[Re: PolyModular]

[Harry Bissell]

Harry Bissell: The "autotune" routines that I'm familiar with don't work quite
this way... They measure the output frequency at several points across the
frequency range (pll if you like) and then the sharp-flat correction data is
stored in RAM (counters if you like). This is dynamically applied
(interpolated data) across the range. 
    The rub is... The circuit has to know what note you are attempting to
play. If it is from the keyboard, and read by a MICRO, then the DAC value is
"pre-corrected". But if the control comes from an ADSR or LFO, the circuit has
no way of knowing which correction factor to use (unless it is going to
digitize the CV input.

I do agree that if I had five VCO's it would be nice to do things with them...
But if we come up with a cost effective package, we could have 10. And if
someone wants a custom DIN cable... go for it.

I also agree that V/Hz is limiting. Maybe the place for a micro is in the
"pentaexpoconvertermodule" I (laugh) proposed. The tuning routines could be in
the form of some kind of feedback from the VCO. Good VCO's are a priority.
Having 20 VCO's is no good if they all suck!!!

My KBD chops, alas, are like yours. MIDI allowed me to produce my CD with the
orchestration I always wanted, and at a price I could afford. I remember long
nights text editing pitchbend data. In research this manipulation would be
called "cheating".... :-) Harry



Rene Schmitz <uzs159 at uni-bonn.de> wrote:
Hi DIYers,


The main trouble to me is the problem of maintaining tracking, and tuning
on a bunch of VCOs. I think proposed linear keyboard control is too
limiting, i.e. MIDI -> CV convertors usually work with expo VCOs. (I'm such
a poor keyboard player)
Considering the increased circuit complexity, I think autotune using a
"D"PLL would be relatively easy.  Using a uC, PIC or even discrete logic.
The circuit would have do disconnect the tuning pots of the vector-VCOs and
supply a tuning voltage corresponding to the reference frequency.
The PLL would consist of a phase discriminator (the lead/lag edge-triggered
type..) one of the VCOs is put into the loop. The filter of the PLL would
be an up/down counter initialized to 0.5 (i.e. 127 in case of 8-Bit) the
output of the counter would be connected to a register followed by a dac.
(Or a dac and a S&H.) The counter would be incremented on a lagging pulse
and decremented on a leading pulse. The output of the DAC is the correction
voltage for the VCO. No need to copy the circuit n-times just the "memory"
for the error signal has to be copied. Either n-times register and dac or n
S&Hs. I'd certainly opt for the S&Hs. Could be el-cheapo 4051 + caps.
I think this could even be done using discrete logic, but with a micro one
can simply put it into one RAM, and refresh the S&Hs from time to time with
the stored values so that the effects of leakage would be minimal. 

The circuit that I tried to describe could of course find application in an
"ordinary" modular or polysynth.

Finally I think that I'm not going to own a PM any time soon, but I have
the idea of haveing mostly all-VC modules and loads of uncommited VCAs on
my modular.
I think it's somehow wasted if you have 5 VCOs and cannot crossmodulate two
of them, use one as LFO and get a bassline from the next one.

Bye
 René

             ,   : (uzs159 at uni-bonn.de)
|") [" |\ | ["   : (http://www.uni-bonn.de/~uzs159)
|"\ [_ | \| [_   : (http://members.xoom.com/Rene_Schmitz)
                       


____________________________________________________________________
Get your own FREE, personal Netscape WebMail account today at http://webmail.netscape.com.