[sdiy] new project in the works

[Czech Martin]

Nice idea!

Why do you divide? There is always this divide by zero problem.
Wouldn't dividing with a resistor network with following
gain block do the trick?

OTHO: it could be done the other way round.
Instead of changing the input signal, you could
control the current fed into the resistor network that
creates the reference levels for the comparators.

A resistor could do the job...
But I doubt if the comparator IC could still be used.
Anyway, a flash converter using some quad comparator
chips shouldn't be that hard to build...

At the output side scaling could be done with the
same reference fumbling trick.
Ever heard about multiplying DAC applications?
Another resistor will drive current into the output network
thus making the amplitude smaller or larger.

So I guess, throwing out the AD633, using some comparators
and resistors would do a similar thing, but perhaps with
less problems and perhaps cost.

The idea of scaling on both sides should prevent loudness
droop, but remember that loundness is not equal to amplitude,
spectrum play a big role also. So perhaps a nonideal scaling
is absolutely no problem.

Perhaps there are some (flash) ADC/DAC converters around that
have programmable reference biasing. SAR types are of no use,
because you need amplitude quantisation, not time quantisation.
Then you could use such a chip, with the advantage that
integration can deliver more comparators then
healthy in a discrete approach.

So far you can change only the granularity of the signal.
The easiest implementation of such a circuit is the well
known PWM, then next a 3 level window comparator (this idea came
up in Electro Notes several times).

We can extend this even more (sorry, but I'm getting exited ;->):
Your idea gives quantisation, but the curves stay monotonic:
rising input will finally give rising output.

A memory between ADC and DAC could change this. Now any 
input level can be translated into any output level.
I guess this would make the thing more spicy.

Ahhh, ... sorry, , back to analog this means that the thermometer code
(flash ADC) via priority decoder
will select (one out of N) potentiometers,
very much in the fashion a sequencer will do it.

If we think a bit more about it, this thing is a sequencer
if we run the input with a slow saw wave. 

More flexibility could be achieved if the input steps could
as well be altered, meaning individual thresholds for
the comparators (but unique, of course).
Variable thresholds at input and variable level at output,
this sounds like a great sequencer (no problem to have
very uneven note value) or audio wave mangle.

The only problem I can see are glitches. But thermometer code
and priority decoder can take care of this, I think.

Call it Wave Mangle...


m.c.





-----Original Message-----
From: Michael Ruberto [mailto:frankentron at hotmail.com]
Sent: Mittwoch, 28. Mai 2003 09:20
To: synth-diy at dropmix.xs4all.nl
Subject: [sdiy] new project in the works


I am currently working on a new wave mangling device which is showing some 
promise. you can read the details here:
http://www.sunrisecomputersolutions.com/samples/decimator.html
so far I am still slaving over a hot breadboard. any suggestion you might 
have are welcome. I should have some more developed schematics by the end of 
the week.
After reading that a decimator is really something quite different from this 
in the world of computer music programming I am considering changing the 
name of this thing to The Cubist Processor. sound OK?

~M

_________________________________________________________________
Add photos to your messages with MSN 8. Get 2 months FREE*.  
http://join.msn.com/?page=features/featuredemail