FM The Grid vs. FM-4

[u-u-u]

The Phase-1 osc module has an internal feedback knob right there in the lower right side of the module. This is to enable you to build a pseudo Phase-4 Instrument in the Grid, where they all can mod each other and also feed back into themselves.

How do you do that? I osc A modulates osc B, how can osc B modulate osc A again? This is possible in Phase 4, but not in the Grid with two Phase-1 osc modules.
I think that's a restriction of the Grid. And I understand that it would things very complicated internally so I'm content with how it is now. Just think that is NOT possible to build a pseudo Phase-4 in the Grid.
However maybe we can code our own modules one day? (would be interested) Then it would indeed be possible to do a Phase-4 module if desired - or other cool feedback (e.g. karplus strong) stuff I'm missing a low-pass filter in the feedback path of the Comb filter.

[Tj Shredder]

The two algorithms of a DX7 which have feedback across more than one oscillator (4 and 6) I realized with a long delay set to lowest time in the feedback. That will have a different sound/result than a native phase-2 or phase-3, but I doubt that it would prevent you from creating the sound you are after... just get patching and listen, a lot of those „limitations“ are theoretical and certainly hard to justify as deal breaker...

[pdxindy]

but I doubt that it would prevent you from creating the sound you are after...

To me that is the thing... I find it more fruitful to focus on the sound I want than trying to duplicate a specific technique to get there. The various Grid Osc's are each much more flexible than the usual FM Operator.

[Tj Shredder]

To contrary, if I would have to explain/teach someone how to program a DX7, I would do it in the Grid. There you can actually see whats going on and how a waveform is changing while you tweak a parameter. Just throw one of the display objects in and watch and listen.
You want to know how different a one sample feedback versus a delayed feedback sounds and looks like? Just patch it and stop asking on the forum, you‘ll get a better answer from the Grid itself...

[kurt008]

@you all: You are really funny! Seems like you don't look at your synth's you're using every day. How many of them use the possibility to feed back portions of their outputs to some of their inputs? How often did you see a parameter matrix inside them? What the heck do you think this damned thing is good for? Looking beautiful????? Feedback in synth's is quite wide-spread and offers many functions you're using quite often. And discussing about pro and con of finding solutions for feedback enabling is like discussing the necessity of subtractive synthesis. You can live without it, but why??? BTW... you often mentioned the possibility to feedback phase1's signal to itself... but you're sure to know what the discussion is all about...?!?!? If you use several (2 or more, hopefully you understood that!) sound sources like oscillators, you might find it useful to feed a possibly processed portion of each output so the other's input?!?!? Could it be, that this give interesting sounds? Could it also be, that this method is already in use of FM4 as well as in Phase4 inside BWS, DX7 (and I'm pretty sure DEXED, a soft synth derived from DX7 as well), ... countless other soft- and hardware synths since a loooong time? So why discussing this very common technique???Only because Bitwig isn't willing to implement it inside The Grid?

[Tj Shredder]

It is implemented! And if you know anything about DSP programming you know also why it is made the way it is, and why its different in native synth engines...

[pdxindy]

And discussing about pro and con of finding solutions for feedback enabling is like discussing the necessity of subtractive synthesis. You can live without it, but why???

Very few synths can have arbitrary feedback loops. In most cases it is limited to specific parameters in specific configurations. It is hardwired.

[wavedigit]

My problem was and still is the inability to use the output of these oscillators to mix them back to their inputs (some kind of feedback, which is the problem according to Bitwig Support). This would be necessary if one wants to make a similar mix matrix like FM4 comes with. Missing this matrix the whole concept of inter-modulation is nearly useless.

Perhaps I'm missing something, but can't feedback be done by utilizing a delay (set to shortest timing) between the output of a module and its input? IIRC that would introduce a couple of buffers of delay in between, but it would still technically be feedback, right?

[wavedigit]

accidentally doubleposted, oops

[Tj Shredder]

The long delay set to minimum is introducing 1 block/buffer of delay. Actual length depends on your setting. Depending on the patch it might lead to different sounding results in different setups. If you set it to a fixed delay above the block size it should sound the same in different setups as long that delay stays above...
Think of a guitar feedback with an amp. The delay introduced by the speed of sound determines the outcome, different distances sound different...

[x.iso]

this is just limitations with DSP, that's where euroracks win.

[wavedigit]

Ah, I see. That does make sense.

Delay time represented in buffers is going to sound different depending on (global) audio settings.

But here's what I don't understand: When delay time is not represented in buffers but in regular time, adding back the delayed signal on top of the original signal results in notch(es) in the resulting spectrum. All fine so far.
However, is that true as well in case of FM or PM? The delayed signal modulates the original signal in this case instead of getting summed with the original signal.

[Tj Shredder]

In principle its true for any feedback. If you feedback a oscillator into a phase input, it would be the same as having a little offset of the phase. Problem is, that offset is different for different frequencies. As with the combfilter you described (feedback and mix) you have sometimes a phase of 180 degrees which zeros out and sometimes 360 degrees, which doubles up depending on the frequency in the signal which results in the typical comblike frequency response...