so I can flip between chips on the fly.

Toggle switch would work you're only changing 2 connections.

Welp, pretty sure I just blew my 1145 doing this. Maybe some other components too.

Shame too, I just got the audio circuit all working with an op amp, rather than the 1034. Worked great.
Mixed up the ACC and AVCC so that the AVCC and RD were connected. Poof, lots of smoke. Dead.
I should be grateful, I suppose... it's been a long time since I've blown something up.

Always remember to triple check your work before you plug in your VCC, kids.

FAACK

That sucks. I always use a red wire for vcc.

Yeah... after messing around with the op amp situation for a good couple hours and getting it all working, I thought I'd just quickly toss in the YM3438 with the DPST toggle switch while I was at it.

So close to finally leaving that thing sealed up for good. New one's on the way :P

Got it all done.
Was having the same crappy impeded noise as on the beginning of this clip: http://www.mediafire.com/?j9scynj7kmll9n1
(Becomes an issue at 0:03 once the game switches on the chip)

Used 100uF caps instead of 10uF and I found it helped reduce the noise. Still isn't completely gone, but it's pretty unnoticeable when music and SFX are playing.
Does anyone know what might be causing this? It seems to change level and frequency, depending on the game.

Maybe a decoupling cap would help?