Archive of the former Yahoo!Groups mailing list: MOTM

previous by date index next by date
previous in topic topic list next in topic

Subject: Re: [motm] Quantizing

From: "J. Larry Hendry" <jlarryh@...>
Date: 2002-02-20

I have l long been thinking about a quantizer.  My thought process was that a very accurate circuit could be built in the 8 bit format IF one used a ADC to get the digital quantities, but avoided the 8 bit DAC by going completely analog on the output.  Then each of the 8 elements could be trimmed accuracy so when you added the elements each together you would have much greater than the accuracy expected from a 8 bit DAC.  I can see Paul shaking his head and sighing at me right now.  But, I have been thinking seriously about a replacement for the 822 when those PCBs are gone that uses pots instead of rotary switches.  So, trimmer count would be considerable less.  Of course, then your basic circuit is a quantizer.  I have done no development yet because of other projects, my need to learn more, and the (cough, cough) significant stock of 822 PCBs I would like to make a dent in first. :)  And, I fully admit to knowing almost nothing about it and being significantly under-qualified for the job.  So, I'll probably play with it anyway.
 
On a related question, where would one normally expect the analog out to track the input relative to step points.  To simplify the example, let's assume 1 volt steps.  Would the output be desired to change from 1 to 2 volts at 1.5 or 1.9999 on the input?  I don't see that is really matters.  But, I am still pretty ignorant on the subject.
 
Seems the 8 bit DAC might be fine for a mono synth and single oscillator, but would never do where oscillators might beat together.  I have a couple of the PAiA 8 bit MIDI 2 CV 8s that I use with stuff other than my MOTM modular.  For my mono synths (like my Micromoog) that 8 bit resolution seems to be OK.
 
Larry
 
----- Original Message -----
From: John Loffink
To: 'sucrosemusic' ; motm@yahoogroups.com
Sent: Tuesday, February 19, 2002 9:12 PM
Subject: RE: [motm] Quantizing

Short answer, no.

 

Long answer:

The miniwave uses a DAC0800LCN.  Full scale error is +-1 Least Significant Bit and nonlinearity is 0.19%.  Just looking at the nonlinearity, 0.19% of 10 volts, or the full scale Miniwave output, is 19 millivolts.  1 volt = 1 octave, 1 semitone = 1/12 = 83 millivolts, and 1 cent = 83mV/100 = .83 mV.  19mV/.83mV = 22.9 cents.  So that’s the accuracy you can expect from the Miniwave.  You may get better results, but that’s not guaranteed.

 

Pitch discrimination varies on the context.  For monophonic lines you may only be able to discern 3-8 cents of pitch resolution.  For chords, where you can hear the beating of the harmonies, pitch discrimination can be better than 1 cent, particularly if you’re using a just tuning.  For equal temperament, 1-2 cent accuracy is adequate.

 

As a comparison, the MOTM-300 VCO published specs correlate to a pitch accuracy of better than 1.4 cents from 50 Hz to 1600 Hz, and better than 4.4 cents from 25 Hz to 6400 Hz.

John Loffink
jloffink@...

-----Original Message-----
From: sucrosemusic [mailto:sucrosemusic@...]
Sent: Monday, February 18, 2002 7:59 PM
To: motm@yahoogroups.com
Subject: [motm] Quantizing

 

Another thing I worry about (thought I don't have a miniwave) is the
relative low resolution of the 8-bit setup in the miniwave... is 255
levels enough to quanitize accurately, in a chromatic way?  Or is it
just better for octave-octave-octave stuff?  Just curious what your
experiences are, and, of course, if there are any "official" plans
for a quantizer.  I agree that using a MV just for quantizing is
probably a waste.
 




Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.