Fundamental extraction: a slightly different method perhaps?

[earthtonesaudio]

I have a project on the breadboard that requires a decent fundamental extractor, but the "pro" circuits I've looked at are intimidatingly huge, and simple low-pass filtering is problematic because of the guitar's note range (some notes' fundamentals are higher than harmonics of lower notes).

For a different approach, I'm thinking you could take a fundamental canceling circuit (like the Mu-doubler on GEOFEX for example), plus the straight signal, and subtract them from each other to get your fundamental.

Still trying to figure this out... now where's my graph paper?

[Scruffie]

I'd be very interested in where this went, I have some projects on my board that could do with fundamental extraction but the schematics I've seen have been huge and made me cry a little so getting it down to a simpler circuit would be great.

[Mike Burgundy]

Interesting idea!
As a thought - most octave down circuits have trouble tracking lower notes - especially on bass. Higher notes work quite well though. What if you stick a simple octave up before that, and then use a fundamental extractor of choice (companders help it track, that's why they're in there - compress first, find fundamental, expand after). The BOSS OC2 tracks fairly well, especially with a compressor in front, and it goes down 2 octaves - could you get cleaner tracking for lower notes by going one octave up, and then two down? This is more or less the other way 'round from your idea though...

[Taylor]

That's a slick idea, earthtonesaudio, I'll be watching this space.

Have you seen the OC-2 fundamental extractor around here? There's a PCB just for that segment of the OC-2 available, and it's not too huge. I built one and had some trouble with it, but it's because I perfed it. Heard good things about it, though.

As for octave up, then down, I think you'd end up with worse tracking that way, because analog octave up creates at least a little distortion, which makes it harder for the fundex to figure out what note you're actually playing.

[earthtonesaudio]

I just looked at the OC-2 schematic again.  It's pretty cool that the comparators driving the flip-flop have their threshold voltages controlled by the envelope of the signal.  I never noticed that before.  That might have something to do with why the tracking isn't so good on lower notes, because the caps holding the envelope voltage can't discharge quickly enough.  It would be interesting to see if it performed better with a smaller cap/FWR rather than half-wave on each side.

Anyway I've sketched out the idea I described on the original post here, but I can't figure out if it would work.  Breadboard time I think.

Does anyone know if the Mu-Doubler distortion adds simply or in quadrature?

[earthtonesaudio]

Expanding on the idea a bit, essentially it's subtracting an error signal from the source signal.  The "error" being harmonics of the source signal.  It follows that if you had a circuit to make each of the harmonics of the source, in the proper amounts, subtracting them would yield a pure sine of the fundamental frequency.  Of course that's ridiculously hard.  But you can probably get close-ish using a doubler and a tripler as your error source, the tripler slightly lower in output than the doubler.

...the primary advantage being you could relax the requirements of the low-pass filtering of the source signal somewhat.  I don't think it would allow you to eliminate the filtering altogether.