MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

FM in Modular?
MUFF WIGGLER Forum Index -> Modular Synth General Discussion Goto page Previous  1, 2 [all]
Author FM in Modular?
drumsofd00m
Hi Moogulator,

it sounds like you know a lot about the topic, but some points confuse me:

- I recently looked into buying Cynthia's ZOe, and IIRC, it's analog.

- Not sure what you mean by "exact in phase" as opposed to "in phase"?

- Furthermore, "in phase" (to me) suggests oscillator relationships, but this would mean a limitation of usable intervals to octaves only, which allow for just a subset of the FM sound palette. So could you maybe clarify if you mean that there needs to be hard sync on a VCO for it to be capable of DX7-/ Chowning-style FM, not in that you would hard sync them to each other, but all of them to the trigger CV of each note?

(In this case I would agree that is required for some sounds. In others, it can take away from the concept of implementing Yamaha-style FM with a certain analog-only charm, because in such all-trigger-synced patches each triggering of a given note will sound exactly the same unless you use routings like "velocity (or random CV) going into a minimal pitchbend on one of the VCOs", etc.)

Danke & greetings
Moogulator
you CAN do phase control in the analog domain, but it's harder to do so. most modules can't fire up 2 or 3 waveforms in phase or control that phase (in degrees)

exact means: analog oscs are free running wave forms, you may be able to reset those and keep the frequency stable but still they won't stay in phase over some time or just some periods which result in another sound / harmonics. that's why FM is made "best" digitally.
-but still it is cool to have analog FM or filter FM, but it's another approach and not as controllable all the way.

in phase 2: not just octaves, it is necessary to keep the sound as it is, if it goes slightly out of phase or changes the phase means there are 2 parameters changing, but you need to control one at once .. that's the reason why phase modulation is (above the audible frequency) the same as fm, so pitches and phase must be perfect and totally in sync. starting at the same phase and if you stop the oscs it should be at the same place, thats something samplers were too cold since without pitch mod and random mod it might feel "dead".. but with FM it's perfect.

no, sync is not necessary - and technically sync isn't really syncing the 2 waves - it's just resetting the waveform to 0 degree as the starting position of the waveform that is synced. - sync isn't envolved in FM synths.
thats for free running VCOs to have some rich harmonics ...
.. the tonal thing is another one - that's just the linearity of the FM between the modulating and the mod oscs (in opposite to anything non linear..)

so it is not about "triggering" it is really about being in phase without sync'ing.

so it's 3 issues or things to be sure about.

right - it' is ok and enough to have just 2-3 analog OSCs free runnung and play, but it will never hit the qualities of controlling the harmonics. since you need the phase and frequency accuracy and never use freerunning oscs at all. if you do, it is still enough to do nice Vince Clarke Metalllic Perc-Sounds ... it sounds wonderful on a System 100m or 700, EMS, Moog, different on all of them ... and it will do the job - and you need more than just 4-5 octaves to get everything from FM .. if you are on a 6 OSC-FM Synth it's about 30-60 times the freq. of some osc.. there are sidebands and other FX.. etc. created by that, but without being in phase and locked you won't need such things since it might get lost..

and you can still FM an osc by 2 or more OSCs etc.. ("Algorithm")


anyway - it all cool, I am really using Xmod and FM a lot on analog synths, but it's just ...

and it's also about the "purist" thing here, with sine waves.
pitch-mod of FM: could be nice.
the coolest thing is to add a VCA to control the env's amount by velocity! or if there's nothing like that - modulate the Filter at self resonance - there's another osc FM'ed and with key-tracking. if you don't keytrack the sound will change as well over the keyboard...(LFO situation)
if you track it with "Oscillators" you have the full range and most LFOs won't reach Main audio OSCs frequencies ...
drumsofd00m
Hey moogulator,

I totally appreciate your indepth reply, but quite frankly, am just as confused as before about the differences you're explaining. Maybe let's talk in German at the next Habby Knobbing or so and see if I get it then ;)

BTW I think the most flexible choice for FM patches is "virtual modular" software because there you get 1) a fully modular environment, contrary to stuff like the DX7, and 2) perfect pitch stability (IF that's what you meant).

(Did I just commit heresy by recommending software in this forum?^^)
Moogulator
ok, maybe we should split it into these:

< Oscillators running free = not in phase - they are not started or "synced" on analog synths.

<linear FM is needed to keep all in tune and to control it because it is seasier to have the same mod amount index for different levels (FM depth) - if not (exp. FM) it would get really hard to have a reference. you CAN do some math or voltage trickery but it is just a thing you don't want if you can have it easy to handle, right?

< if 2 FM'ing OSCs are not starting at the same phase (means from zero to positive - as the start of one period of the whole waveform) it will be the loss of 90% of the harmonic content. thats why PM or PD is basically the same as FM as long as we talk audio frequencies (app. above 40Hz)..

<yep - it is. but only if it's not a freerunning-concept. so you can do that on a Solaris, G2 and Nord Modular (first Gen) but not on the Origin since it does not provide phase stability and it is MADE for free running oscs which makes it as FMable as analogue synths but never beyond even a simple digital 4OP FM Synth or even the Nord Lead 3's FM capabilities.

well - it's software and digital devices that might work best for the buck.
and still there's fun with analog fm until you don't expect the sweetest DX/TG-perfection. wink
Mr. Pete
How do you patch the DIXIE for FM, please?
Can you dumb this down for me? - typically what goes in/out where?
thanx!!! seriously, i just don't get it

misa wrote:
I use a combination of an Intellijel Dixie II, Doepfer LFO A145, Doepfer A-110 Oscillator, Plan B M15 VCO and a synchrodyne as well as my filters (A-120, A-121) in self oscillation mode for FM. There are quite a number of great sounds and timbres hidden in FM, especially if you are willing to look for things beyond harmony...
Navs
Anything you do to change the frequency of a VCO is Frequency Modulation, the results depend on how fast you do it and which input you use.

So, grab two VCOs, patch the sine output of one into the FM input of the other. Go via an attenuator, increase the amount and listen to the sine output of the VCO you're modulating (the carrier) so you can hear the change in colour.

If you use a so-called 'linear FM' input, you can change the colour/ depth of modulation dynamically via a VCA, for example, without getting the bends.
frijitz
Moogulator wrote:
exact means: analog oscs are free running wave forms, you may be able to reset those and keep the frequency stable but still they won't stay in phase over some time or just some periods which result in another sound / harmonics.

I spent a fair amount of time looking at sync effects in my Teezer project:
http://www.electro-music.com/forum/topic-29149.html
I incorporated a variable sync circuit in the module and used this at a low level to sync to a master VCO. There is a tiny bit of distortion because of the sync, yes, but it isn't bothersome at all. I can get wide frequency range sync at very high modulation indices (see the examples in the thread), so it's not hard to get phase stability over a huge range of interesting sounds. Controlling the absolute value of the relative phases could be done by generating a pulse from the SAW of the master, using a variable-level comparator, but that's something for the future.

I disagree strongly with the implication that analog thru-zero FM is somehow inferior to digital just because it isn't exactly the same. It is capable of rock-solid sounds at huge modulation indices over wide ranges of frequencies. Period! In fact, a lot of folks think it sounds better than digital.

But please listen to the demos, watch the videos and decide for yourself:
http://home.comcast.net/~ijfritz/projects/Teezer.htm
http://www.youtube.com/user/frijitz001
Sorry, but to me all this vague, pseudo-theoretical yakkety-yak isn't really helpful in learning about what can be done.

Ian
Mr. Pete
VERY COOL thanks
bubblesound LvL module was very helpful here +/- attenuation
Does it have to be a SINE wave to qualify as FM?
I got some pretty nice sounding results with other waves too

Navs wrote:
Anything you do to change the frequency of a VCO is Frequency Modulation, the results depend on how fast you do it and which input you use.

So, grab two VCOs, patch the sine output of one into the FM input of the other. Go via an attenuator, increase the amount and listen to the sine output of the VCO you're modulating (the carrier) so you can hear the change in colour.

If you use a so-called 'linear FM' input, you can change the colour/ depth of modulation dynamically via a VCA, for example, without getting the bends.
rjungemann
drumsofd00m wrote:
Linear FM is pitch stable, regardless of whether it's through zero or not, and through zero FM is PM.


TZFM is not PM. FM is the time derivative of PM. TZFM differs from regular FM in that the frequency can "go negative" which is counter-intuitive because frequency is a time-dependent measurement, but has the practical effect of mirroring the waveform.

It is true that with the right settings you can get equivalent sounds, but each method responds differently to modulation and to the shape of the modulating waveform.

The DX7 was marketed as being an FM synth but in reality uses PM.
wsy
rjungemann wrote:
drumsofd00m wrote:
Linear FM is pitch stable, regardless of whether it's through zero or not, and through zero FM is PM.


TZFM is not PM. FM is the time derivative of PM. TZFM differs from regular FM in that the frequency can "go negative" which is counter-intuitive because frequency is a time-dependent measurement, but has the practical effect of mirroring the waveform.

It is true that with the right settings you can get equivalent sounds, but each method responds differently to modulation and to the shape of the modulating waveform.

The DX7 was marketed as being an FM synth but in reality uses PM.


Pretty much all true... Chowning's patent wasn't important in that it did FM,
rather that it described how to do it with a minimum of support hardware.
These days, we throw a GHz processor around like a five-dollar bill (which
is what an ARM SOC costs), back then 1 MHz was a $100,000 machine.

The trick is that FM and PM are really two sides of the same sheet of paper. If you
drive an FM modulator with a signal, or drive a phase modulator with the integral
of that signal, you get the same output waveform. Mathematically identical.

Now, the real kisser is that on the DX series, the only waveform you have is sines
(which are identical to cosines with a small timeshift)... AND integral of cosine IS sine.
So it's just a little funky to say the DX series was FM versus PM; it's more of a
"yes" than either in particular. The hardware looked a little more PMish to me
than FMish (where the integrator was), but think of it like this: "Did I divide by
two, or multiply by 1/2 ? "

Now, as to linear (volt/hz) versus exponential (volt/octave) - you can't use an
exponential input to go thru-zero. Why? Because 1 volt/octave raised to a very
large negative number is not a negative number, it's just a very small positive
number (i.e. 0.000001 Hz). Fine for John Cage LFOs, but you're not gonna hear
FM cymbals and bells. You need volt/hz for that.... and you need a circuit
that can handle hitting the limiter in both directions.

The real value of FM/PM is generating INharmonic signals - where there are
multiple frequencies that are NOT simple integer or rational multiples of each other.
Ring modulation can do that (but only two frequencies per ring modulator). So
can hard sync. Best of all is thru-zero FM; it does a much more expansive job,
creating a whole infinite series of frequencies, with interesting roll-offs of amplitude.

Do I want an FM operator? Yes.

How do you control the thing, in a modular context? Good question. I've spent
too much time on it to say it's simple or easy. But I'll keep trying.

- Bill
usw
There are many confusions in your message, wsy confused
Yohda
Quote:
So it's just a little funky to say the DX series was FM versus PM; it's more of a
"yes" than either in particular.

I don't get this : they either use one or the other method.

And they definitely used PM.

PM and FM are related but they still have a lot of differences.

Just one example : try to do feedback modulation with FM, not the same as PM.
Yohda
double post
usw
There are two extreme cases that make having both phase and frequency modulation inputs worthwhile for practical reasons :
- getting a stable fundamental frequency is easier with fm, all needed is a constant voltage, whereas pm requires another oscillator/phasor.
- being able to vary the oscillator's phase to a precise, relative degree within a given period of time is almost impossible to do with fm.

The only things that make them both tricky to use in the analog domain over a wide tessiture or without hiccups, are tracking differences between the oscillators, especially for fm, and slight calibrations/offset issues (this is critical for pm, or you'll get discontinuities/distortions in some applications).
wsy
usw wrote:
There are two extreme cases that make having both phase and frequency modulation inputs worthwhile for practical reasons :
- getting a stable fundamental frequency is easier with fm, all needed is a constant voltage, whereas pm requires another oscillator/phasor.
- being able to vary the oscillator's phase to a precise, relative degree within a given period of time is almost impossible to do with fm.

The only things that make them both tricky to use in the analog domain over a wide tessiture or without hiccups, are tracking differences between the oscillators, especially for fm, and slight calibrations/offset issues (this is critical for pm, or you'll get discontinuities/distortions in some applications).


I'd suggest that you give it a try. Do it in software, and add a slowly varying
noise(infrasonic) to the simulated VCO's base frequency, of a few cents or so
over ten seconds.

Start with a reasonable case: a 3:1 ratio for carrier and modulator. In the digital
realm, this is precise. In the analog (or your simulated analog realm), it's merely
"quite close" - a good analog VCO will hold it's tune to within a few cents per
hour and a few cents per octave (Synthesizers.com claims to within 0.05%
from 256 Hz to 4KHz, one "cent" being about 0.004% shift in frequency -
clearly "within a few cents" is musically usable).

Now examine what the magnitude of the sidebands this error will generate - +/-
the sum and difference - which again, will be at most a few cents. So, instead of
a "pure" 3x frequency, you will hear a pair of frequencies, a couple of cents apart,
at each of the harmonics. Throw it into an FFT and turn your "noising" generator
on and off; the result is well-nigh indistinguishable.

Listen to it! The noised version sounds *better* - it has motion, feel, *life* if you will.
Sure, it's not precise. It's not laboratory T&M quality. It's not meant to be one.

The "beat" caused by the lack of sync and drift on your oscillators
will have a time period on the order of a minute or three; that
assumes that the mere act of triggering the note doesn't cause the PSU voltage to
dip or rise enough to actually wiggle either oscillator's VCO (which I honestly suspect
will be the limiting factor)

Give it a try. It's easy to set this up on AMS and if the version you have doesn't do
thru-zero, I'll send you the patch (I'm the creep that fixed it for the hopefully upcoming
general release).

- Bill
usw
yes, slightly inharmonic ratios and subtle noise modulation are extremely useful smile
wsy
Yohda wrote:
Quote:
So it's just a little funky to say the DX series was FM versus PM; it's more of a
"yes" than either in particular.

I don't get this : they either use one or the other method.

And they definitely used PM.

PM and FM are related but they still have a lot of differences.

Just one example : try to do feedback modulation with FM, not the same as PM.


I was digging deeper into this (and USW's comments) and hmmm.... the WikiPedia talk
page introduces another interesting question: an alternate definition of FM versus
PM, which may be equivalent - "in PM, the modulator and the carrier phase are added;
in FM, they are multiplied".

Looking at the original Armstrong "indirect" or "wideband" FM systems, there's an
arbitrary multiplication factor exists (the varactor sensitivity, for example). So yes,
maybe "FM = modulator multiplies carrier phase" is correct in the Armstrong sense.

If so, then a constant nonzero modulator (a DC voltage) would cause the phase
angle to advance at the FMed rate, and a negative modulator gives a negative
carrier frequency (the sine wave goes backward; just as expected for a TZFM
system) ... but a zero modulator value *freezes* the phase angle, outputting DC.

So, a *useful* TZFM oscillator would need to have a base frequency offset. That's
doable too. The phase angle needs to have infinite precision and extension (i.e. it
never overflows), but in a theoretical analysis, that's free.

Now, is this still any different than PM (<integral-of> modulator) = FM? No, it's identical.
What happened is that the integral in the PM system is now integral of the phase angle in the multiplier-FM system.

In a perfect world, this is fine. But the world is imperfect, and for computational
reasons, the phase angle is not allowed to grow without bound; it needs to
"wrap around" every 360 degrees or else you overflow the digital system
or try to exceed bus voltage on an analog one. So you need to do some
wraparound arranegment, whether digital or analog.

The same issue applies to the "PM of integral of modulation is FM" - the integrator
can grow without bound given a DC input modulation.

So yes, I think I'm reasonably convinced that the alternate definition of

PM := "modulation ADDS to the phase clock"
FM := "modulation MULTIPLIES the phase clock"

is forward-and-backward equivalent to

"PM with the integrated value of the modulator is identical to FM"

Are we in violent agreement yet? Or do we have to do the ritual of snackfood
target practice?

- Bill
MUFF WIGGLER Forum Index -> Modular Synth General Discussion Goto page Previous  1, 2 [all]
Page 2 of 2
Powered by phpBB © phpBB Group