My evil modular plans

[J.D. McEachin]

On Wed, 23 Apr 1997 gstopp at fibermux.com wrote:

>      This is an interesting topic. I'm not sure what the optimum solution 
>      would be - bear with me as I kick around some ideas and realize that 
>      I'm not too sure myself where I'm headed....
>      
>      Okay if you make a modular synthesizer out of modules that are digital 
>      inside, do you bring out all the patchpoints to panel jacks so that 
>      you can use patchcords to connect them to other modules, analog or 
>      digital? If you do, then this means A/D and D/A and the subsequent 
>      potentially cumulative conversion errors.
>      
>      In contrast, if you decide to keep all signals in the digital domain 
>      until the final audio output, and the inter-module signal pathways are 
>      to be virtual connections (what, a 1G ATM bus?), then why the heck use 
>      physical modules in the first place? You might as well do everything 
>      behind a CRT.
>      
>      So, maybe the first idea is the better one, at least for the 
>      traditional type of modular synthesis. Keep the phone jacks (or 
>      whatever), use high-res A/D-D/A, and emulate analog modules. Why do 
>      this, anyway? Well I think it's time to try, since it's getting to the 
>      point where DSP can actually be both afforded and programmed by the 
>      DIY enthusiast. My recent experiments with 18-bit PCM audio DACs have 
>      shown me that intentional aliasing free of clock noise (x128 
>      oversampled conversion) has opened up a whole new type of sound 
>      processing, for example. I think that the vectorboard delay line is 
>      going to end up in one of my modulars rather than in one of my effects 
>      racks!

Why not route SPDIF to the jacks?  You could have a few A/D & a few D/A 
modules for interfacing w/ other stuff, just like you have to have an 
interface module to patch something into a Serge.


>      And now this statement from a die-hard analog-head: digital is 
>      theoretically capable of perfect analog emulation. There, I said it. 
>      In the practical world it's just a matter of granularity plus knowing 
>      what to program. The reason that we keep analog around is that it's 
>      the most cost-effective way to get vast quantities of interacting 
>      electronic processes with all of the nuances of temperature 
>      sensitivity and component tolerances and connector resistance and a 
>      bunch of other things that add up to sounds that are mostly controlled 
>      but a little slippery. "Fuzzy" I guess would be an in-vogue term... 
>      Anyway as digital prices drop and digital tools expand we can consider 
>      analog-like process emulation and get the extremely important features 
>      of total control and total re-configuration.

I totally agree with you.  I'd like better resolution, say 24bit/96kHz, 
but it's already getting good enough to fool me.


>      One thing that bugs me is - if you take your DSP VCO and download a 
>      DADSLSR-whatever envelope generator into it, won't you need different 
>      knobs in different places with different markings on them? What to do 
>      here - maybe make the panel one big LCD with generically laid-out 
>      knobs and jacks? Change the module, change the markings on the LCD? 
>      Ooops better drop back to earth here, this is DIY land... I suppose 
>      dedicated panels would be okay. Maybe we could differentiate between 
>      signal sources (VCO, noise, envelope, etc.) and signal modifiers (VCF, 
>      VCA, phasor, flanger, etc.).

Control surfaces remain the big gotcha in all of these digital 
endeavors.  There's no substitute for tactile feedback.  Maybe in 5-10 
years, VR gear will get good and cheap, and you'll be able to reach out 
and tweak a wall of virtual knobs.

JDM