a clean-ish sounding non-octave interval (or, a very limited harmonizer)

[parmalee]

i've been somewhat obsessed with various tracking, p-v, octave, and harmonizing circuits for some time now.  however, without resorting to digital tech or fairly involved circuits incorporating BBDs--i'm also obsessed with getting as much into as small an encosure as is possible/practicable--a clean-ish, non-octave interval seems rather difficult to obtain.  the designs incorporating PLLs are interesting, but produce only square waves.

but i stumbled upon something when playing around with merlin b's u-boat design--which is awesome, by the way, and many thanks to merlin for that.  i really wanted 2 octaves down, but not without the considerable loss found in the 2 octave down portions of designs like the boss oc-2.  the problem, as i see it, is that in order to achieve the clean-ish 2nd octave, one must switch the already switched output from the first octave down.  but what if you didn't?  with peak crossing designs, you'd get crap; but i've found that with the merlin's zero-crossing switching design, the results are kinda alright.

in other words, rather than sending the switched output from the 1st octave down to the second divider and then switching again, why not just divide twice and then go to the transistor switching portion of the circuit?  while it might look kinda odd on paper or under a scope, it actually sounds pretty good.

and then i thought:  if dividing by 4 prior to any switching (between straight and inverted phase signals) sounds good, what about dividing by 3 (with, say, a CD4017)?  tried it, and it also sounds pretty decent--to my ears, at least.  dividing by 3 results in an interval of a 4th below the fundamental--either just below, or an octave-and-a-half below (results vary, interestingly).  again, this also looks kinda weird on paper or under a scope, but it seems to work.

of course, you can only get this clean-ish 4th--no 3rds or 5ths.  still, it's a cleanish non-octave interval.  please note: trying this with designs that switch on the peaks (to my knowledge, pretty much every design other than merlin's u-boat) will not likely yield favorable results.

[pinkjimiphoton]

we need schematic!!

[R.G.]

The limitations of square wave sounds were one of the motivations for the Rocktave  variation I posted here about a week ago.  It's a different way to sidestep the question, by additive synthesis on the fundamental by digital means. This design does not just produce square waves. Once I get the resistor ratios tinkered it, it's a pretty fair sawtooth wave. That's how organs do it.

I stopped there, but didn't have to. There was an effect from the UK magazine Electronics and Music Maker that was titled "Harmony Generator". It did the basics of squaring up the input signal, then feeding it to a divider. But the divider was a programmable divider to generate 1/3, 1/4, 1/5, etc. by switch selection. With that kind of thing inserted after the signal squaring but before a PLL frequency multiplier, you generate not octaves up/down, but thirds, fourths, fifths, etc. by the multiply-up/divide-down waveform synthesis process. And get non-square waves in the process.

It's something that's easy enough to do once the conversion to square waves and tracking stuff is done. I think it would add one or two more digital ICs to the Rocktave Octave thingy.

[pinkjimiphoton]

r.g., i have the schematic for that harmony generator thing... aren't some of the chips unobtanium?

[R.G.]

I don't think so. At least not the important one, which is the programmable divider CMOS chip. The rest if it's easily designed around and in fact not necessary. The Rocktave "shell" performs better.

[parmalee]

we need schematic!!

sorry about that.  give me a day or two, and i'll get something posted.

in the meantime, take a look at merlin's u-boat schematic:  U-Boat (actually, you should download the pdf linked--he made some changes for the final version)  notice the second portion of the cd4013--the divider part, not the first stage which is part of the fundamental extractor/phase alignment block.  if you replace that with a cd4017, you can divide by 2,3, or 4 and then proceed to the FET switching and final filtering, buffering, and mixing stages.  but again, i'll try to get a proper schematic up soon.

it's kinda interesting, if you take a look at what the wave should (roughly) look like, it certainly doesn't appear promising musically.  but it seems to pan out nonetheless.  essentially, draw one-and-a-half cycles of a quasi-sine wave (three zero crossings).  next draw one-and-a-half cycles of the inverted wave.  repeat ad nauseum.  it certainly doesn't look terribly useful, but--to my ears, at least--i'm actually getting a consistent and fairly decent 4th interval below the fundamental.

Posted by R.G.
The limitations of square wave sounds were one of the motivations for the Rocktave  variation I posted here about a week ago.

your variation was my inspiration for trying this.  if only i could figure out a way to get clean intervals above the fundamental.

[parmalee]

r.g., i have the schematic for that harmony generator thing... aren't some of the chips unobtanium?

i've put that together and played around with it.  as R.G. notes, the most important bits are the PLL chip and the cmos dividers--it certainly works a lot better when you throw a better fundamental extractor in front of it.  again, i seem to get better results with the extractor from the u-boat rather than the oc-2 types.

[PRR]

look at merlin's u-boat schematic:  U-Boat

Link should be  http://www.valvewizard.co.uk/uboat.html

(Don't use quotes around a URL.)