ratio 1:1 should result in a saw-like wave but i assume with higher index a real feedback would create more unpredictable sounds…until it ends in almost white noise.
As I think more about it - I’m certain you’re right - it is not equivalent, because the output of the oscillator being fed back on itself, if truly feedback, is no longer a sinusoid (assuming you were using sine oscillators to start with).
So I don’t need any convincing. But if you still feel up to making a video, this is good education for us all.
this is my face trying to decipher FM
I tried to fend off buying an Akemie’s Castle by getting the demo version Bengal. Pretty cool for exploring FM beyond Operator, on this email machine I’m typing on right now.
@Patatelemage, after watching @NeilParfitt’s 28th video set today and thinking about it for a while, I believe it is possible to have an oscillator (operator) feed back on itself. Have a watch and let me know if you think the result looks correct. It seems to me that it is doing exactly as @kilchhofer described:
Looks good to me @Joe
Earlier version of the FM feedback here using a dedicated channel rather than a global chain
Oh nice! Yeah, I guess the advantage here is that you don’t have to worry about running out of channels (just CPU). Looking over the original DX7 algorithm chart, I guess they never had more than one Op with a feedback loop. Though in a couple of algorithms, several ops outputs are mixed and then fed back into an Op.
Oh, but wait, this isn’t the DX-7. It’s the ER-301! We could have all kinds of feedback loops in our custom FM algorithms (or other patches). Hmm…
I made a monster FM patch last night with all kinds of weirdness - I didn’t think it through properly and just half randomly threw some sine osc units together. After a while I was getting some very nice FM sounds, loads of scope tweaking the parameters, to the point it was doing things that I had no idea how it was doing them.
One effect in particular had me wondering if it was just an emergent effect from the interoperating nodes or if there was a ‘slowing down’ of the calculations, it sounded really like tape warble!
It’s definitely worth adding the noise and filter operators as described in that other thread, they totally add another dimension and make a huge difference to the range of possible sounds!
I should probably reiterate the word of caution about your speakers and the possible very high levels that can occur doing this kind of experimentation.
It really is good fun!!
I need to re-read the other thread all the way through. Haven’t read it since before I got the 301.
Quite a lot has changed since then, but the general ideas are still relevant.
I can’t quite get my head around if the new fm options on the sine osc actually improve resolution or not, I’d have to load up the old firmware and compare, unless anyone has a definitive answer?
Yes it seems to work,i ´ll do some experiments today.
Thats better than using another out because of the delay between an out and an in.
Local chain seems to be the soluce.
A little comparison between, concerning feedbackFM :
The Zero and the Solaris have the same sort of behavior, going in bass freq then suddenly cut.
The SY77 only has 7 values for the modulation VCA.
The ER301 is capable to do what is doing the SY and much more, but is not acting the same as Zero and Solaris does.
In my memory the DX7 was capable to go in the chaos sound of the ER301, but my DX7 died a long time ago, and my memory does what she wants.
Very interesting… thanks for doing this
I’m not sure what to say to be honest… they all sound good!!
As mentioned above, I would recommend trying phase modulation - very similar, but you may get different outcomes, it’s possible that this is what the Zero and Solaris are doing, but it really is just a wild guess, I don;t actually know this.
Speaking of phase modulation, I tried adding PM to the Sine Osc last night. It has much more stable feedback behavior.
Ooh - I’m eagerly awaiting built in PM. In my experience (mostly using the Cyclonix Shapeshifter), the result can be very similar to FM, but it is much easier to maintain control of the pitch across a large range since you aren’t actually modulating frequency.
This is fascinating. Thank you for doing this. You are ultimately going to cause me to pull my DX7 mk II out of the closet and see how it behaves…
Intuitively I think this makes sense because you’re not trying to programatically recreate an electrical circuit with very tight constraints and instant response time, rather the phase offset naturally absorbs any slight delay there may be. Is there any truth to this?
I had loads of fun with it a couple of months back - should dust off that custom unit again
I’m no expert and my aging memory plays tricks on me sometimes. But if my memory is on today, I think the ER-301 and SY77 are behaving exactly as I’d expect them to for self-feedback with linear TZFM. The Zeroscillator and Solaris appear to be doing something else? Different wave shape transformation than I’d expect and they are also changing the pitch…
Here’s one more point of comparison. Intellijel Cyclonix Shapeshifter. Osc 1 is multed, patched to the scope and audio output as well as to the Shapeshifter’s FM input. Sweep of the feedback amount. Seems to behave most like ER-301 and SY77 to me. What happens after it passes unity gain seems more gnarly and digital than what happens with the ER-301 though.
Heheh, and FPGAs in the Shapeshiter right?
the dx7 uses phase modulation too, not frequency modulation. thanks @Joe for your demonstration on how to setup a feedback loop. i really like the behavior shortly before it gets weird and by inserting a filter into the path i’m able to expand on these weird tones, especially by introducing resonance.