FM programming tips? Well, this is what I've come up with so far, between the V50 and the Fusion:
* The easiest place to start, and probably the starting point for 90% of FM sounds out there, is a pair of operators - a carrier and a modulator, with a feedback loop around the modulator. In the Fusion, you can have 3 pairs per voice - and unlike the DX7, all three modulators can have feedback.
* If you like sawtooth waveforms, you can produce them with just one operator, with a feedback loop - applying an envelope to the feedback loop sounds not dissimilar to a 12dB/oct filter.
* The difference between a carrier and a modulator is generally one less than the harmonics that get emphasised. So, for example, 1:1 produces sawtooth-like sounds, 2:1 produces square-like sounds, and so on.
* Up to a certain point, feedback in FM acts a bit like resonance in filters, adding warmth and "squidginess" to the sound; however, feedback at too high a level produces white noise.
* Two different modulators feeding into two identical carriers behaves the same as two modulators feeding into a single carrier. Likewise, two identical modulators feeding into two different carriers behaves the same as a single modulator feeding into two different carriers.
* Changing the phase of a modulator with respect to its carrier drastically alters the resulting sound - the effect could be described as akin to PWM, except for sounding nothing like it.
So detuning a modulator slightly with respect to a carrier will create a waveform that changes periodically.
* A stack of more than one modulator quickly gets very difficult to predict... but it gets a lot easier if you start with a stack of two (modulator and carrier) and then add another carrier to the bottom, rather than adding a modulator to the top. (This is something that's difficult to do in FM synths, so well done Alesis! In fact, Alesis have done even better than that, because you can hear what's going on at any point of a stack simply by routing that operator to the output as well. This is a GOOD thing!)
* The Fusion has a filter, of course, so the pressure to achieve ALL of one's tonal variation via FM is considerably lowered; if you get stuck with a sound that's too bright, just use the filter to knock the edge off it.
* And of course, FM is not the only mode in which FM is available! It's just the place in the Fusion where it gets most flexible... and to be honest, most usable.
And as anyone familiar with FM the Yamaha way will realise, the two synths are not the same at all. Here are some differences, both obvious and non-obvious:
* DX operators are set by choosing their frequency ratio; Fusion operators are set by setting a relative pitch. That has two implications - you need a ratio/pitch conversion chart handy; and any ratio that's not a power of 2 can't be expressed exactly on the Fusion. Likewise, DX fixed-frequency operators are set by absolute frequency, whereas Fusion fixed-frequency operators are set by pitch relative to middle C.
* DX output levels are not linear! They're super-exponential, in fact. Fusion output levels ARE linear, though.
* DX envelopes are rate/level; Alesis are ADSR with a delay before the A and a rate at which the sustain decays. DX7 envelopes cannot be exactly duplicated on the Fusion... BUT those of 4-op DXs (except the DX9) CAN, because of their much simplified structure.
Curves of +50% for attacks and -50% for everything else will work well - this simulates the DX's linear envelopes feeding exponential output generators. Actually, the best bet is to use linear envelopes, and set the tracking curve to +50% in the modulation routing.
* Feedback not only works differently on the DX, but sounds different too. The reason is that the Fusion double-samples any operator with a feedback loop on it, and the DX didn't - so a feedback loop on the Fusion has effectively only half a sample's delay on it, rather than a whole sample as on the DX. (It also becomes more prone to aliasing as a result. Did Alesis mean to do that?) To get the same approximate result as a DX, you need to set the feedback level relative to (and apply the same envelope to it as) the output level - for a feedback of 7, set them the same; for 6, halve the feedback level; for 5, halve it again, and so on.
* DX synths have hard-coded modulation routings, which is wasteful if two operators have the same envelope (say), but does allow every operator to have different scalings. If you try to replicate patches that take advantage with that in the Fusion, you will run out of modulation slots - a fully-loaded DX7 patch would probably need about 50-60 routings in the Fusion. A 4-op patch might be reproduceable as is. But it's fair to say no other kind of patch will eat your mod routings as quickly as an FM patch.
Some essential links for understanding FM:
www.soundonsound.com/sos/apr00/articles/synthsecrets.htm and
www.soundonsound.com/sos/may00/articles/synth.htm - part of Gordon Reid's excellent Synth Secrets (links to all of which can be found here:
www.soundonsound.com/sos/allsynthsecrets.htm - this should be considered required reading for anyone interested in synth programming). Also see
www.soundonsound.com/sos/1997_articles/sep97/synthschool3.html for Paul Wiffen's take on it
oak.cats.ohiou.edu/~phillipm/BasicFM.pdf is a PDF with some useful stuff in it
www.angelfire.com/in2/yala/2fmsynth.htm is a masterful guide to both FM synthesis in general and the DX implementation of it. Has some scary maths in it, but if you thought a Bessel was something that went with a mortar, just be reassured that scary mathematicians could theoretically predict the harmonic spectra of any given set of parameters for FM voices - and since inverse Bessel functions do exist, could theoretically produce a set of FM parameters for a given harmonic spectrum, which opens the door to FM-based resynthesis. (But that way lies TRULY scary maths, and probably a good dollop of heuristic programming too...)
www.maths.abdn.ac.uk/~bensondj/html/dx7.html - Dave Benson's huge pile of DX7 information - there are a couple of good FM tutorials there as well.
Is that any use to anyone?
So detuning a modulator slightly with respect to a carrier will create a waveform that changes periodically.