Amiga audio channel split by hardware?

[Amiga1992]

Is it by any means possible to split each of the channels of the Amiga into two to have 4 separate MONO outputs?

I know it has nothing to do with the way the Amiga works, but the AY/YM series of chips allow this, more prominently on the Atari ST, where you can take advantage of using a better amp instead of the one it comes with because it's faulty.

[rkauer]

Yes, look into the schematics and use an extra pair of op-amps and two more RCA jacks to output in four channels.

[Shadowfire]

If I understand the question correctly, you want each of the four channels on a different mono output, not having all four channels on four outputs. IIRC Paula only outputs left & right channels, so the two channels are mixed internally inside Paula and what you request isn't reasonably possible.

[Amiga1992]

Originally Posted by Shadowfire

If I understand the question correctly, you want each of the four channels on a different mono output, not having all four channels on four outputs. IIRC Paula only outputs left & right channels, so the two channels are mixed internally inside Paula and what you request isn't reasonably possible.

Yes it's exactly what i want, to separate each channel to get the 4 mono outputs of each. There really isn't any way to jump in before they get mixed?

[rkauer]

Glenn is right: the channels are mixed inside Paula. Oddly I remember a device that splits the four channels for using a laser canons presentation controller...

And it surely was using an Amiga for that.

[Amiga1992]

:O
I need to know about this device!
This would make a major breakthrough for us music makers

[pandy71]

use 2 Paula's - detect adress and prevent access to ODD/EVEN pair of channels or mirror (decode adres and buffer data) from AUDxDAT (and maybe also AUDxVOL).
you can read this with SYSCLCK/16 ie somewhere around 1.79MHz.

[Amiga1992]

You can daisychain Paulas?

[Shadowfire]

Originally Posted by Akira

Yes it's exactly what i want, to separate each channel to get the 4 mono outputs of each. There really isn't any way to jump in before they get mixed?

No. The left pair and right pair are mixed inside the IC (Paula has digital & analog sections inside the same IC). Only the resulting left and right audio streams are accessible outside of the chip.

You would need to build custom hardware (one route would be a daughter board that fits into the Paula slot, on which sits an fpga (which emulates Paula in hardware), 4-channel audio codec & support circuitry). A huge undertaking for something of questionable use, so likely a hobbyist project, if ever done at all.

Pandy's suggestion is somewhat impossible, since there is no DMA control hardware for a 2nd paula in Gary. Even it was possible, you would require custom software to use the 2nd Paula.

[8bitbubsy]

Speaking of emulating paula, is there any source code for an accurate paula emulator out there? I mean, someone that translated the schematics to code or so. Preferably assembly or C.

[Shadowfire]

You might want to check out the WinUAE sources.

[Amiga1992]

I understand, Shadowfire. Thanks for your valuable input!

[whiteb]

Originally Posted by Shadowfire

You might want to check out the WinUAE sources.

Or learn VHDL, and use the minimig sources to create a secondary Paula and split your channels that way, and route the new channels out to the Spare I/O.

You cannot physically splice the channels out, if the mixing is done inside the Paula Chip, at least on a REAL amiga, so hence, the Winuae Sources (And thank Toni for them), or make your own custom Minimig

[Toni Wilen]

UAE sources are quite messy.. It is not made for generic emulation and resampling is very tightly integrated.

100% correct emulation for module-like players will be very simple ("nothing more" than a 4 channel D/A converter), it gets complex when interrupt timing, latch timing etc.. needs to be emulated. (normally only needed if you want to emulate an A500 not just Paula audio)

Check also HRM audio state diagram, it explains many things, but it isn't easy to understand

[pandy71]

Originally Posted by Shadowfire

Pandy's suggestion is somewhat impossible, since there is no DMA control hardware for a 2nd paula in Gary. Even it was possible, you would require custom software to use the 2nd Paula.

No, my suggestion is guite ok - 2 Paula'a, share same adress space but different levels for volume on each Paula: first Paula serve as a source for 0 and 1 and second Paula as a source for 2 and 3, it can be done transparently to the system. Only need to change level for AUDxVOL that oposite channels are muted or repplace data for always 0 that particular pair of channels are not in use (is not outputed to particular R and L lines of Paula) - this can be made with relativelly simple logic (simple CPLD shall be sufficient)


but Toni what with second idea? - store values from AUDxDAT - all we need to do is decode RGA then store in buffer/latch value for particular channel ? maybe similar for AUDxVOL - then we have four pairs (sample + volume) of data that can be used outside - if we sample this 1.79MHz then we have all informations that we need.

[Toni Wilen]

Originally Posted by pandy71

but Toni what with second idea? - store values from AUDxDAT - all we need to do is decode RGA then store in buffer/latch value for particular channel ? maybe similar for AUDxVOL - then we have four pairs (sample + volume) of data that can be used outside - if we sample this 1.79MHz then we have all informations that we need.

It should work if you don't need to be 100% accurate, you can't know when exactly samples switch without period counting, switch when AUDxDAT is written and when to use high 8-bit sample or low 8-bit sample.

btw, DMA bus access rate is ~3.5MHz, ~28MHz main crystal divided by 8.

[pandy71]

accordingly to the http://www.bel.fi/~alankila/modguide/interpolate.txt - You have right Toni - my mistake (i dont know why i think on SYSCLOCK/16)

So if we decode RGA (access to the AUDxDAT, AUDxVOL), store values written to those registers (assume that value not changed remain the same) we have exactly the same data as Paula - then we know how state machine work - seems that it is simple Zero Order Hold then we can perfectly emulate general behavior of Paula.

So in CPLD we need "only" to:
-adress decoder
4x16 bit register to store AUDxDAT
4x6 bit register to store AUDxVOL
some glue logic that detect access (new Write to the AUDxDAT and AUDxVOL then strobe to change values stored in our registers)
state machine to "play" 16 bit value as 2x 8 bit values
then simply join 6 bit from stored AUDxVOL and the 8 bit AUDxDAT (upper 6 bits are xVOL, lower 8 xDAT) - such 14 bit word can be outputed directly to the 14 bit DAC (or we can decimate this by some reasonable factor - 32 ... 128 to lower DAC speed)


But IMO doing this only to have DC galvo control for laser seems to be unreasonable - currently there is plenty of multichannel DAC with SPI data input - easier is to connect this to the computer with using some PIC or AVR than reinvent the wheel.

PS

I know why i mentioned about 1.79MHz:

"If the sampling rate is set much higher than the normal maximum sampling rate (approximately 29 KHz), the two samples in the buffer register will be repeated. If the filter on the Amiga is bypassed and the volume is set to the maximum ($40), this feature can be used to make modulated carriers up to 1.79 MHz. The modulation is placed in the memory map, with plus values in the even bytes and minus values in the odd bytes." - source obviously AHRM

[Amiga1992]

I don't want to do this for controlling lasers, I am a musician and I want all 4 channels with independent audio outputs, that's all!
I thought a mod similar to the easy YM2149 channel split mod would be possible

[rkauer]

I know there was a laser projector for the Amiga: http://www.pangolin.com/products/pricelist-lsd1000.htm

Thing is: I cannot remember where shows how to split the four channels to use the laser control hardware inside the Amiga...

But I'm sure it is possible at least in the 500/2000 or it was only for the A1000?

[pandy71]

so seems that lsd doesn't split channels - i think that they use 2 channels but key to the success is to have DC coupled DAC - this is possible in Amiga with some small modifications.