New Documentation: An authoritative reference on the YM2612

AamirM · Post by **AamirM** » Sat Oct 18, 2008 6:50 am

Hi,

@ TmEE
Thanks a bunch for posting that recording. Now....

@ Snake
In that Spiderman recording, notice that a intrument is played using channel 3 (alternating between left and right channels) has quite low level compared to all other emulators (except Kega) and I would like to know how do you do that?

stay safe,

AamirM

Lord Nightmare · Post by **Lord Nightmare** » Sun Oct 19, 2008 12:48 am

In response to ym2612 versions, there appear to be 7 variants of the MD system *in terms of ym2612 and VDP*:
MD1 VA0 thru VA6: use a real yamaha ym2612 and a real 315-5313 VDP (note that va0 and va4 (and likely the intervening models, but please correct me) have 3 asics, while at least va5 and va6 have only two, one of which is the vdp)
MD1 VA7 thru MD2 VA1, plus MD2 VA3: use a FC1004 ASIC with ym2612 and VDP functionality integrated. The amstrad mega-pc also uses this same ASIC.
MD2 VA2: need more info on this
MD2 VA4/MD3 VA1: uses 315-5960 asic, need more info on where the ym2612 is hiding
MD3 VA0: need more info on this
MD3 VA2: uses 315-6123 "FQ8007" asic, need more info
Nomad(and megajet?): uses a 315(or is this 513?)-7000(or is it 5700?) ASIC with ym2612 and VDP integrated (as well as the z80?)

So for sound analysis, we need to consider all of these 7 (and there are likely more) as separate.
Most emulators probably should try to emulate the original MD1 config with the separate ym2612, alternately choosing the FC1004 asic which has a 'cleaner' ym2612.

Edited 10/20/2008, thanks to Tmee for corrections

LN

SmartOne · Post by **SmartOne** » Sun Oct 19, 2008 3:09 am

Wow, but do those four really sound that different?

TmEE co.(TM) · Post by **TmEE co.(TM)** » Sun Oct 19, 2008 9:17 am

Real YM2612 sounds one way (poorer), and ASIC variants all sound same. I have systems with all 3 in them, and no difference whatsoever. My VA1 MD2 ( http://www.hot.ee/tmeeco/MYSMD2.JPG ) has 315-5660, the VA1.x plays no role to what's in the machine, 5487 is old, 5660 is newer, and 5700 is newest from these types of ASICs. MD2 VA2 uses discrete YM2612, and different ASIC, MD2 VA4 uses same setup as one of the 2 known Genny3 setups. I know nothing about MD2 VA4 nor Genny3s.
The last Genny3 setup, or maybe the one also used in MD2 VA4, has some things altered in them which causes incompatibilities with few games, and around around 10% US SMS games have some glitches (so I've been told).

Snake · Post by **Snake** » Sun Oct 19, 2008 11:42 am

TmEE co.(TM) wrote:Real YM2612 sounds one way (poorer)

Blasphemy!

Certainly, with your audio mod, an MD2 sounds better - but without it - ugh.

But with a similar audio mod I think you'd find a real YM2612 doesn't sound much different, if it all.

TmEE co.(TM) · Post by **TmEE co.(TM)** » Sun Oct 19, 2008 12:46 pm

It makes things nice and clear, but the sheer amount of the "ladder effect" noise which is otherwise filtered out ruins all experience for me... it sounds like a damn 6-bit WAV (not as bad, but you get the idea...).

HardWareMan · Post by **HardWareMan** » Sun Oct 19, 2008 2:46 pm

TmEE co.(TM) wrote:It makes things nice and clear, but the sheer amount of the "ladder effect" noise which is otherwise filtered out ruins all experience for me... it sounds like a damn 6-bit WAV (not as bad, but you get the idea...).

Indeed. YM2612 "ladder" effect was clearly hear at very low volume on every FM channel (did not tested it for DAC). But I had MD2 with integrated YM2612+Z80+M68000+Arbiter+IOPort in one big chip. It did not produce this "ladder" effect but it did mute some very low volume sounds! Very noticeable cut off some weak sounds.

TmEE co.(TM) · Post by **TmEE co.(TM)** » Sun Oct 19, 2008 3:13 pm

Some cutting happens, but at these volumes, its hard to make out what you're listening anyway...
My MD2 has 68K, Z80 and all RAMs separate... also, PSG carries VDP activity noise

Lord Nightmare · Post by **Lord Nightmare** » Mon Oct 20, 2008 2:36 pm

I've updated the post above with the systems list with tmee's corrections, and to include late MD2 and MD3. Please add comments.

Also, which MD2 has the z80 inside the giant asic? VA4?
Thanks to tmee's correction, I know all the va1.x ssytems are really 'VA1' systems regardless of the second number, but I'm still missing a couple of md2 systems:
FC1004 ASIC chips: (208 pins) 315-5487, 315-5487-01, 315-5487-10, 315-5660, 315-5660-02, 315-5700, all are supposedly interchangable
MD2 VA0: FC1004, separate? z80
MD2 VA1(.x) (and x'eye and multimega): FC1004, separate zilog z80 or 315-5676/315-5619 which seems to be a z80
MD2 VA2: NEED INFO on this system, preferably a pcb picture. tmee says it has a separate ym2612, a non-yamaha version?
MD2 VA3: FC1004, ? z80
MD2 VA4: 315-5960 asic (160 pins), may have z80 on it? do these systems have separate yamaha chips or is that inside an asic?
MD3 VA0: NEED INFO on this system, preferably a pcb picture
MD3 VA1: 315-5960 asic (160 pins), may have z80 on it? do these systems have separate yamaha chips or is that inside an asic?
MD3 VA2: 315-6123 FQ8007 (208 pins, not compatible with FC1004, contains z80?)

LN

TmEE co.(TM) · Post by **TmEE co.(TM)** » Mon Oct 20, 2008 6:02 pm

315-5487, -5660, -5700 all are VDP+I/O+YM2612+PSG. 68K, Z80 and RAMs are separate on these systems. 5660 and 5700 are definitely pin compatible, 5487 uses same VRAM chips as what are in MD1 and I have no idea if they're compatible with the single VRAM chip used in later designs. All should be considered same.

MD1 VA0...VA4, all have no TMSS, all have 2 part I/O controller, everything is with discrete chips. VA3 has one small extra digital chip in it, which function is unknown to me, possibly related to clock signals... earlier models possibly too.
MD1 VA5 and VA6 are same as previous models, but have single chip I/O controller and also have TMSS.
VA2 MD2 has discrete YM2612, which is made by Yamaha. Goes under MD1 category.

315-5950 has everything but VRAM and 68K RAM integrated in it. http://www.sega-16.com/forum/showthread.php?t=5079
All newer designs should be like that chip, with everything integrated...

YM2612_fan · Post by **YM2612_fan** » Tue Oct 21, 2008 1:50 am

TmEE co.(TM) wrote:315-5487, -5660, -5700 all are VDP+I/O+YM2612+PSG. 68K, Z80 and RAMs are separate on these systems.

What about the 315-5708 ASIC? I don't see a separate Z80 on my board (MD2 - VA1,) unless the KM428C64J-10 is a clone.

TmEE co.(TM) · Post by **TmEE co.(TM)** » Tue Oct 21, 2008 12:01 pm

VA1 boards have to have Z80 (I've only seen clone chips on all units I've had access to), 68K and RAMs... The Z80 is a small square chip near reset switch. I think it had 40 or 44 pins.

5708 is a 5660 and 5700 compatible then... how many variants Sega has !?

Eke · Post by **Eke** » Wed Oct 22, 2008 1:43 pm

I've fallen accross this picture recently and made me think about this topic:

http://img523.imageshack.us/img523/4518 ... ordhf9.jpg

apparently, the Amstrad MegaPC had a YM3812 inside (instead of YM2612 ?) and 315-5487 (bus arbitrer + VDP/PSG ?)

Shiru · Post by **Shiru** » Wed Oct 22, 2008 3:27 pm

Eke wrote:Amstrad MegaPC had a YM3812 inside (instead of YM2612 ?)

YM3812 is OPL2, it's not compatible with YM2612.

TmEE co.(TM) · Post by **TmEE co.(TM)** » Wed Oct 22, 2008 3:40 pm

315-5487 is the first VDP+I/O+PSG+YM2612.
The OPL2 was in the MegaPC for Adlib compatibility.