Self-Oscillating Sine FM

Hey everyone, first time posting here. I'm a synthesis student at college in Texas and we've been covering FM synthesis the last few weeks. The professor mentioned self-oscillating FM one day, and I thought I'd give it a shot in Reaktor, but I seem to be running into an issue. Aside from having no idea if I'm even going down the right path, I couldn't find anything online on how to accomplish this, (or if it's even a thing). Not so much as a single diagram!

So the issue I'm seeming to run into is that once my index of modulation gets to a certain point it begins to affect the pitch of the fundamental frequency. The min and max for the index are set to 0-200; perhaps that's just too large and wrap around is screwing with everything? I'm not super familiar/comfortable with the maths surrounding this all, hence why I'm reaching out!

If anyone could help me crack this puzzle I'd greatly appreciate it! Here's a screen shot of the ensemble.

 

What you have there is a self modulating sine not a self oscillating sine

 

Now I have more time...

There are (at least?) two common ways to create a sine wave "oscillator".
#1 Generate a ramp or saw waveform and use a waveshaper to convert it into a sine wave.
#2 Build a filter with feedback that can self oscillate and generates a sine wave as a result.

I'm not sure if your professor is referring to #2 of something else... maybe the idea of FM modulating the sine oscillator with it's own output (to create some sort of pseudo saw) I would call that self modulation rather than self oscillation.

#1 is maybe more common in hardware?, and definitely more common in software.
#2 some hardware sine oscs work this way and are great - e.g. Doepfer A-110-4.

In hardware its difficult to get a really pure sinewave using #1, but maybe easier using #2
In software it's easy to get a really pure sinewave using #1 - just send the ramp through an accurate math sine function!... but difficult with #2 - this is because of aliasing, you need some sort of clipping in the feedback to limit the self resonance, and harmonics resulting from from the clipping are magnified by the feedback process, and are hard to prevent - these don't exist in the hardware version, so in software you get an impure sine with various extra harmonics.

 

Here's an example using two self oscillating Monark filters and a Bento seq8. try tweaking the FM amounts no the two filters and or the cutoff settings. Lots of interesting results to be had - e.g. try modulating the FM amount with envelopes or LFO.

 

I think he means phase modulation , with good ol primary

 

Thank you, can't wait to take a look at this later today!

 

i gotta imagine your prof meant an oscillator that modulates itself, and not a sine wave from a filter.

 

I think gentleclockdivider's structure shows what your prof may have been talking about. If you crank the feedback and watch the waveform on a scope, you'll see it morph from sine to saw to an interesting slope that rapidly jumps above and below zero. That might be what was meant by self-oscillation. Beyond that it starts to sound like noise. Very interesting stuff.

 

Is this a morphing waveform?

 

Yeah. I think SubHarmonic generates its waveforms this way (aside from the Formant wave).

 

It's phase modulation ( yamaha dx style ) made with primary modules , and yes you can morph from sine to saw if you modulate the feedback amount
Phase modulating an operator by itself gives us a saw wave
And modulating it by itself ( rectified ) gives us a square .
I 've read that yamaha patented this (so called new ) square feedback for their reface dx synths .
The problem however with the feedback system in the reface dx is that it's pre envelope , which basically means if your amp.envelope has a decaying curve to zero , the feedback amount stays the same .
While on the original dx synths , a self feedback operator with a decaying amp envelope will gradually fade from saw to sine , sbecause the amount of gain in the feedback loop is gradually reduced ( because of the envelope )
It's really a pitty they didn't do that on their reface dx

 

Yep , good old subharmonic is indeed phase modulation with core

 

Is this a truly new and patent-able idea, or are Yamaha just doing the usual corporate abuse of patent system thing?
I mean, doesn't this just fall out of the maths, in which case it would be something that is 'obvious' at least to a professional DSP Engineer?

 

It was on gearslutz where a user named ''acreil'' pointed me to the patent , there was mention of negative feedback ( absolute value ) for achieving the square wave ...
He's a real ''pure data'' mastermind (check some videos on you tube from acreil ')
I'll try to search for the post on gearslutz.

 

You could probably square the feedback signal and get similar results.

 

Yes, I tried that already
It's more of a toothy shape than the skewed pulse you get from rectifying it.

 

In my MomoFm ensemble I used the following strucuture inside a feedback loop ,for a continious change between feedback saw , and rectified feedback (square )

 

I try your curcuit in older synths from 2008/2009. IllCognito with it's 6 osc's got real new flavors.

 

A worlds first ?
Primary phase modulation with cables YUY