[sdiy] Super cheap synthisizers

phillip m gallo philgallo at attglobal.net
Sat Jun 15 02:09:51 CEST 2002

Understanding that discussion to date qualify as "brain storming, a lot of
good input has been exchanged.  I just wanted to get a feeling for specific
issues, not from a product manufacturer stand point (i am not), but that of
a "constructor" who would (at least theoretically) construct and implement
such a device.

>(1) Expected cost per channel;

--> Speaking for myself, time is more costly than component costs (within
reason of course). Anything that implements "quickly" from a construction
and packaging point of view has a reduced cost of implementation for me.

>(2)is it fully digital or a hybrid;

--> Again i didn't revue this thread from a product standpoint. Instead i
thought of it from a project standpoint, where an individual module design
allowed system growth through aggregating additional modules.

>  (3) MIDI driven or CV?;

-->A MIDI implementation eases the design from a parts and cost standpoint.
Would a voice module provide external "synch" signals (i.e. Trig/Gate) for
integration with existing VC modules?

>  (3a) if MIDI does each channel operate as a stand alone MIDI
>client or is there a master Unit which "terminates" the MIDI input and
>distributes it to "slaves"?

-->I too thought a mono-synth (per module) was envisaged.  I also thought
the thread indicated a similarity of voice assignment with the ability to be
"Channel Addressed" (voice per MIDI channel). If separate modules will
co-exist upon a single MIDI channel then a "Master Assignment" mechanism is
required. This could implement many ways spanning a specific Master Module
responsible for assignments to an assignment algorithm where module 1
assigned to "currently sounding highest note", module two = "currently
sounding second highest note" etc.

-->If each voice = MIDI Channel then the only requirement would be
terminating the MIDI current loop and passing the MIDI stream unfiltered to
additional module (perhaps at TTL levels).

>  (4) is it a "reduced" performance synth or
>expected to provide "state of the art" audio.

-->By reduced performance i meant is it a cute little synth module wonderful
for what it was but not necessarily able to compare to it's analog brothers.

So I imagine this as the big picture.

In short. Use as much uC power as you can to give you the most bang/buck
where possible leaving budget for analogue stuff where it really really
counts. IE: Filters and wave shaping.

--> Three micro's may make sense for a lot of implementations. My interest
in this thread is that it allows another way to objectively consider my
current digital voice project. I am at the front end of this project which
was envisioned somewhat differently from the existing thread.  The
difference is that i want a complex multi-voice module, i want very good
fidelity, i want it to co-connect to my existing VC gear as just another

-->My solution implements two DS89c420's with the workload distributed
differently.  The two uproc's share a dual port RAM.  The Dual Port Ram
contains the wave tables, and mailbox's for the passing of new phase
increments and other variables between the proc's.

--> 1 uproc connects to the two 16-bit DACs and deterministically
(non-interrupt/non-timer) indexes and accumulates from Wave tables existing
in a Dual Port RAM.  This processor also stores the current value of 2 16
bit CV input conversions (to allow outboard CV gear to drive the module).

--> The second uproc (1)terminates and processing the MIDI Messages of
interest, (2) Updates (filtering, amplitude scaling) the individual Wave
Tables, (3) reads, averages, and stores front panel control values,
(4)generates ADSR's which drive the individual wavetable filters, and
amplitude scaling.

-->At this time i am building and testing the DCO accumulation routine.  A
single voice index takes just about 1us per voice. So accumulating and
outputting 8-12 DCO's across 2 DACs doesn't look to be the challenge. The
issue is maintaining 12 independent Wave Tables with individual Adders for
PWM/DCF/DCA operations.


Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.370 / Virus Database: 205 - Release Date: 6/5/02

More information about the Synth-diy mailing list