[sdiy] F2V with PIC or AVR?

[Harry Bissell]

Usually its some kind of either fixed LPF (usually a very aggressive one), or some adaptive
or tracking LP filter.  Electro-Harmonix has some examples of each, the guitar micro-synthesizer uses the
18db filter, the Deluxe Octave Multiplexer uses a compressor and tracking filter (the tracking filter
only works with a constant amplitude input signal.  There are other methods, such as a bandpass filter and
zero crossing detection (Roland GR-500), detecting one peak and a zero cross, detecting successive peaks (Gentle Electric),
and other methods. 

I used the New England Digital (Synclavier) detailed in the Calman Gold patent, however I did that for all the wrong reasons :^)
I had to feed back the CV into a tracking filter to reduce jitter.  

and thats just methods for guitar, other instruments are easier or harder than that...

H^) harry



----- Original Message -----
From: David G. Dixon <dixon at interchange.ubc.ca>
To: 'Harry Bissell' <harrybissell at wowway.com>
Cc: lanterma at ece.gatech.edu, synth-diy at dropmix.xs4all.nl
Sent: Mon, 11 Apr 2011 13:24:30 -0400 (EDT)
Subject: RE: [sdiy] F2V with PIC or AVR?

> The really tough part is to condition the signal so that you 
> can get ONLY the fundamental frequency.

So, what are the steps one must take to isolate the fundamental?  I'm
assuming some or all of the following: AC coupling (to centre the signal
about ground), LP filtering (to remove extraneous wiggles which might cross
zero but not represent the fundamental, such as HF resonance), and
comparating with hysteresis (to generate a clean square wave).  The LP
filter could even be tuned to the derived frequency plus a discrete distance
(like an FBI agent on your tail).

> Pitch-to-Voltage is the easy part.  Most of the signal 
> processing involves figuring out when a note begins, when it 
> ends, and when you can no longer trust the input to be valid 
> (enter hold mode etc.)

So, in other words, envelope following.  Is there a lot more to it that
that?  Is the EF best done on the raw signal or the filtered signal?

> The PV-1 will not work (correctly) unless you have a good 
> input signal.


-- 
Harry Bissell & Nora Abdullah 4eva