[sdiy] Boss Octave - was Re: Frequency shifted from BBD?

[brianw]

This is very interesting. I had assumed the Boss Octave uses an Envelope Follower and VCA to impart the dynamics of the input onto the synthesized sub-octaves. Seems like the switching / ring-mod would be much cheaper than a VCA. The caveat, of course, is that ring-mod produces sum and difference frequencies, and a lot of those new frequencies need to be filtered (unless you want a really spaced-out effect).

Benjamin mentioned companding to restore dynamics, but I think it would be different from dynamics compression and expansion. Converting the input to a square wave is a form of compression, but the output would need to follow the amplitude of the input, which is more of an Envelope Follower function than an Dynamics Expansion function. i.e. A square wave has no dynamics to compress or expand, so a compander would get rather confused unless it was highly modified.

Brian


On Oct 8, 2024, at 3:54 AM, Roman Sowa wrote:
> Which is basicaly amplitude modulation with supressed carier, or as synth nerds like to call it - ringmodulation. Input is modulated with square wave of half frequency, so the resulting spectrum contains sums and differences of the previously filtered input and square wave spectrum. Upper part can be filtered, and we get shifted down by half fundamental frequency output.
> 
> Roman
> 
> W dniu 2024-10-08 o 12:31, Tom Wiltshire pisze:
>> There's actually one further twist to this. The flip-flop signals are not used directly, but instead used to switch between non-inverted and inverted versions of the input wave. The effect of flipping every other input wavecycle is to produce a new waveform with the fundemental an octave lower. The advantage over using the square directly is that the resulting output changes with changes in the input in a way that the square wave doesn't. This makes the pedal seem more "responsive", since it's not just blasting you with a constant volume square wave all the time - instead the dynamics and harmonics of the octave-down signal can both change.
>> Tom
>>> On 8 Oct 2024, at 08:20, brian wrote:
>>> 
>>> The Boss Octave does not actually shift pitch. The algorithm is completely different.
>>> 
>>> 
>>> The analog signal is boosted to the point of extreme clipping, making it effectively a square wave. There's some EQ ahead of this to favor the fundamental rather than the harmonics, but you've probably noticed that harmonics can still slip through. Once the input signal is clipped to a square wave, it is processed with CMOS gates to cut the frequency in half, and the again to one quarter. These represent one octave down and two octaves down. Each of those are square waves, too, so they're processed with a low-pass filter to remove (redundant) harmonics. Finally, the original input is mixed with the one-octave down and two-octave down signals. Voila!
>>> 
>>> The Boss Octave technique cannot produce any interval other than one octave down (and two octaves down by repeating part of the process).
>>> 
>>> FYI, the digital gate that can drop a square wave by one octave is very simple and cheap, and often there is more than one in the same chip. Thus, the two octaves down signal is practically free.
>>> 
>>> Brian
>>> 
>>> p.s. I made a decent recreation of this in software until I realized that Logic Studio Pro already has such a plugin that works better than mine.
>>> 
>>> On Oct 7, 2024, at 9:49 AM, Didier Leplae wrote:
>>>> Thanks for all the replies to this question! It’s all very interesting. I used to have a Boss Octave pedal back in the 80s. I wonder if it also worked with BBDs.