For younger folks, the SID is the sound chip of the C64. It's infamous for being partly an analog circuit. That is, it uses transistors in states between 0 and 1, as you would in a transistor-based amplifier, or an analog radio. This gave the sound of the C64 its unique texture.
My understanding is that this behavior is impossible to reproduce with modern chip fabrication techniques, so what's the solution? Digital emulation?
Seen from the perspective of microcomputers, which mostly either used simple square waves, or FM synthesis to generate sounds, the SID is unusual. But seen from the perspective of sound synthesis the SID is an analogue subtractive synthesizer[0] which, at the time of its design, were commonplace. As I say, that was pretty unique amongst microcomputers, and is why the SID was and is revered by fans of the 8-bit era: it has a really fantastic, crunchy analogue sound, that other computers of the era just don't and, indeed, can't reproduce.
I'd go as far as to say that Commodore missed a massive opportunity with the Amiga in terms of use for music: if it had the PCM based Paula chip, plus a SID chip (or even two - a man can dream), it would have been absolutely incredible.
In recent years analogue synthesizers have experienced something of a resurgence although, in truth, they never completely went away. As such ICs with analogue circuitry are actually relatively commonplace. Now that doesn't mean you can just remanufacture the SID chip really easily, but it is absolutely possible to build ICs with analogue circuitry today: as a trivial example, analogue to digital converters are commonplace.
[0] I think strictly speaking it's also a paraphonic synth, since all the voices share a single filter, but I can't remember off the top of my head whether they also all share a single amp (again, a hallmark of a paraphonic, rather than a true polyphonic, synthesizer).
Emulating the distortion is done with varying levels of accuracy.
Additionally the fabrication varied from part to part and from rev to rev. So the sound of any SID is a combo of its batch of substrates limiting bandwidth and the intermodulation distortion (IMD) of its FETs' imperfect emulation linear adding and multiplication functions increasing the bandwidth. Note that these non-linearities are in a feedback loop.
Furthermore the external caps ("matched") used for the filter varied considerably, but this is more easily reproduceable.
The envelopes are unconventional too, but more easily emulated than the filter. They are a mix of linear and exponential and use the non-linear FET multiplier.
Also the oscillators are not simple analog oscillators. They're high frequency lookup tables with digital aliasing related to the key the music is written in. The noise waveform is not analog, either, but a variable frequency LFSR, although this is easy to reproduce in FPGA and somewhat in software, although the max frequency can be challenging.
Finally the entire signal chain is replete with IMD so any melody/rhythm played on one voice is affected and influenced by what is played on the other voices.
In summary the SID was designed by electronic musician Bob Yannes who modeled it after typical analog synths but was technically limited by the fabrication of the age to make it less clean than even the grungiest cold war soviet synth. Still it was better than everything else at the time and it got a lot of loving attention from devs and musicians who squeezed every ounce of juice from it.
IDK - the first DAWs were made for the Amiga. If the Amiga hadn't already died-out by the early-ish 90's then I'd imagine Pro Tools would have been created on that and there would be a very real possibility we'd still be using Amigas today.
In a world where the Amiga didn't need an external MIDI interface, I suspect things would have been different.
1. Built-in MIDI interface.
2. High-resolution screen available, 640x400, 60Hz monochrome.
3. Simplistic operating system, the DAW software had 100% control of all hardware and timings.
What's difficult is replicating it with generic programmable platforms like FPGAs because those are mostly digital gates.
But if we'd just want to cook a bunch of SID chips that's no problem. It'll just be hard to get enough interest to warrant all the design and setup work. You'd need to do a large batch to make it worth the investment.
That said, if I were to do music production, I'd use my real chip, because I prefer the sound of it.. But, on a PC, I'd use one of the libraries, like libsidplay which emulates the sound better than the ArmSID.
The same could be said for the VIC2 chip, I've yet to encounter two that give the same image, they look slightly different. There, I do prefer the look of the kawari on s-video, it's just.. easier to see ^_^
Might just be old age? At some point, IC's may be degraded to a point where few (if any) work exactly like original part when it was new. And thus also differ between specimens.
At some point, original parts may become unreliable / quirky / glitchy enough that picking some kind of re-implementation (rather than original) as the "gold standard" is the only way forward.
Not a big issue for 'pure' digital parts like CPUs, glue logic etc. But (partly) analog IC's like sound or video chips... different story.
I was really into Koyaanisqatsi at the time and you'll definitely hear the influence there.
FPGASID - https://webstore.kryoflux.com/catalog/product_info.php?produ... reDIP SID - https://github.com/daglem/reDIP-SID