That cool SSI 2130 oscillator chip that many different synth builders have been using lately lists thru-zero linear FM and phase modulation among its features.
Linear FM changes the core frequency up and down equally by some number hertz per volt. So if the modulation signal has a +/-1V range a 200 Hz carrier might swing from 100 to 300 Hz, which averages out to 200. At +/-2V, it will swing from 0 to 400 Hz, which is still in tune, no problem! But at +/-3V, it’ll try to go from -100Hz to 500Hz.
If the carrier oscillator doesn’t support thru-zero, it’s going to have a minimum rate of 0Hz. So the average will be 250Hz, which means it goes sharp. The FM sidebands are “wrong” too.
If it does support thru-zero, “-100 Hz” is just 100 Hz with a reversed phase — so it starts going backwards at 100Hz, and the average frequency works out to 400Hz and it sounds right. The SSI2130 does this using a simple external circuit that feeds a “Time Reverse” input.
But their literature also claims it does phase modulation… and then shows simply using the TZFM with a DC-blocking capacitor (aka a highpass filter) in front. That’s cheating.
*If* you have a sine wave for the modulation signal, and it’s above the cutoff frequency, you’ll get the same results with TZFM and PM. (Except that PM is less strong than TZFM given an equivalent amplitude.)
But there are uses for phase modulation at less than audio rates. For instance, controlling the relative phase of two different oscillators, because you’re going to mix them in some way. Or simply using a slow LFO for gentle modulation.
And if your modulation signal isn’t a sine, you get serious differences.
This is Shapeshifter, with the carrier as a sine wave and the modulator at a 2:1 ratio. First we hear a sine wave as the modulator, using TZFM and then PM. Very similar (aside from the PM having to go through conversion at a lower rate and resolution than the internal workings of the module).
Then at :09 I switch the modulator to a triangle and again, demonstrate TZFM and then PM. Very different sounds there, which is the math at work.
At :18 I switch the modulator to a square and repeat. Notice TZFM with a square sounds a lot like PM with a triangle? A steady linear increase in phase is the same as a static increase in freqeuncy, and a square wave is just flipping up and down between two levels, so this totally makes sense.
And then at :26 the modulator is a sawtooth. It’s pretty awesome sounding with TZFM and nasty with PM.
I realize that doesn’t argue that Sound Semiconductor is cheating people out of something great — the TZFM just sounds better anyway. But I don’t think they should have called it phase modulation.