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

freq/phase/ring + other modulations with oscillators
MUFF WIGGLER Forum Index -> Modular Synth General Discussion  
Author freq/phase/ring + other modulations with oscillators
lilakmonoke
im doing a lot of sound design experiments with pure data and have found phase modulation being much more musically useful than frequency modulation. its easier to control the sideband harmonics and just sounds better to my ears because its more related to the carrier frequency.

from what i understand from miller puckette mathematically its supposed to be the same because both just move the phase around. they are both actually related to ring modulation because the side band frequencies are centered around the carrier signal. they sound totally different to me which is confusing. just like a casio cz-1 sounds much better than a dx-7 :-) while both are actually phase modulation based ..

read: http://msp.ucsd.edu/techniques/v0.11/book-html/node87.html

im wondering why all analog oscillators have frequency modulation but not phase modulation? i just know the wmd pdo but but it is hybrid digital/analog. is it that difficult to implement?
teezdalien
Analog oscillators generally implement exponential FM, meaning the sidebands aren't spaced evenly around the carrier frequency. Digital implementations are most often linear with evenly spaced sidebands because they're not using the 1v/octave protocol.
For example a modulator signal varying between +1v and -1v causes a carrier set at A440hz to vary between A220hz and A880hz, an asymmetrical modulation 220hz downwards and 440hz upwards, thus detuning the center pitch significantly. Not really ideal for tonally based music. I think Serge have an analogue VCO with a linear implementation.
adam
the fm aid phase modulates any input
listentoaheartbeat
The Malekko Wiard/Richter Oscillator is the only one I am aware of. No idea how they implement it. Would be interesting to know.

Even in the digital domain there are few VCOs which allow for external phase modulation, and the input converters are a bottleneck for audio rate modulation. The only exceptions are supposed to be the Mungo w0 (and the d0 will give you phase modulation with any pair of VCOs), and maybe the Shapeshifter? The e350 does it well, but it breaks apart at higher frequencies and gets a bit noisy in this application. I guess this is also down to general issues of audio rate modulation and wavetable lookup. Jim Clark pointed this out in the Shapeshifter thread I think, stating that there is less aliasing when using audio rate modulation with the calculated waveforms in the Cyclebox.

I think one issue with external phase modulation in general is that the original implementation (DX7 etc.) perfectly locks the phase of the operators. This is hard to implement if you break the operators apart into seperate modules, and even harder in analog.
lilakmonoke
also ring modulation is really interesting and much underrated, i find it mostly used for drones. i think it sounds great even with tonal sequences. im right now figuring out a pulse train oscillator that is ring modulated by a sine = instant formant oscillator. a pulse train also is also based on manipulating the phase of a sine, narrowing the peaks basically.

im guessing analog oscillators are not that much different from digital ones? at least saw core ones must have a phase based saw generator. why is it then so difficult to modulate the phase in the analog domain?
Navs
adam wrote:
the fm aid phase modulates any input


No. Like the Richter, the FM Aid needs a sawtooth input to function properly.

The Richter uses a sawtooth animator circuit. Search for the Electronotes paper to get an idea. It's similar to the Doepfer A-137-2.

The FM Aid uses a different method.

For another approach you could try this:

http://navsmodularlab.blogspot.de/2014/09/patch-tips-27-fm-equivalent. html

As to why more analogue VCOs don't have this, I'd guess it's because it's difficult to get clean waveforms from the processed sawtooth.
Navs
lilakmonoke wrote:
why is it then so difficult to modulate the phase in the analog domain?


It's not. The difficultly is deriving a nice triangle/ sine.

e.g. using the sawtooth animator method, you are cutting and splicing the saw twice to get a triangle after the fact. If you were to build an animator for your own system - i.e. make it trimmable - you can get acceptable results. But the sine can still be buzzy and will only give you roughly the same amount of modulation depth as normal linear FM.

The FM Aid method allows much deeper modulation but also suffers from unclean end waveforms. It's just the nature of the game. Get a Clavia Micromodular and be done with it lol
Graham Hinton
lilakmonoke wrote:
im guessing analog oscillators are not that much different from digital ones? at least saw core ones must have a phase based saw generator. why is it then so difficult to modulate the phase in the analog domain?


The usual sawtooth or triangle core integrates a control current which may be more or less, but does not change polarity so it can only move forward in phase. Same for a digital oscillator with an accumulator. Only a "through zero" VCO changes direction as the modulating wave changes polarity. A digital oscillator would have to signed addition instead of add only.
ear ear
listentoaheartbeat wrote:
I think one issue with external phase modulation in general is that the original implementation (DX7 etc.) perfectly locks the phase of the operators. This is hard to implement if you break the operators apart into seperate modules, and even harder in analog.


Ian Fritz describes phase-locking two Teezers here.
slow_riot
I think both analogue and digital techniques require surrender to their strengths and weaknesses. Phase modulation may work perfectly in software, but the methods which are available to create this in analogue may be limited by different physical laws, restricting the success or failure of such a technique.

I think the Livewire AFG has the most in depth form of phase modulation, splitting a square waveform into positive and negative pulses, with 2 pulse width controls, and 2 controls for relative phase position of each. If you approach the circuit from an accepting frame of mind I'm sure you can find a lot in there.

Frequency modulation is natively something that is perfectly achievable through analogue techniques, and I don't think there is a good reason to rule it out for generation of complex harmonic spectra.
lilakmonoke
Navs wrote:
lilakmonoke wrote:
why is it then so difficult to modulate the phase in the analog domain?


It's not. The difficultly is deriving a nice triangle/ sine.

Get a Clavia Micromodular and be done with it lol


absolutely valid point there, navs. waveshaping via modulation is all based on precise waveforms, phase alignment and simple number frequency relationships .. which are all unanalog qualities.

i advise you all to read the miller puckette book, its a masterpiece of content compression and even with skipping the math like i do you get 90% of the important stuff. what can be done with two frequencies is simply mind expanding to me especially if you can experiment with the example pd patches and hear whats going on. i might post some sound samples soon ...

http://msp.ucsd.edu/techniques/latest/book-html/

.

Quote:
I think the Livewire AFG has the most in depth form of phase modulation, splitting a square waveform into positive and negative pulses, with 2 pulse width controls, and 2 controls for relative phase position of each.


that escaped me completely, i have to have a look at my afg again, thanks!
.
ear ear
This thread is interesting.
lilakmonoke
im renaming this thread into a more general discussion of spectral design with modulating oscillators ...

here is an example of ring modulation that i find quite fascinating: sine a ring modulates sine b which is fix at 280 hz. sine a sweeps from 140 to 280 in 7 minutes and back. all this goes into the schippmann vcf-02 lpf and a long delay.

as ring modulation generates two freq a+b and a-b you are hearing one freq sweep 140 hz - 0 hz and another from 420hz - 560 hz so it goes through all kinds of harmonics/disharmonics. i made this to to "cure" an old archtop jazz guitar, after a week in front of the speaker with this it sounded totally different ...

http://tindeck.com/listen/ahzl

.
.
ear ear
....or the OP's thread from a couple of years ago.
lilakmonoke
;-) exactly! but ive figured out a few things in the meantime ...
ear ear
I thought this was interesting in the Phase Distortion thread:

wsy wrote:
That's the trick, I think. Not just phase speed, but maintaining a total of 360 degrees of phase at the desired frequency even though the nominal 180 degree point is getting shifted back and forth.


and:

Navs wrote:
I'd still like to find an analogue equivalent to phase modulation that doesn't involve comparators i.e. allow me to generate the PDO's PM sine tones, not saws or pulses.


although, in the FM in Modular? thread:

frijitz wrote:
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.
frijitz
lilakmonoke wrote:
im guessing analog oscillators are not that much different from digital ones?

Pretty different, I'd say. (Having designed both.)

Quote:
at least saw core ones must have a phase based saw generator. why is it then so difficult to modulate the phase in the analog domain?

The FM input of an analog VCO modulates the instantaneous frequency of the signal (by changing the rate the integrating cap is charged). If you want phase modulation you need to differentiate your modulating signal (because frequency is the derivitave of phase). IIRC, somebody once reported actually doing this in analog.

But consider this: If your modulating signal is sinusoidal, say sin(wt), then its derivative is wcos(wt). Still a sinusoid, but multiplied by the frequency w. So you have to be very careful with headroom/dynamic range if you try this. I believe this is the reason phase modulation is not usually done in analog.

Phase and frequency modulation are equivalent, but only if your context is a single frequency.

Ian
lilakmonoke
so if i get a certain spectrum with fm at a certain frequency there is a setting with pm that produces the exact same spectrum? how would you figure that out, fm being in frequencies and pm being in index?
cretaceousear
That was interesting to listen to.. thank you

I don't have the energy to think it through right now but was wondering why the pattern shift in first half is in a cycle of around 14 seconds .. must be the speed the peaks are passing each other, right?

Off topic 1:
Is there a reference to the theory of playing sounds at an acoustic instrument to alter its behaviour.. sounds intriguing.. I half heard of it, but thought the changes induced were tiny?

Off topic 2:
I started a Coursera topic on Audio Signal Processing (gave up as my maths was being stretched too far).. but the intro to the Fourier transform might shed light on your quest.. at least I think it might
lilakmonoke
cretaceousear wrote:
That was interesting to listen to.. thank you

I don't have the energy to think it through right now but was wondering why the pattern shift in first half is in a cycle of around 14 seconds .. must be the speed the peaks are passing each other, right?


there are a few things going on in that ringmod example but when you are superimposing two frequencies the most relevant effect comes from interference or moiree patterns. so i think what you are hearing is those two numbers shifting from harmonic ie. no interferences to detuned to disharmonic to harmonic ie. whole number relationships. as the tunings shift very slowly its easy to hear.

Quote:

Off topic 1:
Is there a reference to the theory of playing sounds at an acoustic instrument to alter its behaviour.. sounds intriguing.. I half heard of it, but thought the changes induced were tiny?


i just read about it and tried it myself, with acoustic guitars the difference is huge. this archtop is an old epiphone from the 50s which was hardly played and and has a rather stiff top so there was no low end. guitars also tend to have resonance holes because they are mainly played in certain tunings. the ringmod track goes through all possible frequencies between 0 and 560 hz so it hits resonances that are not possible with 6 strings. any music will work though, just put the guitar on a stand as close as possible to the speaker and wait ...
frijitz
lilakmonoke wrote:
so if i get a certain spectrum with fm at a certain frequency there is a setting with pm that produces the exact same spectrum? how would you figure that out, fm being in frequencies and pm being in index?

The modulation index is defined as

m = d_w / w_m,

where d_w is the change in carrier frequency produced by the modulation and w_m is the frequency of modulation. This can also be written in fractional form as

m = (d_w/w_0)(w_0/w_m),

where w_0 is the unmodulated carrier frequency. In terms of the modulating CV this is also

m = (d_V_b/V_b)(w_0/w_m),

where V_b is the control (bias) voltage.

These formulas are for the case of linear fm. It's not particularly easy to measure or characterize these numbers in a practical analog system, but if you are running simulations you should be able to get the same spectrum.

Warning: some people are "casual" about the meaning of the term "index" and use the modulation depth instead of the correct definition. Big difference.

Ian
Sleipnir
lilakmonoke wrote:

im wondering why all analog oscillators have frequency modulation but not phase modulation? i just know the wmd pdo but but it is hybrid digital/analog. is it that difficult to implement?

I don't know if you're Euro or 5u or what, but you should try out the PDO - that thing is da bomb for complex waveforms (also one of the wildest LFOs out there).
This week I was tinkering in the studio and just for background sound I put 3 slightly offset lfos into the 3 inputs of the Triple Bipolar VCA (ribboned to the PDO) fed into some reverb and just let that go. I listened to it for about 2hrs solid, just enjoying the constantly evolving drone.
something wonderful
I should try that ringmod trick on an ES175 I have laying around - it could use a tune up like that.
lilakmonoke
@ian ... great, im going to see if i can verify that in pure data. i built an óscillator that can do fm/pm/rm but i have a hard time finding parameters for fm/pm that are comparable. index is the maximal phase displacement in pm, correct? thanks again!

@sleipnir ... i have the dpo, its great, i also have a casio cz-1 which is awesome. you can download the ringmod file from tindeck and throw it at your es-175. i had a sixties es-175 that i sold like an idiot because it sounded totally dead. had i only known ... ;-)

here is more ringmod fun with discrete numbers and pure data. the oscillator i use here is straight out of the miller puckette book. its a variable bandwidth pulse oscillator that can go from sine to needle squashed sine pulse + ringmod. in other worlds from bloodless to subhuman. all the numbers im using to animate this are based on even subdivisions so the resulting spectren are not disharmonic but not nice either.

there is also a tremolo and a delay that gets switched around with integers so depending on how the numbers line up you get very different grooves. all this goes into the korgasmotron for final rectification. i think this would sound great through a marshall stack.

sound: http://tindeck.com/listen/nfoa

patch:



pulse train waveform (c):

lilakmonoke
here is another version of this, this time the carrier wave is synched so it becomes discontinuous. im using even subdivisions again but this time divided by three so you get harmonics that are related to pure fifths intervals. keep in mind there is only one saw oscillator and some number magic behind all this.

the track is "humanized" by pushing the ms-20 filter and recording it to tape which never fails to improve things. this is a true shapeshifter oscillator, i find this much more useful than wavetables. it would be great as a digital ringmod module, imagine if you add another level of ringmodulation, instant death metal!

audio: http://tindeck.com/listen/dpnw

carrier waveform:

CZ Rider
lilakmonoke wrote:
im wondering why all analog oscillators have frequency modulation but not phase modulation? i just know the wmd pdo but but it is hybrid digital/analog. is it that difficult to implement?


There was the Aries AR-338 "PMS" oscillator that was based on the SSM2030 chip. They called this Phase Modulated Sync though and may not be exactly the same as PM, but did look interesting.


And a lo-rez pic of the schematic.


Not sure what it would sound like? I have a 15 module Aries system here, but not many of those later Rivera (RMS) designed modules from 1979. Surprised no one has cloned that one, but could be due to the rarity of the SSM2030.
nigel
CZ Rider wrote:

There was the Aries AR-338 "PMS" oscillator that was based on the SSM2030 chip. They called this Phase Modulated Sync though and may not be exactly the same as PM, but did look interesting.

No, it has nothing to do with phase modulation, although it is interesting. Most sync inputs just reset the slave oscillator on the rising or falling edge of the input signal. This circuit feeds the sync input into a comparator. If you input (for example) a triangle wave into the sync, you can effectively generate an edge at any point of the waveform. As a result, the slave oscillator can be offset against the sync oscillator.
You can of course do this with any oscillator which has a sync input, if you have a spare comparator - input an audio frequency triangle wave and an LFO into the comparator, and the comparator output into the other oscillator sync input. You may want to mix the two oscillators to make the effect obvious.
ear ear
CZ Rider wrote:
Not sure what it would sound like? [...] Surprised no one has cloned that one, but could be due to the rarity of the SSM2030.


The Nonlinearcircuits VCO has features from the Aries VCO.
lilakmonoke
these vcos look interesting, especially the nonlinearcircuits one, very sergeseque!

im beginning to think digital its the way to go in modulation oscillators as my pure data synth is sounding extremely nice. here is a sound sample called "antarctica", kind of a ppg wave 2100 ;-)

http://tindeck.com/listen/opgv

this is variable width pulse train oscillator -> double ringmodulator in just tunings -> tremolo -> mono delay -> schippmann vcf-02 vcf -> slightly detuned stereo delay
lilakmonoke
btw: does anybody know any other interesting frequency modulations besides fm/pm/rm? anything that generates a new spectrum out of 2 frequencies ...
extra testicle
it seems like all modulations if done right will make new frequencies. smile have you found any that haven't?

i really like pulse width. sync and pulse code are good too. smile

phase mod is just inverted am? tbh, i get confused on the terminology going between software, hardware, theory, visual + audio...i don't know if i feel like i really need any more modulations! wink then when you start combining them! the "problem" when you're patching up is you can go much faster than you can understand it. (it's not really a problem lol ) then throw some slews in anywhere and start modulating the slew amount...and feedback...

but i think my favorite modulation is direction... it seems like not enough stuff let's you do that without mega patching. what is direction modulation called? amplitude, frequency, phase, sync mod? like string vibration or striking an object...

or jitter? lol the analogy of how it works "physically" that i remember reading is that it's like a wave hitting a wall and it bounces back if not dampened. tbh, that's a pretty great sound to me! it's the kind of thing i'm always taking the time patching up at least...lol

doepfer quad vc lfo gets nice syncing the phase flip with the sync and direction. even though it's only 8 stages i like the a152 voltage addressed switch to help visualize sync to get my sines working.. i usually use nord modular or software, but they get their own weird jitter too... smile

demodulation is really fun also. smile imo both together are best. lol
MUFF WIGGLER Forum Index -> Modular Synth General Discussion  
Page 1 of 2
Powered by phpBB © phpBB Group