Viewing Issue Advanced Details
ID Category [?] Severity [?] Reproducibility Date Submitted Last Update
01065 Sound Minor Have not tried Feb 9, 2008, 08:03 28 days ago
Tester Kold666 View Status Public Platform
Assigned To couriersud Resolution Fixed OS
Status [?] Resolved Driver zaccaria.cpp
Version 0.113u3 Fixed in Version Build
Summary 01065: monymony: Some bugs of sound.
Description Just for documentation I recorded the sound from my Money Money PCB. I played the first loop of the game.

Lord Nightmare described the bug. I can break this into three bugs:

1: one of the output channels of the ay-3-891x has a filter attached to it, and ONLY one. This produces the sawtooth-like sound on one channel.
Edit: after reading the sourcecode in zaccaria.c, i realize it isn't ONE channel, but ALL the channels which have filters! its just most noticible on the one 'tromba' channel.
I don't know if MAME even SUPPORTS having filters on SPECIFIC OUTPUTS of the ay-3-8910 at all! In fact, the mixing method of the ay-3-891x should be emulated too, since theres two ways to mix the channels: the 'right' way, by having a seperate buffer on each channel and then using some op-amps or a mixer IC to add them together, or the 'wrong' way, like the MSX does with its ym2149 (an ay-3-8912 clone, which is actually all that MAME emulates right now. yes, there ARE differences, and yes, MAME doesn't properly emulate a real ay-3-891x at all!) is to tie the three outputs together right outside the chip, which results in some downright strange amplitude crossover mixing issues. The FPGAArcade ym2149 vhdl implementation emulates this screwy mixing using a 3-d filter table based on measurements from the real ym2149.

2: the tms5220 has some bugs in it, which screw up the voice. I'm working on fixing this, but the tms5220 code really needs to have a good amount of it rewritten. Jarek B. did a nice job of this when he wrote the tms5110 core based on it, but he didn't port back a lot of the fixes he made to the tms5220 code. Current problems include the interpolator being implemented wrong (I think this is true for the tms5110 too, actually), the FIFO being implemented in a suboptimal way with a hack to clear it in one place (I'm working on replacing it with a new one based on Jarek's code right now), and possible errors in the lookup tables.

3. The global filtering for all chips is not implemented. We need the schematics to do this, and we also may need Derrick Renaud to bite the bullet and implement multiple-stream-input for the discrete core to do this right.
Edit: apparently nicola has a copy of the instruction manual from this game. hopefully I can get the schematics from that.
Steps To Reproduce
Additional Information
Flags Verified with Original
Regression Version
Affected Sets / Systems monymony
Attached Files
aac file icon monymony0113u3ora.aac (882,120 bytes) Feb 14, 2008, 17:57
Relationships
There are no relationship linked to this issue.
Notes
4
User avatar
No.00819
couriersud
Developer
Apr 26, 2008, 13:55
1. and 3. of the list have now been implemented. Could you please check again whether the problems persist?
BTW: are any schematics available for this game or Jack rabbit?
User avatar
No.00987
nicolasa
Developer
May 16, 2008, 13:01
as a reminder, the Money Money schematics ARE available
User avatar
No.01043
couriersud
Developer
May 21, 2008, 21:43
Just had a look at the schematics ... Lots of opamps and 4066 (analog switches) around. This will unfortunately take a while. This is definitely not a 8910 plus simple filter and mixer setup.
User avatar
No.16361
couriersud
Developer
28 days ago
We now have netlist sound for monymony. This should be closed.