[sdiy] CC-Modular

[xmurz at gmx.de]

08.09.02 20:08:21, Seb Francis <seb at is-uk.com> wrote:

>Hi Hans,
>
>xmurz at gmx.de wrote:
>
>> I woulnd't start doing all LFO's and EG's in a pic, this would make little sense to me,
>> but you could stick a couple LFO's and EG's in the Pic aditionally to some analog ones
>> if you got some processing power left.
>
>My reason for this would not be because I don't want to use analog modules for this .. it's just some way to try and reduce the amount of inputs and outputs which must be patched together.  A 
32x32 switch matrix requires 64 16-1 Analog Mux chips, whereas a 64x64 switch matrix needs 256.  Plus every CV which needs to be controlled by a "pot" needs a D/A.
>
>Moving 2 LFOs and 2 env gens to the digitial domain could save approx 20 inputs and 12 outputs needing patching, and save approx 20 D/A channels.
>
>I wonder whether anyone can really hear the difference between an analog and digital LFO or env when using 12bit DACs?  Also maybe there are interesting possibilities in the digital domain such 
as different shaped envelope ramps, and exotic LFO shapes...

I'd say that one CAN hear the difference (depending on the application). Why else do we run 24bit at 96kHz now? If you would use 20bit DAC's at a high frequency,
that would be enough. But any way you do it, you have to set priorities. For the system as you describe it, _I'd rather do a huge patching matrix with all analogue
(and maybe some extra digital logic/LFOS)  than mixing things up too much. You could as well run reaktor on you pc :)

>> Anyways, that's the way I'd do it.
>> The PCI idea is also very slick. Though that would mean you'd have to write drivers for
>> different operating systems and the card would have to be well shielded. (or only for one
>> operating system if you want once and only for you..)
>
>Since I would only write the "patching" software for one OS anyway, multiple OS support isn't really a consideration.  And it is important that all control can be done via MIDI in order to record any 
realtime parameter control in my normal sequencer (Cubase).

If you stuff all in a pc chasis, you can use a standard pc midi interface.

>> As you need so many ins and outs you will end up with an external box anyways.
>> OR you could do a custom PC chasis with additional linear PSU and stick all stuff
>> in one PC  custom chasis and have the posts on the front panel of the PC.
>> You can do that with a 19" 4U chasis that you build youself.
>> A small pentium would be sufficient, so you could do without a fan (->no noise)
>> and a zip drive to boot (which goes quet after transfer).
>> You would have to shield the analog circuitry pretty good, too.
>>
>
>Nice idea to use a cheap PC for all the digital synthesis and control (rather than a PIC or similar board).  
>After all my Mackie mixing desk has a PC at it's core.  Would have to use some better OS than windows though - 
>I don't fancy all the overhead of a GUI OS.  
>There's some "realtime" Linux kernel extensions which might be useful for this kind of thing - a synth which you can SSH/telnet into :-)

this system would leave very much room for additions (USB-audio/midi, telnet, SAMPLING/wavforms, FFT resynthesis,
digital delay/reverb, VST(i) support, video input/output....)

>I would need to be careful with shielding and power supply separation though.

not only the separation, you shure want the case to be dead-quiet. (better no HD, or go for the Barracuda IV which is VERY quiet)
I would build a case-withing-the-case to shield all the analog stuff from the computer noise. So you may be better
off building the whole thing, either in a big pc chase (19" 6U) or do a PCI card and a seperate chase (which I wouln'd like
that much...

If a Pentium 150MHz is enough you would be able to build that system without the use of any fan!

I would use seperate PSU's for both the analog and the digital part.

I really hope that your will build this system, and that you show it to us!

- Hans