Try to detect different AY/YM chip types

[tomcw]

From @roughana:

Given French touch recent expose reporting of 8913 being buggy, is there anything you can do to discern which of AY /YM is being used? Or is that only possible by connecting directly to the YM2149?

[roughana]

From this blog

So the information about the extra bits was correct, just with a caveat. If you write 31 to register 1, you will get 31 back on a Yamaha YM2149, and 15 back on an AY-3-8910/2. All other bits and values work the same on both devices.

[roughana]

From this comment

The difference between the "original" AY-3-8910/8912 and the YM2149F, is that the latter has a 5 bit volume register (32 steps), whereas the AY - only 4 (16 steps). That translates into smaller DR on the AY, it sounds more compressed - thus certain sounds sound louder in the overall sound picture on the AY, making them sounding somewhat more "dense" because they were designed for 4bit volume scale. The same music on the YM basically sounds at half of it's volume - that is what we hear. Otherwise, it's not a bug, but a feature, widely utilized by Atari ST software, for instance (that has the YM2149 on board).

[roughana]

Chris Torrence commented in his video review of the A2FPGA that he felt that the A2 speaker output was at a different level to the mockingboard output and expressed a desire to be able to balance them somehow in future. Given that the A2FPGA implements the YM2149F, perhaps this gives a reason as to why there is a difference. Perhaps the A2FPGA should shift the volume register to compensate for the expectation of the music being produced by an AY3891x.
Perhaps the mbaudit can shift the volume register if it detects YM214Fs and lets compare shifted / unshifted.

[tomcw]

I can reproduce this using my hybrid RM Mockingboard (ie. Chip-A = AY-3-8913, Chip-B=YM2149F, and both 6522s = WDC W65C22S6TPG-14).

This doesn't work (and I just get 0x0F for either AY/YM chip):

• Select reg
• Write value
• Select reg
• Read value

I can only reproduce when using this sequence:

• Select reg
• Write value
• Read value

For my hybrid card:

• the AY-3-8913 always returns 0x0F.
• the YM2149F successfully returns these values that I've tried:
  • reg1: 0x1F, 0x3F, 0x7F, 0xFF
  • reg3: 0x1F, 0xFF
  • reg5: 0x1F, 0xFF

See: 1b04096

[tomcw]

mb-audit v1.59 now detects YM2149s (but not for MB4c and Phasor's 2nd-ary AYs).

[tomcw]

Detect AY-3-8910 vs AY-3-8913:

• Write ENABLE (reg-8) with 0xFF (set both AY ports as outputs)
• Write PORTA (reg-E) with 0x55
• Write PORTB (reg-F) with 0x69
• And read the 3 values back

My hybrid RM Mockingboard

• AY-3-8913: 3F, 0E, 0F
• YM2149F: FF, 55, 69

Phasor

• AY-3-8913: 3F, 0E, 0F
• AY-3-8913: 3F, 0E, 0F

MB-K

• YM2149F: FF, 55, 69
• AY-3-8910 (AR's dud 3): FF, 55, 69