[sdiy] CC-Modular

Paul Maddox (QinetiQ) P.Maddox at signal.qinetiq.com
Mon Sep 9 10:15:06 CEST 2002


Seb,

What you're talking about has been discussed on the opensynth mailing ist.
----- Original Message -----
From: "Seb Francis" <seb at is-uk.com>
To: <synth-diy at dropmix.xs4all.nl>
Sent: Sunday, September 08, 2002 2:25 AM
Subject: [sdiy] CC-Modular


> Hi,
>


> It's been interesting reading the discussions about the computer
controlled Walsh Module.  I've had a dream for a long time now to build a
complete modular synth system where it's possible to save and recall patches
instantly.  Ok, so it's good to always make new sounds with every new piece
of music, but sometimes it's nice to be able to call up and re-use an old
sound without having to manually copy it from a patch sheet (to say nothing
of having to manually "store" the sound on a patch sheet in the first
place!)
>
> Being a relative beginner in the world of analog audio electronics I'd
like to share my ideas, invite comments, and even ask a few questions to
some of the experts on this list :)
>
>
> Design goals:
> - The synth must sound analog (so digital modelling is not a solution)
> - The synth must be as flexible and extensible as a normal modular
> - Electrically compatible with other common modular systems (e.g. MOTM,
etc..)
> - Completely "saveable" patches
> - PC based graphical patch editing
> - MIDI control of every parameter
> - Physical control of every parameter
> - The synth must sound good (did I mention sound already .. it's
important, otherwise I'd just buy a NORD modular! ;)
>
>
> Overview:
> (see also picture: http://burnit.co.uk:8888/overview.gif )
>
> Clearly no pots can be used, so physical control would be done from a
digital device.  I plan to use some nice quality (but not too expensive)
ALPS continuous rotary encoders.  The ones I'm thinking of have 24 sine
pules per rotion, and because they are sine output you can derive much more
detail than 24 notches per rotation.  Each "pot" would send pairs of MIDI
controller messages (14 bits) and have a dispay as you turned it to show the
exact digital value.  Exactly how many turns make up the full value scale
depends on how fast you turn the knob (i.e. turn slowly for precise control,
faster for bigger jumps).  When a "patch" is loaded on the PC, the software
would send MIDI CC messages to initialise the value of each "pot".  Names
for each pot would also be sent by SysEx, so you get the name displayed as
you turn each control.
>
> So far, so good.  This whole controller could be build around a PIC or
other microcontroller and a few auxiliary components.
>
> Also no problem is the PC software (softare I can do!) - some nice GUI app
built with something quick like Delphi, working in a similar way to existing
"virtual" modular synths.  No need to worry about special hardware drivers
since MIDI is the method of communication.
>
> But now it comes to the tricky part .. the interface between computer and
analog modules.  Every analog module must have no pots or switches (or at
least if it does have pots they must be kept at the same position - I'm not
about to start mounting motors in the analog modules!)  So every parameter
must be controlled by CV, and every input must be able to be "patched" from
every appropriate output.  In addition control voltages generated from the
14bit MIDI CC data must be able to be patched to every appropriate input.
>
> My original idea (see the picture) was to implement this literally - e.g.
32 MIDI CC -> CV converters, and a 64 input x 64 output digitally controlled
analog switch matrix (even this amount of CVs and switches isn't enough for
a very large modular).  Then I started working out how many chips I'd need
for the switch matrix (*4096* 16-1 analog multiplexers!) and I realised why
there isn't anything like this made already.
>
> To try and solve this without compromising the design goals too much I
considered 2 things:
>
> (1) Limit the flexibility in ways which would not matter (e.g. mixer
outputs never need to connect to mixer inputs, don't need to use noise for
things like pulse width control, etc..)
> - It would be hard to do this with an ever extending modular, since the
digital circuit design needs to have some idea what inputs can be patched to
what outputs.
> - Actually when you look at it, there are not so many outputs/input
combinations which might not give some nice synthesis possibility.
>
> (2) Move some of the synthesis functionality into the digital domain (i.e.
the microcontroller).  LFOs and envelope generators seem the obvious choice.
It seems unlikely this will compromise the sound too much, and neither of
these modules generally need input from any other analog modules (except
from other LFOs and env gens).  A large number of cross-patchable LFOs and
envelope generates could be programmed, limited only by the microcontroller
speed and how many CV outputs are needed.
>
> I believe a combination of these 2 approaches will make this idea a
practical proposition, but I am open to other ideas :)
>
>
> And now I come to my questions ..
>
> Will digital LFOs and env gens sound as good as analog?  I can't think of
any reason why they shouldn't, but maybe I've overlooked something?
>
> I want to make it electrically compatible with as wide a range of modern
modular systems as possible:
> What are the generally used voltage ranges and current levels?
> Analog switches that handle 0->15V (or -7.5->7.5V) signals are quite a lot
cheaper than switches that can handle -15->15V - I wonder if I can get away
with using all, or at least some, 0->15V (-7.5->7.5V) ones.
> How much current must each switch and digitally generated CV signal be
able to handle?
>
> How much effect do you think the choice of analog switch will have on the
sound - could I get away with using really cheap things like HEF4067
(datasheet says 0.04% sine distortion, but no mention of THD), or perhaps a
mixture of HEF4067 for CVs and some more expensive (e.g. ADG428) switches
for the audio signals?
>
> Could I get away with cheaper 12bit DACs (~3.7mV per click on a 15V
scale), or would 14bit DACs (~0.9mV per click) be noticeably better for
things like LFO and Envelope ouputs?
>
> Is it ever needed to patch an output into multiple inputs in a modular
synth?  Or is this electrically not so good?
>
> Am I mad at even trying to attempt this?
>
>
> Thanks in advance for any replies.  Actually it's been good just writing
my ideas down in pen and ink (or do I mean 0 and 1s;)
> Seb
>
>
>




More information about the Synth-diy mailing list