The Viva Analog Nyquist Aliaser: A what-if question

[Mark Hammer]

I pitched this in the Lounge but thought it might get a better reply here.

Someone in the lounge had asked about the division of the amplifier output between 2 speakers such that phase-splitting in a Class AB amp continued on to the speakers, with each speaker handling a different half cycle.  I didn't think this would work terribly well, but the idea got me thinking.

The Nyquist Aliaser circuit (http://www.lynx.bc.ca/~jc/nyquistAliaser.html) provides for variable "undersampling" of an audio input signal.  Think of it as a very fast sample & hold circuit where, if you turn the sample rate down a bit it starts to get dirty and jittery.

It occurred to me that. although the Nyquist is applied to the whole audio signal, there is no reason is has to be.  That got me thinking: What if the undersampling was applied to only one half-cycle selectively.  The simplest case I can think of would be something like the Green Ringer  circuit, where the 2N5457 used as the Nyquist's gate essentially forms a clock-controlled switch inserted between one of the diodes after the Green Ringer's phase splitter and the base of the transistor where the two half cycles meet again.  Note that to undersample one half cycle, you'd need to flip the diode in that part of the Green Ringer around so that you don't get full-wave rectification, but rather a phase-splitting and recombining.  The high-speed gating gets applied to one half cycle but not the other.

I haven't tried it, so this is a "gedanken" experiment for the moment, not something I can say produces a feasible or interesting sound.  I *am* curious about whether folks think it might have some potential or not.

[aron]

I don't know what it would sound like but you should be able to do this via a plugin on a computer.

[dano12]

How interesting, I just learned today about clipping half of the wave on another thread.

Do you have any theories about what it would sound like?

Also, Bugbrand over in the UK has a few schematics up for his BitCrusher.

http://www.bugbrand.co.uk/pages/electronics.htm

I wonder if that could be a good starting point....

[Arno van der Heijden]

The Bugcrusher looks very interesting (certainly for such a simple circuit).
Too bad there are no soundsamples...

[disantlor]

in a somewhat related note... I wonder what it would sound like to split the wave in half and then delay one half and recombine it later.  I imagine it would completely obliterate the sound, which may be fun.

[dano12]

The Bugcrusher looks very interesting (certainly for such a simple circuit).
Too bad there are no soundsamples...

There are a handful of clips here: http://www.bugbrand.co.uk/pages/sounddevices.htm#bugcrusher

[Eb7+9]

Mark,

one condition imposed is the ability to split the waveform while maintaining absolute relation to DC throughout ... the superFullWave octaver I introduced a while back could be used for disassembling a waveform this way (the feedback res is zero in that case) ... it does so without distortion all the way down to the zero-crossing point since super-diode op-amp circuits do this by nature ... the SFW is a DC coupled circuit if you want it to be, same as the typical FET sampler pair, so it seems that the re-assembly condition can be met using these types of circuit blocks in tandem ... taking this path you'd have to build yourself two different SFW circuits (diodes flipped) for extracting each opposite phase, then you could apply a DC coupled sampling circuit on one channel ... following this a DC level shift would be necessary to reassemble both sides and bring them together at the right intersection to undo the other DC shift taking place in the sampler channel which is set by the Vgs voltage in the follower device ... your idea is doable in principle  ... another variation on this would be to sample each half-wave at different frequencies before reassembly - even LFO'ing each clock if you want to go nuts ... this reminds me, simply decomposing and recomposing the wave sides in different ratios (w/o sampling) might be somewhat interesting harmonically, and fairly easy to pull off ... just sticking a mixer pot between the outputs of these two SFW circuits and you'd have yourself variable lobes !

~JC