Balanced/Four-Quadrant/'Ring' Modulation vs Frequency Shifting?

[Snufflepuff]

At one point I was reading Suzanne Ciani's "String Patch" in her famous grant proposal docs. She mentioned getting superior richness and realism from a frequency shifting module. I was baffled as to how she got a consistent timbral shift with a changing melody, given that none of the frequency shifters I'd ever seen tracked 1 volt/octave in a consistent way. Once I read about the Buchla module in question, it finally made sense: said module allows the user to plug in a second signal (in addition to the processed signal) whose frequency determines the amplitude of the frequency shift. So Ciani could just plug in a. copy of the melody to that input and get a shift equal to the fundamental pitch.

But wait, is that any different than (theoretical) balanced/four-quadrant modulation with a sine wave equal in frequency to the carrier wave?

I say theoretical because I'm not sure which modules produce the actual theoretical sum and difference of two input signals. P. Elsea's book includes the frequency spectrum of a diode-produced ring modulation, and the harmonics produced are far, far more than sum and difference and much less organized.



In VCV Rack, I tried balanced modulation with Blinds and got only a sum of the two input signals. I didn't expect Warps' "crude digital model" of a diode to produce a sum and difference (it didn't) but I was expecting its not-crude digital module to get closer. Instead, that algorithm produced a very trebly mass of upper harmonics.

Do any analog modules produce the theoretical sum and difference? How does an analog frequency shifter differ from a ring modulator? Thanks.

[Yes Powder]

A frequency shifter produces sum and difference products as separate signals, rather than a four-quadrant multiplier which produces the sum and difference within the same signal. The telecommunications equivalent of this is Single Sideband Modulation.

A transformer/diode ring modulator is a very primitive device and is not going to give you a perfect sum-and-difference, since the diodes have a minimum voltage which must be met to pass current. This means some irregularities will be introduced as the zero-point is crossed.

The Doepfer A-114 is four-quadrant multiplier based, and will give a very clean product.

[Wavtekt]

"How does an analog frequency shifter differ from a ring modulator?"

Ring modulators are used in part of Frequency shifter's circuit ... see Doepfer's A-126-2 block diagram.
https://doepfer.de/a1262.htm

Analog ring modulator that use diodes aren't that perfect, so they tend to give additional color.

[Garnoth]

You can patch a frequency shifter by using a sine as a carrier for a ring modulator, take a second ring modulator and take the cosine (90 degrees phase-shifted sine) as a carrier on the second one. Taking the sum of the outputs gives you the upshift and taking the difference gives you the downshift. So you can patch it using a quadrature sine oscillator, two ring modulators and a mixer/inverter. As said above, a ring modulator will give you both sidebands in a single signal.

[Yes Powder]

You can patch a frequency shifter by using a sine as a carrier for a ring modulator, take the output to a second ring modulator and take the cosine (90 degrees phase-shifted sine) as a carrier on the second one. Taking the sum of the outputs gives you the upshift and taking the difference gives you the downshift. So you can patch it using a quadrature sine oscillator, two ring modulators and a mixer/inverter. As said above, a ring modulator will give you both sidebands in a single signal.

Can you post a sample of the result of this, modulating something recognizable like a voice? I'm intrigued by this, but it seems incomplete; I thought frequency shifting required the Audio input to be phase-shifted as well, usually being accomplished by a Dome filter. I'd like to hear and maybe analyze the results before I rush to throw something capable of quadrature oscillations into my rack.

[Garnoth]

You can patch a frequency shifter by using a sine as a carrier for a ring modulator, take the output to a second ring modulator and take the cosine (90 degrees phase-shifted sine) as a carrier on the second one. Taking the sum of the outputs gives you the upshift and taking the difference gives you the downshift. So you can patch it using a quadrature sine oscillator, two ring modulators and a mixer/inverter. As said above, a ring modulator will give you both sidebands in a single signal.

Can you post a sample of the result of this, modulating something recognizable like a voice? I'm intrigued by this, but it seems incomplete; I thought frequency shifting required the Audio input to be phase-shifted as well, usually being accomplished by a Dome filter. I'd like to hear and maybe analyze the results before I rush to throw something capable of quadrature oscillations into my rack.

Ah you’re completely right, this patch only works with two quadrature sine inputs. Otherwise you do need to phase shift the audio for the second ring modulator and that’s not easy to achieve.

[cloudleft]

I’m trying to do the above mentioned frequency shifter setup with two doepfer balanced modulators (A-133-2), two sine VCOs at 90 degrees apart, and various combinations of summing and inverting and not getting freq shifty enough sounds. Sounds vaguely ring moddy though. Is phase shifting the input required to make this thing work?

[Synthiq]

Yes, the input signal must also be split into two signals with 90 degrees difference. If more information is of interest, read the US patent 4,399,326 covering the Bode frequency shifter: https://www.freepatentsonline.com/4399326.html

[Snufflepuff]

Thank you, everyone.