PoKey emulation?

[Lynxpro]

If any pokey experts have an fpga board I'd be very interested in finding features I need to fix/improve on.

 

What about the Rapidus board? Supposedly, it does Quad POKEY and Dual SID emulation…and possibly at the same time...

 

[foft]

Getting closer...

 

http://www.scrameta.net/autobuild/instrumentarium/Instrumentariumv2.mp3

 

I re-did the plls - timing should be much closer (2% vs 0.1% wrong - I can't get exact for PAL).

I fixed 2-tone I think (bit 7 forces serial out to 0 - I was doing this after the 2-tone logic).

Poly counters now pass Perry's linked test.

 

Still not sure about the poly 5 issue though - does the problem still exist?

Edited August 23, 2014 by foft

[thorfdbg]

One example:

...

You can find more here. I hope it's now clear enough.

 

http://atariage.com/forums/topic/149261-a-new-pokey-tune-unusual-instruments/?p=1820884

 

 

Thank you a lot! I fixed this - actually, this was caused by the pokey "channel mute" logic being a bit overly ambitious as it disabled the high-counter of a high-low channel pair even though the high-counter is required to drive the reset-logic of the low counter. Thus, next release will get this right!

[foft]

Still not sure about the poly 5 issue though - does the problem still exist?

So from the times with problems I guess the issue I have is with:

----------------------------------------------------

Distortion 2 at 1.79 MHz: (0:46 - 1:09, 1:32 - 1:39)

 

Channel: 1 or 3

Pokey Distortion: 2

AUDCTL settings: x1x0x0xx or xx1x00xx

Custom Notation Table 1

 

Remarks:

-This instrument gives better-tuned high notes, independent of 15 kHz or 64 kHz mode

(see http://atariage.com/forums/topic/149261-a-new-pokey-tune-unusual-instruments/?p=1820884).

 

I tried to reproduce this in basic, but I get what appears to be a near identical sound.

 

10 P=53760

20 TL=250

30 POKE P+8,64

40 POKE P+1,74

50 POKE P,TL

60 GOTO 60

 

i.e. 1.79MHz channel1, distortion 2 (010 - 4 bit and 5 bit together)

 

Is this the sound the problem is heard with?

[analmux]

@ foft

 

No, distortion 2 only plays the poly5, not combined with poly4, 9 or 17. Do you use a real atari by the way?

[analmux]

Ah, I see in line 40: POKE P+1,74 is equivalent to distortion 4: use hi nybble: 74 = 16*4 + 10, thus distortion 4, volume 10. Yes, this is poly4+poly5.

 

POKE P,TL, with TL=250 gives the real pitch: 250+4 = 254 = 2*127, which isn't divisible by 3,5 and 31: it's a maximal sound combination w.r.t. poly4 and poly5.

[foft]

Thanks - I see. So 0010, not 0100... I'll make that change and try tomorrow.

I thought you meant 2 in the 3 bits as described here:


Yes I have a real 600XL and XEGS here to compare with.

Edited August 24, 2014 by foft

[foft]

Reproduced. Removed inversion on the poly 5 output (its on the decap schematic, which also has the wrong terms to the xnor) and it matches now with my test.

 

Thanks for the help, this should be pretty close now.

 

Sure enough instrument starting just after 45seconds ish match frequency better:-)

Instrumentariumv3.mp3

Edited August 25, 2014 by foft

[foft]

Another interesting case:
10 P=53760
20 POKE P+8,64
30 POKE P+1,10+32
31 POKE 53775,11
40 POKE P,20
50 GOTO 50

Relatively low frequency on my 600XL, higher frequency on my implementation and on Altirra 2.5 (not checked latest test).

Perhaps down to two tone delays? I haven't adjusted delays with 2-tone mode. Two-tone extra delay description. Have to connect up the scope/LA I guess...

[phaeron]

I've known about some inaccuracies in Altirra's emulation with two-tone mode at 1.79MHz rates but haven't gotten around to tracking them down. Last time I ran into it was when I was working with PWM audio, where some max amplitudes worked in emulation but crackled on the real hardware.

 

If I simplify the case to a lower pitch square wave and hold down timer 2 so it doesn't interfere, it sounds like the timer 1 period needs to be lengthened by two cycles. However, that's not enough to fix the case above, where there is some other interaction going on.

[emkay]

Anti forget propaganda

 

The Video shows some different tunes played in Altirra. Seems the 60Hz plays "speed stable". Not sure about the timing.

Some tunes use "instruments" with special timings and some also use "Filter-Modulations" .

Minimod has only some adjustments and the main filter runs free (for example)

 

All of them were played at 50Hz speed. Not sure about the channel mixing. To me it seem "over" compared to the real thing.

 

It's all the same volume and the same Altirra-settings. Just different usage of the soundgenerators.

 

Edited August 30, 2014 by emkay

[analmux]

... Which mathematics are going to help me out here?

F.e. this link I hope:

 

atariage.com/forums/topic/124994-pokey-help/?p=1524134

 

It's an example of sawtooth mathematics, i.e. square roots, quadratic equations, prime numbers, relatively prime pairs.