Octave Up Achieving... Ideas, Thoughts, Experience & Theory?

[earthtonesaudio]

Based on recent experiments, I've found you can also get octave up-ish tones by frequency modulating the audio, then demodulating with a simple high pass or low pass, rather than the normal bandpass or tuned circuit.  By matching the oscillator's center frequency and the filter frequency "just right," you can get it to output the absolute value of the input waveform.  Also, by adjusting the filter parameters you can sort of blend the octave in or out.

However, it's overly complex and doesn't sound very good...   

[Mark Hammer]

Looking at the Maxim notes referred to ( http://www.maxim-ic.com/appnotes.cfm/an_pk/881 ), and the Escobedo circuit ( http://www.geocities.com/tpe123/folkurban/fuzz/dof.gif ), and marrying that with RG's early experiments, perhaps it might be suitable to use the DOF with a frequency-divided signal (i.e., several parallel paths for "preferred" frequency groupings/bands), and a delay amount for the "other" XOR input that produces a waveform of desired duty-cycle for that band.  As shown, the delay introduced by the 100k/2nf network will result in a duty-cycle for the resulting octave-up which may be suitable for some notes, but not for others.  This seems like one of those cases where band-splitting and tailored delay for the other XOR input would/could lead to consistent duty-cycle across the spectrum.

[earthtonesaudio]

That would be awesome, Mark!

Another approach (which I'm attempting on the breadboard right now) would be to run the signal through two XOR doubling networks, effectively quadrupling the input frequency, then run that signal through a flip flop to divide it back down to the single octave up, but now with 50% duty cycle.

[brett]

Hi
While there are lots of octaves out there, this one should be good (minimal gating, variable fuzz/drive).
It is unverified.  Does someone have time to build it?
http://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=40366&g2_serialNumber=1
The stompswitch pins are 1, 2 and 3 (top to bottom, LHS), 3.4. and 5 (top to bottom, middle) and 7,8 and 9 (top to bottom, RHS).
I'll see if I have a layout somewhere.
cheers

[brett]

Hi again
here's a 3PDT-mounted PCB for the Octavium grandis.  It allows plenty of space for an easy build.
*NB*  I haven't tested this design yet (but it SPICEs up nicely.)
http://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=40374&g2_serialNumber=2
(please excuse and delete the crosshairs.  I should read the DipTrace manual and find out how to avoid them.)

And the corresponding overlay:
http://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=40371&g2_serialNumber=1

Fingers crossed that it works well.
cheers

[murrayatuptown]

Yes, I'll post despite the age (of the post, not me...)

Well, as I read all three pages here, I had answers to thoughts I had about knowledge & experience I did not...

I was reading the data sheet for an SA571 compander in a wireless microphone I was given, and began thinking about the balance circuit for the SA/NE571...it has an inherent squaring artifact that is tuned out to minimize even harmonic distortion.

This led me to the thought about JFET's for squaring. Then I found a circuit board with some AD533's on it (PITA tot unsolder round case), and pondered multiplying a signal with itself and realized the JFET was probably a lot simpler...wondered what the square rooter would sound like...wondered it that is a square root of an amplitude or of the actual freq. Decided it surely must sound like krap and dismissed it.

Today I read here that doubling via FWR has significant side effects and I thought 'well, use a precision recitifier!' and your sanity check tells me to just forget the whole damn idea because my idea of distortion is probably something like a Peppermill on an archtop...
I

So, thank you all for saving me a lot of trouble & expense messing with doublers....

Does a maladjusted compander offer any useful unintended sounds?

I saw a comparator with hysteresis the other day unintentionally creating a doubled signal but of course a square wave...if it's doubled do the odd harmonics contained in a square wave become even after doubling?

I'm talking myself out of some things....

maybe reverb, delay and phase shift are more subtle things for me to check out...

OK one more thought/question....

'Bessel response' filter...not a great LPF, much less attenuation per octave or decade than Butterworth, but constant phase delay...has anyone already tried phase shifting this way? (Maybe not adjustable or not enough range of adjustment?)

I better pack a lunch, cause they aren't free (except in New Hampshire...Eat free or die).

:O)

[Taylor]

I've done some digital squaring (multiplying a signal with itself, not squaring like clipping) and it does indeed square both the frequency and the amplitude (not surprising when you think about what's happening. So you get a dynamically expanded signal along with your new harmonic content.

[Satchmoedie]

All the old octave ups were "fuzz" oriented devices in the 60s. I have found that a lot of the chips from the 80s pertaining to A/d D/A conversion sound better than a lot of the new stuff. I continually buy old Roland or whatever 25 year old rack mounted devices DOA and find usually a solder joint fixes them. Maybe since this was the period at which I was playing 6-24 hours a day I think they sound better. A clean octave up will be elusive at best with analog, unless you wish to just play one note. It actually might fly with some bands, but not very many. I was shocked at how much the old Roland digital delay sounded like my Maestro solid state which I started using for live gigs, and just keep the tube Ep-2 with SOS at home. I suppose with selected parts something might work, but good luck producing it in mass quantities. The theoretical optimum op amp does not exist, nor does any other theoretical perfect device, so I shall continue bashing my head into that brick until I find just the right angle. Ring modulators? Yeah real musical, especially to people who are on mass doses of MMDA. They have their place, but not really useful for a clean octave up analog signal. If digital is done right the sound is darn close. Maybe the older lower bit rate had something to do with it???

[Quackzed]

I've been messing with octaves, and have a few tricks you could try...
basically they involve fwr - then use a diode ladder to 'smooth' the pointy side... take a look... don't think it's been done commercially this way but it seems to hold some promise for a 'smooth-er' octave up.   ---disconnect the ground from the diode ladder to see the difference with and without...
http://tinyurl.com/43mo4ke

huh- tinyurl - who knew? thanks for the tip! 

[amptramp]

Now that's a candidate for tinyurl if I ever saw one!

There are absolute value circuits here that incorporate the diodes into the feedback network of an op amp so that the diode drop is divided by the gain.  This eliminates the gating effect:

http://sound.westhost.com/appnotes/an001.htm

Some people have mentioned multipliers and square-law circuits based on FET's, but there is another possibility: the log / antilog generator.  If you take the signal from a log generator, double it and feed it to an antilog generator, you have a squaring circuit that could be trimmed.  If you want tube sound, you could select 1.5 times the log and get a sound that is musical and contains some of the second harmonics you desire.  Varying the exponent from 2 to something else may produce something more musical than straight square law.