New Documentation: An authoritative reference on the YM2612

HardWareMan · Post by **HardWareMan** » Mon Apr 09, 2012 3:41 am

OK, I've finished it. This is my first experience of gluing pictures, so do not kick me. :3 So, this picture 24516x21770 pixels and 116MBytes. It's too big for FileDen, and all pichostings, so I placed it on filesharing service. Someone of you can rehost it to better place.

1. DAC is 8bit, not 9bit.
2. DAC seems to be one not only for 6 channels, but for left/right output too. But I don't get it: oscilloscope shows channel synchronization:

Perhaps they used the capacitance to store the current signal level. DAC place at upper right corner on picture. All analog pins can easily be traced to him.

The first step is to identify all the blocks in the picture. Interestingly, below the center at right side there are some matrixes, perhaps this is the operator unit and the calculation tables for the level of the envelope and sine. And they're seems 14-bit. On the left side above the middle is likely array of registers.

Learning, thinking and posting here.

PS Any part of picture can be recaptured with better quality. Source PSD file was 1,5GB. :3

TmEE co.(TM) · Post by **TmEE co.(TM)** » Mon Apr 09, 2012 9:49 am

I'm not very sure but I think the 2 big zig zags (capacitors ?) right next to each other under DAC part are part of channel switching. Final analog lines end up in the circuitry connected to them and there's some digital lines coming into the circuitry too...
There are 6 of something (triggers ?) that could hold panning settings for each channel... ?
It could also explain that when you overload one channel it affects the other.

HardWareMan · Post by **HardWareMan** » Mon Apr 09, 2012 11:08 am

TmEE co.(TM) wrote:I'm not very sure but I think the 2 big zig zags (capacitors ?) right next to each other under DAC part are part of channel switching.

Yeah, it could be capacitors. Or it could be power current amplifier. I'm not sure for now. But, there are lot of capacitors, that interconnect some power rails on die. Look at bidirectional pin:

And analog output pin:

Analog output has only one transistor, wich connected to AVCC. Thus, analog output can only source current, not sink. External resistor required.

And here those "zig-zags". Definitely this is a two transistors with some control circuit.

TmEE co.(TM) · Post by **TmEE co.(TM)** » Mon Apr 09, 2012 1:14 pm

I was just thinking... there not needs to be 2x DACs for both sides, only a switch that turns off sides.

Code: Select all

digital data -> DAC -+-> Switch -> L Output
                     |      ^
                     |      +--LR registers
                     |      V
                     +-> Switch -> R Output

HardWareMan · Post by **HardWareMan** » Mon Apr 09, 2012 3:58 pm

Interesting idea. This chip can't do smooth L/R panorama, you can simply enable or disable the output to the L/R channel, right?

TmEE co.(TM) · Post by **TmEE co.(TM)** » Mon Apr 09, 2012 4:36 pm

This is exactly what got me thinking this, especially since output is muxed not summed. If it was summed you'd need 2x DAC or have no sync between 2 sides with single DAC.

Nemesis · Post by **Nemesis** » Mon Apr 09, 2012 8:43 pm

Great work HardWareMan! Permanent mirror:
http://nemesis.hacking-cult.org/MegaDri ... tiched.jpg

Nemesis · Post by **Nemesis** » Mon Apr 09, 2012 10:46 pm

I haven't had too much time to look at the die shot so far, but one thing I can add is the function of pin 10. It was mentioned earlier in this thread that pin 10 is actually the "TEST" pin. So far, the only known function of this pin is that it returns the realtime "busy" state of the YM2612, the same as what can be read from the status register. I've confirmed this through hardware tests. See this post:
viewtopic.php?p=7409#7409
Pin 10 is also formally identified as the TEST pin in the official YM3438 manual.

It's likely that this pin can perform other functions, depending on the state of the test register. I don't know of any at this stage, but I haven't investigated it too thoroughly. Looking at the die shot however, it does appear that this pin can act as an input line as well as an output line...

HardWareMan · Post by **HardWareMan** » Tue Apr 10, 2012 3:07 am

Nemesis wrote:Great work HardWareMan! Permanent mirror:
http://nemesis.hacking-cult.org/MegaDri ... tiched.jpg

OK, applied to my post.

Nemesis · Post by **Nemesis** » Tue Apr 10, 2012 11:58 pm

HardWareMan was kind enough to supply the original psd, so I've uploaded it to my webspace:
http://nemesis.hacking-cult.org/MegaDri ... hed_PSD.7z
The file is massive, almost 700MB compressed. To be honest though, after comparing it with the jpg file, the two images are virtually indistinguishable. I don't think anyone will find any extra details revealed in the psd file that can't be seen in the jpg version, but here it is for the sake of posterity.

HardWareMan · Post by **HardWareMan** » Wed Apr 11, 2012 5:52 am

Exactly. JPEG compression was set above 10. Any thoughts about crystal?

TmEE co.(TM) · Post by **TmEE co.(TM)** » Wed Apr 11, 2012 9:51 am

I'd need a crash course in IC topology so I would know what kind of components am I looking at...

HardWareMan · Post by **HardWareMan** » Wed Apr 11, 2012 5:17 pm

Go for it! And I can proceed to SN76489AN (someone not sure about noise channel). :3

TmEE co.(TM) · Post by **TmEE co.(TM)** » Wed Apr 11, 2012 6:23 pm

What I want to see is 315-5478 / 5660 / 5700 / 5708 to finally settle why it has the least noisy output, plus the other tidbits that fix trailing noises and other oddities that only happen on YM2612 and not on YM3438...

Nemesis · Post by **Nemesis** » Thu Apr 12, 2012 1:22 pm

Yeah, it would definitely be interesting to see the embedded version of the YM2612/YM3438 on the Mega Drive 2, if nothing else just to see how "custom" it is, or if it is literally a "YM3438 in a box" if you will. Getting a good look at the DAC will also be quite useful, since we know it's a completely different DAC than in the discrete YM2612.