Further details on the frequency shifter all-pass filter network:
If one was only dealing with two (sine) frequencies it would be possible to use a single all-pass filter and adjust the cutoff point accordingly. However given that ring modulation typically produces multiple frequencies (especially when applied to a complex input signal) a network of filters is needed to cover the entire audible frequency range.
My first searches on the web brought me across Hordijk’s Frequency Shifting patch for the Nord G2 Modular, as well as Jürgen Haible’s descriptions and schematic drawings which filled in a little more detail on the Hilbert transform filter network. Fortunately I also came across an article on Analog Wide Band Audio Phase Shift Networks with a diagram (see fig. 4) and a table of frequencies that I could use as a point of departure.
Via Don Tillman’s collection of Moog Patents I could download a PDF of the Bode Frequency Shifter that Moog produced in the early 70s. That helped fill in the final pieces of the puzzle, showing how the sine/cosine oscillator (which I’d already encountered in Hordijk’s Harmonic Oscillator) connects up with the Hilbert filter network (or ‘Dome Filter’ in Moog parlance) to enable the ring modulation and phase cancellations.
I’ve uploaded a few demos to the Audulus Forum.