I fell into this KVR thread last night where they were discussing FM. Now, they said that the DX7 was actually Phase Modulation, and not “real” FM, as were all the ones that emulate it, like FM8. Is the digitone “true” FM? Am I real?
It’s phase modulation, just like most other digital FM synths. It’s a more stable method, and allows for the use of feedback in a controlled manner.
The “FM concept” by Chowning was likely implemented as phase modulation by Yamaha since it was cheaper [less CPU cycles] to add than multiply. The use of feedback in the context is probably a happy discovery after the PM implementation!
Culturally, this type of synthesis method is colloquially known as ‘FM synthesis’, which is why we use the term as well.
In a stable digital system however, Linear FM and Phase Modulation can sound very similar. PM has some really good advantages though.
12 Likes
As far as I know, where the operator waveforms are periodic, there’s no real difference between the results of phase modulation and frequency modulation anyway. Assuming digital accuracy (and ignoring the complication of feedback) here!
FM of analogue oscillators is different for a few reasons, not least that the oscillators are likely to be free running (ie the phase relationship of the oscillators will not be the same for every note) and they will be less likely to have such an accurate frequency relationship, owing to tuning differences. These are just two of the reasons why digital FM (or PM) can produce amazing emulations of organ sounds that analogue FM will struggle to match.
So I’m basically of the view that FM and PM are pretty much interchangeable terms in digital synth land. Without being told otherwise, I would always expect any digital FM polysynth (or VSTi) to be based around a PM implementation.
3 Likes
Yep! Although, feedback with LFM even in a digital system will detune the oscillator/operator.
I agree about LFM being more interesting with analog synths, especially the through-zero implementations.
4 Likes
Hi Simon,
quick question, because I just tried to plot this in macOS Grapher. Are we talking about the difference between
y=sin(x*sin(ax)) vs
y=sin(x+sin(ax))?
Thanks for enlightening me.
Best,
Hans
There are a couple of differences to note.
When modulated slowly or by DC, PM will adjust phase while FM will adjust frequency.
Another difference is that when the modulation is removed, in PM the carrier will be back to its original phase while in FM the carrier will be in an arbitrary phase (which is related to how long you have been modulating).
Yet another difference happens when you use non-sine oscilaltors.
But when the operators are sine waves and the modulation is in the audible range there is little sonic difference.
2 Likes
Yamaha caused a lot of confusion even until today by calling their PM FM… And everyone followed.
I read somewhere here that the Preen FM is actual FM and thus an exception
1 Like
FM (linear): y = sin( t * 2 * PI * (F1 + ( A * sin( t * 2 * PI * F2 ) ) ) )
PM: y = sin( ( t * 2 * PI * F1 ) + (A * sin( t * 2 * PI * F2 ) ) )
(i think that’s right 
)
Where F1 is the carrier frequency and F2 is the modulator frequency and A is the modulator amplitude and t is time.
2 Likes
Thanks for the extra info, very interesting, and yes I wasn’t thinking of outside the audio range there.
Linear FM in analogue is more predictable and less chaotic than analogue exponential FM, PM is more predictable than analogue linear FM, especially when you have more than 2 operators/oscillators. I like them all for different things, even though of course there is some overlap.
I’d like to see the Digitone allow external signals to be used as modulators, that would be incredible.
2 Likes