#96

Yeah the thing to remember is not only do you want to control the amount of input from one oscillator to the next, but controlling modulator and carrier pitch (within a specified range) is very important. A lot of the work I was doing was around 2 modulators in a series, the middle one being low frequency that I would ramp from an audio range (like 120hz) down to LFO range (.25hz) with a quick attack and a variable release. this would give you a pretty good impact noise and then some cool wobble as the middle carrier acted like a pitch LFO. Add some enveloped feedback to the middle carrier and just with 3 oscillators and a bunch of control points you got a massive range of sounds.

2 Likes
#97

I like some things that “DX-style” FM can do, but modular really opened my third eye to other kinds of FM. I’m not even in it for the noise that much, but there’s something gorgeous about linear TZFM on the Hertz Donut when the modulator is off-tuned a little, and in whatever the hell E370 is doing in its linear TZFM that makes it so angry, and in exponential FM on the Double Helix whether using a sine, saw or square as the modulator, and in getting FM to fight with oscillator sync or a PLL…

1 Like
#98

… I am ABSOLUTELY gonna see if the Pitch EG on my DX7 is capable of doing this tonight bc that seems like it would be amazing fun. If not I guess I finally have a great reason to build an FM Max patch.

ETA: it can’t. hello MSP!

Abstractly as I accidentally get sucked further and further into the intricacies of DSP for music, I’m beginning to really appreciate the usefulness of FM, especially when digital was more uncharted territory—the fact that if your operators are sines you get a finite spectra really improves the situation of aliasing, since there will only be a finite (and usually very small) number of aliased harmonics if there are any.

1 Like
#99

hey @zebra @cannc

is this possible to run on norns? what changes might be needed to use this as a synth engine?

2 Likes
#100

yeah its possible, would need a fair amount of cleaning up but nothing too substantial

2 Likes
#101

@Starthief Do you feel like accounting for your provocation that “2-op FM is better than 4-op or 6-op”? I’m intrigued.

1 Like
#102

It’s partially just personal preference. In a certain sense, I like working with simpler parts and assembling a whole from them.

I had a DX-100 back in the day, and then a DS-8. Later when plugins came around, I never really recaptured the love I had for those. FM was something I used occasionally for weird digital breakup noises, basses or glassy pads (usually with two operators, or an algorithm with a pair of detuned modulator-carrier pairs) but not a major part of my sound.

One of the first things I did when I got my first module (Tides) was to FM my Microbrute with it. And that was pretty glorious. I discovered there are all kinds of flavors of FM beyond Yamaha-style phase modulation. It was a bit like when I realized Chinese food goes way beyond the fried rice my parents had always ordered :laughing: And then I got deep into West Coast style complex oscillators, wavefolders and LPGs… 2-op FM at its best.

I tried a Volca FM and a Reface DX a few months ago and found I just didn’t like either all that much; I also haven’t been messing with FM8 or other “traditional” FM software at all in the last couple of years.

So now I have:

  • Hertz Donut, which is great for dynamic linear TZFM and those Yamaha-ish bell-like tones and has noisy digital shapes, but there is no restriction on frequency ratios so you can set up wildly inharmonic tones or some nice subtle beating, or you can sequence them separately, you can hardsync the VCOs to each other or something else, cross-modulate, and so on. Or FM eiither or both VCOs from another source, for linear or exponential FM. But most often I just use its internal FM bus for 2-Op FM.

  • E370, in which each of its four VCOs has a sort of tame and gentle 2-Op FM mode but you can apply them to any wavetable, not just sines. It also has a phase mode where you can use audio rate phase modulation for a similar, but subtly different effect. And on top of that, it’s got FM inputs, which can be assigned to exponential or linear TZFM modes at three different strengths, and the linear comes out really dirty and weird compared to everything else even if you filter the modulator signal. And since it’s a wavetable synth, you can give it banks that were created through FM synthesis and stack it even further. But honestly: with all those possibilities I tend to use it mostly for 2-op, exponential FM.

  • Double Helix, which is an analog saw-core complex oscillator that does exponential FM only, but is full of character. Using saw instead of sine for the carrier sounds pretty nice.

  • Rings, which has a poorly hidden 2-op FM mode, and an exponential FM input

  • Plaits, which has a similar 2-op FM mode.

Also there’s both Aalto and Arturia Buchla Easel V repping the West Coast in software, and now PortaFM for the old Yamaha sound. And I find I still like it, but mostly when I do crazy things with it instead of electric pianos and bass and brass :slight_smile:

8 Likes
#103

Yamaha phase modulated electric piano and brass sounds, while classic, are what I blame for a host of aspersions cast on 80s synth-centric music. Our ears got tired of those sounds long before they stopped getting overused.

2 Likes
#104

just curious what was it called? the pd patch i mean

#105

wow, I’m very glad I asked, thanks for the in-depth response!

It’s interesting, I’m finding that since the DX7 does its little 80s best to appeal to the pianists of the world, I find myself programming sounds that respond well to pianistic expressions. Mental note… see if I can get it to growl…

The various kinds of FM you describe do sound really exciting, and I think the modular ones really well suited for a modular approach.

@jasonw22 I definitely have my own e-piano rendition (really buzzy and dark, since I was trying to mimic the sound in a Cure song) and… I’m not sorry :joy:

7 Likes
#106

An important point there. Most of the sorts of timbres I work with aren’t ideally suited to polyphony, much less pianistic playing. If I were playing chords, or runs of notes that overlap, I would be much less interested in slightly-offtuned ratios and thick textures, and would probably appreciate those extra operators to bring in more transients.

I started out noodling on keyboards because that’s what was there – the family’s beat-up old upright piano, the little Magnus chord organ my grandma got me, the good ol’ Casio VL-1.

Back in 1988-89 I played keyboards (badly) in my (bad) high school jazz band and had some crash Suzuki Piano lessons in an attempt to suck a little bit less. I can jam a little bit but I’m definitely not a piano player in mindset. I’m probably at the least keyboard-player-ish point now that I’ve been since I was 8 years old or so :slight_smile:

2 Likes
#107

OMG i dont even remember! Most likely something clever like 4OP_Matric_FM. I never published it. It was before there was an easy way to make PD patches widely available.

#108

I would like to ask a really basic analog FM question. What is the difference between a given module’s V/Oct pitch CV in and an FM in? I assume there’s a scaling difference. But, if I run a signal into a ‘pitch’ input, isn’t that modulating frequency?

Not concerned about the practicality of this or trying to get a steady recognizable pitch or bell tone or such, just understanding what kind of effect is happening. Thanks for any insight!

#109

A 1v/o input on a module is an exponential FM input. There is no difference to other exponential FM inputs, apart from a potentially present attenuator knob or internal vca to set the fm depth (you wouldn’t want that on the 1v/o input for obvious reasons).

There is a difference to linear FM inputs, but you probably know that.

3 Likes
#110

There is basically no difference, although an FM input is more likely to have an attenuator, so you can add more or less modulation. A v/Oct input has no attenuator because it is precisely calibrated.
V/Oct VCOs all have an ‘exponential converter’ which turns the linear voltage (0 to 1v is the same as 3 to 4v) into a signal that can control a frequency that is exponential (1 extra volt doubles the frequency, 2 extra volts quadruples it, 3v = 8x frequency). The ‘linear fm’ input bypasses this exponential converter.

6 Likes
#111

when i teach my students about FM synthesis (i usually use Ableton Live Operator for this) one of the aspects i insist mostly on is the expression, how expressive a simple fm patch can be if you start to add velocity to operators, and aftertouch. often they remain highly surprised by how a digital form of synthesis like dx style phase mod can be dramatically more expressive than an analog synth…

7 Likes
#112

well, if fixed-architecture analog synths commonly had SIX envelope generators to control various parameters, maybe it’d be a fair fight :stuck_out_tongue_winking_eye:

5 Likes
#113

i think this may be my favorite “how to” and to show new people. Helped me an absolute tonne

https://www.youtube.com/watch?v=ziFv00PegJg&t=0s&list=WL&index=73

4 Likes
#114

Found this thread since I published a FM7 engine for Norns and have been in the FM spiral for a few weeks… it’s goes deeper than I expected.

Ironically, I don’t actually know much of the DX7 history and I’m learning as I go. The engine is in the master branch of dust.

3 Likes
#115

The master branch of dust has an FM synth that uses a different UGen. I’ll be adding envelopes next week.

@zebra would you happen to know the differences between the FM7 UGen and this DX7Clone?