digger wrote on 2023-09-04, 16:13:

Maybe you could increase the "demand" for such a CVX5 by adding Stereo-on-1 support to FastDOOM, @ViTi95? 😇

EDIT: Ah, chicken-and-egg, huh? You're probably waiting for it to become available so you'll be able to test such an implementation.

True, didn't add support for it as I don't have one 😂

darry wrote on 2023-09-04, 17:21:

That is an interesting idea, except for these points, AFAIU : […]
Show full quote

That is an interesting idea, except for these points, AFAIU :

a) Output would likely need to be upsampled to at least 32KHz (I do not have a anything with an S/PDIF input that supports less than that, AFAIK)
b) Low pass filtering would need to be done in the digital domain

The above would increase the cost, but might well be worth it. That being said, if the horsepower to do that were to be added, why not scope creep it a bit more and add an FM synth option too (real chip or emulated) ?

I was joking, and I assume that Scali was as well. 😁 Not that it wouldn't be a cool project, but this is a typical case of scope creep. 😅

Stereo-on-1 compatibility would already be a really cool improvement.

Well I was, sortof...
I mean, it should be possible. The Stereo-On-One uses one of the outgoing status signals to select between a left and right DAC.
There are 4 outgoing status signals on the printer port in total.
So in theory you can decode this as a 4-bit selector, allowing you to select between up to 16 DACs. So 5.1 should be easy.
In fact, you could probably create a 16-bit DAC this way: select between low-byte and high-byte, and use a 16-bit buffer before the DAC. Then you output 16-bit samples as two separate 8-bit outputs.
You could then create an 8-channel 16-bit PCM DAC over a single printer port: 1 status signal selects between high-byte and low-byte, and the remaining 3 status signals can select between 8 different 16-bit outputs.

Although, at some point you'd run into the bandwidth limitations of the printer port. The standard 150 kb/s wouldn't be enough for 44.1 kHz at 16-bit. But with 2 MB/s of an EPP you should be able to do the 8 channels easily, in theory.

My experience has shown that a vanilla LPT port (at least on on 486 and newer hardware) tops out at more or less exactly 1MB/sec (1µs cycle time for data and control ports access) so having 6x 8bit channels is able to give 1000000 / 6 * 2(one data and one select write) = 83.333kHz but of course all the CPU time is lost doing that since the CPU will be stalled to maintain that 1µs cycle time in order not to upset printers etc.

Hi,

The problem of the Stereo on one is that there is a need to write to a register to select the channel Right/Left
Instead of just alternate the channel each time a data is sent for example.
The 2nd problem is that the PC need to program a timer at the output frequency, this is extremely slow, here come the FIFO 😀
Plus of course a FIFO (Programmable size) with IRQ When "Almost empty" to have time to start the IRQ at the computer side.

Then, support of ECP/DMA 😀

A Raspberry Pi Pico (Or other microcontroller) can be an excellent to do this, advantage of the Pico is that we can transform this to ZIP Drive or network card emulator as well 😀

But... all those modification will require a specific code to work. Patch of existing software may be not simple.