[Question] Operator connections and feedback

[freq-mod]

Hello. Glad ESFM is finally being de-mystified, turns out this chip is much more impressive than previously thought... Still, few things perplex me..

And there are no pre-set 4-op configurations, as on the OPL3. Instead, ESFM features separate controls for output and modulation level on each operator -

Does that mean users can make any connection of 4 operators, any way they can think of, also emulating OPN family 4op connections (like, op1 modulating op2, op3 and op4 carriers at once, or 4 carriers). Also feedback - on OPL3, AFAIK, only op1 could have the feedback affecting it. What about ESFM? Is it still per channel, or per pair of ops(like envelope) or per op itself (like frequency)?

P.S: one operator can both produce sound, and modulate another operator, simultaneously. - wonder how that one is implemented...

[jwt27]

Does that mean users can make any connection of 4 operators, any way they can think of, also emulating OPN family 4op connections (like, op1 modulating op2, op3 and op4 carriers at once, or 4 carriers). Also feedback - on OPL3, AFAIK, only op1 could have the feedback affecting it. What about ESFM?

Not all combinations are possible - you can't modulate op3 by op1 directly, only via op2. Feedback is only possible on the first op.

I found this patent that shows how it's connected (but doesn't show feedback): https://patents.google.com/patent/US5581045

Maybe I could add an ascii version of that diagram in the document.

Is it still per channel, or per pair of ops(like envelope) or per op itself (like frequency)?

Each op also has it's own envelope, of course - you just can't trigger each individually.

[freq-mod]

Oh, so not every 4op alg is possible, feedback is still like OPL3....

[jwt27]

They had to maintain OPL3 compatibility, so I guess they couldn't make it too dissimilar.