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

Why are internal LFOs so common in phasers?
MUFF WIGGLER Forum Index -> Eurorack Modules  
Author Why are internal LFOs so common in phasers?
Footkerchief
From what I can tell, most Eurorack effects get their modulation via external CV. However, phase shifters seem to be an exception. All of these have at least one internal LFO:
https://www.modulargrid.net/e/xaoc-devices-kamieniec
https://www.modulargrid.net/e/emw-phaser-12
https://www.modulargrid.net/e/emw-phaser-4
https://www.modulargrid.net/e/analogue-systems-rs-400-phase-shifter
https://www.modulargrid.net/e/pittsburgh-modular-phase-shifter
https://www.modulargrid.net/e/cwejman-sph-2-

Does anyone know why this is common?
emdot_ambient
Because without an LFO there is no phasing effect?

Seriously. It would be like selling a chorus effect that didn't include an internal delay circuit.
thermionicjunky
emdot_ambient wrote:
Because without an LFO there is no phasing effect?

Seriously. It would be like selling a chorus effect that didn't include an internal delay circuit.


I'd say it's more like a chorus effect without an LFO. In either case, there is a static delay/shift. My DIY phaser has no internal modulation. My chorus does, but only because JH put it on the PCB. There is no reason to limit these effects to periodic modulation. It's not a guitar pedalboard.
ispeakhopelandic
it's a hold over from guitar pedal implementation
Bogus
A phaser is gonna need CV and free running LFO circuit cost only a fraction more than dirt soooo hihi

Doepfer's A-101-3 (and maybe lesser models) has no internal LFO if that's what you're in the market for.
flo
As far as I can see all the ones you linked also have CV inputs for the center frequency (only have experience with the SPH2, no idea what the mod ins on the EMW is), so what's the problem? seriously, i just don't get it
ispeakhopelandic
Bogus wrote:
A phaser is gonna need CV and free running LFO circuit cost only a fraction more than dirt soooo hihi

Doepfer's A-101-3 (and maybe lesser models) has no internal LFO if that's what you're in the market for.


it's part of what makes the 101-3 so appealing to me.
mojopin
flo wrote:
As far as I can see all the ones you linked also have CV inputs for the center frequency (only have experience with the SPH2, no idea what the mod ins on the EMW is), so what's the problem? seriously, i just don't get it


It would still be nice to be able to bypass the lfo.
flo
Ah, of course! Looks like the Pittsburgh and XAOC don't have a depth control? The others seem to have one, as said I can only speak for the Cwejman! but that one can certainly attenuate the internal modulation to zero. That is a must indeed even with such a nice internal modulation source as on the SPH2 hihi
thetwlo
also, the Frequency Central does NOT:
https://www.modulargrid.net/e/frequency-central-continuum-phaser-ii

same goes for the Doepfer A-125
flo
A true shame seriously, i just don't get it
SOFTWIRE
I love putting stepped sequences into the 2 doepfers ones that i have. And in combination with a resonant filter either before or after SlayerBadger!
mantiwhore
flo wrote:
Looks like (…) XAOC don't have a depth control?


There are Rate and Depth knobs within the internal LFO section on the Kamieniec.
mojopin
mantiwhore wrote:
flo wrote:
Looks like (…) XAOC don't have a depth control?


There are Rate and Depth knobs within the internal LFO section on the Kamieniec.


ooohh, the LFO OUT is a nice touch!
flo
Good thumbs up Couldn't really read the labels on that pic...
Footkerchief
Bogus wrote:
A phaser is gonna need CV and free running LFO circuit cost only a fraction more than dirt soooo hihi


The circuit may be cheap, but the panel controls require extra HP. I have lots of external modulation, but not a lot of free space.
GGW
Why not both?

I've got the EMW Phaser 4 and I can just turn the Depth to zero and put CV into the Mod In's. I find that it is interesting to crack the LFO depth just a bit to get more randomness to the main CV.
mbartkow
The reason for having an internal LFO in phasers may be that often the waveform of the LFO has a special shape optimized for the particular phasing circuit. Depending on the technology (e.g. FETs or vactrols) an optimal LFO may be a hyper-triangular or some other unique shape that takes into account the highly nonlinear response of these elements. Hyper-triangular is rather uncommon in a modular environment and would require several extra modules to replicate.

I've seen many demos of modular phasers on youtube which sound bad or uninspiring, because an inappropriate external modulation has been patched in.
peripatitis
It could also be though, that a phaser in contrast to other processors needs to keep moving. If it stops there is nothing interesting going on or at least nothing meaningful...
thermionicjunky
peripatitis wrote:
It could also be though, that a phaser in contrast to other processors needs to keep moving. If it stops there is nothing interesting going on or at least nothing meaningful...


Sorry, but I strongly disagree. If it never stops, then there is nothing interesting going on. Static phase shifters are useful. Stepped modulation, manual adjustments, envelopes, audio and other CV sources are useful. I prefer to use my shifter as a complex filter rather than that ceaseless swoosh-swish.
cinnamonjuly
Well I just got the Schippman phaser and it has No modulation on board. Initially I was like sad banana but I guess it's more interesting to patch something else in to do modulation.
mbartkow
cinnamonjuly wrote:
I guess it's more interesting to patch something else in to do modulation.

Having an onboard LFO doesn't prevent you from doing so. It gives you more options, not less.
cinnamonjuly
mbartkow wrote:
cinnamonjuly wrote:
I guess it's more interesting to patch something else in to do modulation.

Having an onboard LFO doesn't prevent you from doing so. It gives you more options, not less.


I agree, I was slightly disappointed tbh but having spent that kind of money on it I'm trying to stay positive about it!
Mefistophelees
cinnamonjuly wrote:
Well I just got the Schippman phaser and it has No modulation on board. Initially I was like sad banana but I guess it's more interesting to patch something else in to do modulation.


I've used the Schippmann a lot without any modulation so I've never missed it.

OTOH I had it hooked up to a syncable LFO the other night so I could use it as a rhythmic effect with drum sounds.
moofi
From what I can tell a phaser is nothing but a (5-10ms)-delay (or several) put onto the original signal, thus through the shift in phase and resulting phasecancellation (combfilterlike) creating the typical sound. Like already mentioned the effect would be static if it was just for that, then the LFO modulating the delaytime is responsible for the characteristic wandering.
Footkerchief
mantiwhore wrote:

There are Rate and Depth knobs within the internal LFO section on the Kamieniec.


What does the Rate knob do when using external modulation? I assume the Depth knob attenuates.

Really wish it was 8 HP... doesn't look like the ergonomics would suffer, and I'm not ready to take the odd-numbered-HP plunge.
mantiwhore
Footkerchief wrote:

What does the Rate knob do when using external modulation? I assume the Depth knob attenuates.


The RATE and DEPTH knobs affect only the internal LFO. The sensitivity of the EXT MOD input is fixed at 1V/oct. This allows turning the internal modulation off while still having a full range modulation by external sources, as well as mixing both both modulations to your taste. The internal oscillator is also scaled to 1V/oct and goes up to audio range, thus it may be used for additional FM coloration of the phaser effect.
flo
moofi wrote:
From what I can tell a phaser is nothing but a (5-10ms)-delay (or several) put onto the original signal, thus through the shift in phase and resulting phasecancellation (combfilterlike) creating the typical sound. Like already mentioned the effect would be static if it was just for that, then the LFO modulating the delaytime is responsible for the characteristic wandering.


That would be a flanger. A phaser is made up with an all-pass filter. This article explains the difference nicely: http://www.soundonsound.com/sos/mar06/articles/qa0306_1.htm


ndkent
There is a constant source of confusion since almost day one of phasers. One can get into some finer points of argument, but there is no question that a flanger is delay based and a phaser is all pass filter based.

Phasers historically were meant to be a tapeless simulation of tape based flanging effects but it was soon realized that the effect was different as to the frequencies involved at any time in the cycle.
flo
Wasn't phaser intended as a Leslie emulation? hmmm.....
thermionicjunky
flo wrote:
Wasn't phaser intended as a Leslie emulation? :hmm:


The Uni-Vibe was intended as a Leslie effect, and is a simple phase shifter.
moofi
Aah yes, I see, wasn´t familiar with this specific difference though phaseshifting through an allpass-filter has been familiar and I remember it being mentioned in that context. Basic sense of the LFO modulation principle, wether it´s a phaser or flanger, still applies for answering the original posed question, reason an LFO is usually implemented in both.

flo wrote:
moofi wrote:
From what I can tell a phaser is nothing but a (5-10ms)-delay (or several) put onto the original signal, thus through the shift in phase and resulting phasecancellation (combfilterlike) creating the typical sound. Like already mentioned the effect would be static if it was just for that, then the LFO modulating the delaytime is responsible for the characteristic wandering.


That would be a flanger. A phaser is made up with an all-pass filter. This article explains the difference nicely: http://www.soundonsound.com/sos/mar06/articles/qa0306_1.htm


thetwlo
i understand a modular phaser having an LFO if that is specific to the clone or the designer's intent, so that's what was desired, but the ability to override it is huge.
Personally, I've never cared for LFO's much, at least fast enough to notice, and that's what put me off of phasers for a while. I don't hate that sound, but it always has *that sound* and that is the LFO.
After realizing that phasers are amazing! in feedback loops with crazy CV, just tweaking... they are endless, yet with an LFO the felt like a gimmicky and limited effect.
wsy
moofi wrote:
From what I can tell a phaser is nothing but a (5-10ms)-delay (or several) put onto the original signal, thus through the shift in phase and resulting phasecancellation (combfilterlike) creating the typical sound. Like already mentioned the effect would be static if it was just for that, then the LFO modulating the delaytime is responsible for the characteristic wandering.


Short delays are one way to create a phaser, but the classic method is the "all pass" filter.

That's because delays are expensive (now) or unobtainable (20 years ago) compared to all-pass networks, which require just one op-amp stage, three resistors, and a capacitor (see the wikipedia article).

- Bill
TonvaterJan
And that brings me to the question:

Are there any Allpass-Filter Units in Eurorack-Land?
Because in the Analogue World, the only Allpass Filter-Mode
in a Synthesizer, that I know of, is the Oberheim Xpander/Matrix 12.

I guess only the Doepfer Filter A-106/6 can do that trick, but I guess that´s it,
or am I mistaken?



Greetings, Tonvater
Mort Rouge
ndkent wrote:
There is a constant source of confusion since almost day one of phasers. One can get into some finer points of argument, but there is no question that a flanger is delay based and a phaser is all pass filter based.


I've even often heard this extremely confusing explanation: flanger is delay, phaser is extremely short delay.

Doesn't make any sense.

wsy wrote:
moofi wrote:
From what I can tell a phaser is nothing but a (5-10ms)-delay (or several) put onto the original signal, thus through the shift in phase and resulting phasecancellation (combfilterlike) creating the typical sound. Like already mentioned the effect would be static if it was just for that, then the LFO modulating the delaytime is responsible for the characteristic wandering.


Short delays are one way to create a phaser, but the classic method is the "all pass" filter.

That's because delays are expensive (now) or unobtainable (20 years ago) compared to all-pass networks, which require just one op-amp stage, three resistors, and a capacitor (see the wikipedia article).

- Bill


As stated, delay-based comb-filtering is called flanger. Phaser is built around an all-pass-filter.

TonvaterJan wrote:
And that brings me to the question:

Are there any Allpass-Filter Units in Eurorack-Land?
Because in the Analogue World, the only Allpass Filter-Mode
in a Synthesizer, that I know of, is the Oberheim Xpander/Matrix 12.

I guess only the Doepfer Filter A-106/6 can do that trick, but I guess that´s it,
or am I mistaken?



Greetings, Tonvater


Lot's of alternatives concerning phaser/allpass-filter in euro:

https://www.modulargrid.net/e/modules/browser?SearchName=&SearchVendor =&SearchFunction=21&SearchSecondaryfunction=&SearchTe=&SearchTemethod= max&SearchBuildtype=b&SearchMarketplace=&SearchIsmodeled=&SearchShowot hers=1&order=newest

As well as A-106-6.
mbartkow
wsy wrote:

Short delays are one way to create a phaser, but the classic method is the "all pass" filter.

A simple delay is an allpass filter as well. The important property is that its phase response is linearly proportional to frequency.

wsy wrote:

That's because delays are expensive (now) or unobtainable (20 years ago) compared to all-pass networks, which require just one op-amp stage, three resistors, and a capacitor (see the wikipedia article).

I respectfully disagree here. A short delay is much easier and cheaper to build: a BBD line is just a single chip, as it was 20 years ago. Indeed, it was unavailable in the 1960's, that is why filter-based delays were the dominant option.

Both the flanger and phaser effects are based on similar principle - the delayed signal is mixed with the original, which creates a series of notches in the frequency response, that correspond to frequencies where the delay creates a 180 degree phase difference which yields phase cancellation. Thus, both these effects may be classified as comb filters.

I believe the main difference between the sound of a phaser and flanger is related to the way the notches in frequency response are distributed along the frequency axis. In a flanger, the distribution is linear due to linear phase response, i.e. they create a harmonic pattern. Thus, the cancellation makes a certain periodic signal with all its overtones to be cancelled at the same time and this yields the characteristic metalic sound of a flanger.

In a phaser, the phase response is usually nonlinear, beacuse it is a result of accumulated nonlinear phase responses of each individual filter stage (that typically resembles the arc tan function of frequency). After mixing with the original signal, phase cancellation occurs at a series of frequencies that are not related harmonically.

There is some variety of phaser sounds which is a result of how these notches are distributed, and this is related to how the individual filter stages are tuned wrt each other.

Note that in a flanger, the number of notches is only limited by the audible frequency range and all they are created by a single delay. In a phaser, the number of notches is related to the number of stages divided by two, which is a result of the shape of the phase response arc tan curve. A combination of these two could be an interesting and powerfull effect.

I wonder, how much the appeal of high-order phase shifters (the particular fluid-like timbral effect) is related to the increased number of notches and how much of it depends on the Doppler frequency shift related to the modulation of the group delay. The clue may be in the increased similarity to the timbre of a flanger, which is also attributed to the Doppler effect.
wsy
mbartkow wrote:
wsy wrote:

Short delays are one way to create a phaser, but the classic method is the "all pass" filter.

A simple delay is an allpass filter as well. The important property is that its phase response is linearly proportional to frequency.

wsy wrote:

That's because delays are expensive (now) or unobtainable (20 years ago) compared to all-pass networks, which require just one op-amp stage, three resistors, and a capacitor (see the wikipedia article).

I respectfully disagree here. A short delay is much easier and cheaper to build: a BBD line is just a single chip, as it was 20 years ago. Indeed, it was unavailable in the 1960's, that is why filter-based delays were the dominant option. [...]

Having worked in a lab where we actually used the Reticon BBD arrays for Real Stuff (not audio), I respectfully disagree.

They were expensive as all frak - an engineer's month's salary. They were rare - months-long wait to get them, and we only got them at all because we were
at $BIGCORP and working on $DEFENSEPROJECT, otherwise it would have been years.

Sure, nowadays they're on Ebay. But that's what happens after 20 or 30 years....

Quote:

Both the flanger and phaser effects are based on similar principle - the delayed signal is mixed with the original, which creates a series of notches in the frequency response, that correspond to frequencies where the delay creates a 180 degree phase difference which yields phase cancellation. Thus, both these effects may be classified as comb filters.

I believe the main difference between the sound of a phaser and flanger is related to the way the notches in frequency response are distributed along the frequency axis. In a flanger, the distribution is linear due to linear phase response, i.e. they create a harmonic pattern. Thus, the cancellation makes a certain periodic signal with all its overtones to be cancelled at the same time and this yields the characteristic metalic sound of a flanger.

In a phaser, the phase response is usually nonlinear, beacuse it is a result of accumulated nonlinear phase responses of each individual filter stage (that typically resembles the arc tan function of frequency). After mixing with the original signal, phase cancellation occurs at a series of frequencies that are not related harmonically.

There is some variety of phaser sounds which is a result of how these notches are distributed, and this is related to how the individual filter stages are tuned wrt each other.

Note that in a flanger, the number of notches is only limited by the audible frequency range and all they are created by a single delay. In a phaser, the number of notches is related to the number of stages divided by two, which is a result of the shape of the phase response arc tan curve. A combination of these two could be an interesting and powerfull effect.

I wonder, how much the appeal of high-order phase shifters (the particular fluid-like timbral effect) is related to the increased number of notches and how much of it depends on the Doppler frequency shift related to the modulation of the group delay. The clue may be in the increased similarity to the timbre of a flanger, which is also attributed to the Doppler effect.


You've got the answer already in what you wrote. It is true that the all-pass filter does ONE shift and done per frequency. On the other
hand, when a BBD filter is cancelling frequency (or harmonic) K, it's also cancelling 2K, 3K, 4K... so all of those frequencies drop out
at once, rather than the one-at-a-time of an all-pass.

That's why a "phasey" sound may or may not have high harmonics (usually does) while the flanged sound often sounds very
bandlimited (and also frequency-shifted; that's part of the magic).

Now you have me lusting to listen to the Bubba Hotep soundtrack again. Best flanged guitars EVER, literally brings tears to my
eyes it's so beautiful.

- Bill
thermionicjunky
TonvaterJan wrote:
And that brings me to the question:

Are there any Allpass-Filter Units in Eurorack-Land?
Because in the Analogue World, the only Allpass Filter-Mode
in a Synthesizer, that I know of, is the Oberheim Xpander/Matrix 12.

I guess only the Doepfer Filter A-106/6 can do that trick, but I guess that´s it,
or am I mistaken?



Greetings, Tonvater


Plan B Model 12 State Variable Filter
Doepfer A-101-3 Vactrol Phase Shifter (stage outputs)
Xoac Devices Kamieniec Phase shifter (stage outputs)
Shippmann PHS-28 (stage 8 ouputs)
mbartkow
TonvaterJan wrote:

Are there any Allpass-Filter Units in Eurorack-Land?
Because in the Analogue World, the only Allpass Filter-Mode
in a Synthesizer, that I know of, is the Oberheim Xpander/Matrix 12.

I guess only the Doepfer Filter A-106/6 can do that trick, but I guess that´s it,
or am I mistaken?


One can pretty easily get an allpass response from any 2-pole (12dB/octave) state variable filter, by mixing the original (input) signal with 2x amplified bandpass output. OTOH, if you want a notch (band-reject) response, mix them with 1:1 ratio.

Adding is required, not subtracting, due to the phase shift introduced by the filter.
emdot_ambient
TonvaterJan wrote:
...in the Analogue World, the only Allpass Filter-Mode in a Synthesizer, that I know of, is the Oberheim Xpander/Matrix 12...


Those synths were based on the CEM3320 filter chip, where a 4-pole all-pass filter could be implemented. (http://www.synthtech.com/cem/c3320pdf.pdf)

A lot of '80s analog synths used that chip but most of them didn't bother with the all pass. I've got a small handful of those chips and have a modular design with each pole switchable between LP, HP, and AP.

Too bad those chips are so rare and expensive.
soundwave106
emdot_ambient wrote:
A lot of '80s analog synths used that chip but most of them didn't bother with the all pass.


TBH I can't think of too many examples where an all-pass filter would be very musically useful in a typical *standard* fixed VCO-VCF-VCA synthesizer architecture. I might be wrong (probably am, actually), but wouldn't it be "just a phase shift"? If so, not terribly useful to me at first glance.

All pass filters are of course potentially useful in *combinations*, such as if there is a feedback loop, or if it the APFed signal was mixed either with the unmodified signal or a parallel APF route. Perfect, in other words, for modular type experimenting.
thermionicjunky
soundwave106 wrote:

TBH I can't think of too many examples where an all-pass filter would be very musically useful in a typical *standard* fixed VCO-VCF-VCA synthesizer architecture. I might be wrong (probably am, actually), but wouldn't it be "just a phase shift"? If so, not terribly useful to me at first glance.

All pass filters are of course potentially useful in *combinations*, such as if there is a feedback loop, or if it the APFed signal was mixed either with the unmodified signal or a parallel APF route. Perfect, in other words, for modular type experimenting.


Vibrato is a good application for APFs. Aside from that, feedback loops are great.
mojopin
thermionicjunky wrote:
soundwave106 wrote:

TBH I can't think of too many examples where an all-pass filter would be very musically useful in a typical *standard* fixed VCO-VCF-VCA synthesizer architecture. I might be wrong (probably am, actually), but wouldn't it be "just a phase shift"? If so, not terribly useful to me at first glance.

All pass filters are of course potentially useful in *combinations*, such as if there is a feedback loop, or if it the APFed signal was mixed either with the unmodified signal or a parallel APF route. Perfect, in other words, for modular type experimenting.


Vibrato is a good application for APFs. Aside from that, feedback loops are great.


Can you explain the vibrato patch?
thermionicjunky
mojopin wrote:
thermionicjunky wrote:
soundwave106 wrote:

TBH I can't think of too many examples where an all-pass filter would be very musically useful in a typical *standard* fixed VCO-VCF-VCA synthesizer architecture. I might be wrong (probably am, actually), but wouldn't it be "just a phase shift"? If so, not terribly useful to me at first glance.

All pass filters are of course potentially useful in *combinations*, such as if there is a feedback loop, or if it the APFed signal was mixed either with the unmodified signal or a parallel APF route. Perfect, in other words, for modular type experimenting.


Vibrato is a good application for APFs. Aside from that, feedback loops are great.


Can you explain the vibrato patch?


This is an application that does call for cyclic modulation, or at least a signal that remains in motion. It usually requires just a couple of 100% wet allpass stages in series.
wsy
soundwave106 wrote:
emdot_ambient wrote:
A lot of '80s analog synths used that chip but most of them didn't bother with the all pass.


TBH I can't think of too many examples where an all-pass filter would be very musically useful in a typical *standard* fixed VCO-VCF-VCA synthesizer architecture. I might be wrong (probably am, actually), but wouldn't it be "just a phase shift"? If so, not terribly useful to me at first glance.

All pass filters are of course potentially useful in *combinations*, such as if there is a feedback loop, or if it the APFed signal was mixed either with the unmodified signal or a parallel APF route. Perfect, in other words, for modular type experimenting.


Quite true. The phasing effect (the cancellation of different frequencies) happens only
when you add together the straight-thru signal and some of the all-pass phase-shifted
signals.

Luckily, you can do that with one op-amp (I think).

- Bill
mojopin
thermionicjunky wrote:
mojopin wrote:
thermionicjunky wrote:
soundwave106 wrote:

TBH I can't think of too many examples where an all-pass filter would be very musically useful in a typical *standard* fixed VCO-VCF-VCA synthesizer architecture. I might be wrong (probably am, actually), but wouldn't it be "just a phase shift"? If so, not terribly useful to me at first glance.

All pass filters are of course potentially useful in *combinations*, such as if there is a feedback loop, or if it the APFed signal was mixed either with the unmodified signal or a parallel APF route. Perfect, in other words, for modular type experimenting.


Vibrato is a good application for APFs. Aside from that, feedback loops are great.


Can you explain the vibrato patch?


This is an application that does call for cyclic modulation, or at least a signal that remains in motion. It usually requires just a couple of 100% wet allpass stages in series.


That sounds like just a phaser then and if anything, a tremolo effect. Vibrato can only come from modulating an oscillator.
thermionicjunky
mojopin wrote:
thermionicjunky wrote:
mojopin wrote:
thermionicjunky wrote:
soundwave106 wrote:

TBH I can't think of too many examples where an all-pass filter would be very musically useful in a typical *standard* fixed VCO-VCF-VCA synthesizer architecture. I might be wrong (probably am, actually), but wouldn't it be "just a phase shift"? If so, not terribly useful to me at first glance.

All pass filters are of course potentially useful in *combinations*, such as if there is a feedback loop, or if it the APFed signal was mixed either with the unmodified signal or a parallel APF route. Perfect, in other words, for modular type experimenting.


Vibrato is a good application for APFs. Aside from that, feedback loops are great.


Can you explain the vibrato patch?


This is an application that does call for cyclic modulation, or at least a signal that remains in motion. It usually requires just a couple of 100% wet allpass stages in series.


That sounds like just a phaser then and if anything, a tremolo effect. Vibrato can only come from modulating an oscillator.


That isn't true. Vibrato can be applied to an input by modulating allpass filters or delay lines.
mojopin
Hmm..I will have to try it but I don't understand how filtering can change the harmonic structure. The delay lines, yes, you will get pitch shifting.
thermionicjunky
mojopin wrote:
Hmm..I will have to try it but I don't understand how filtering can change the harmonic structure. The delay lines, yes, you will get pitch shifting.


The word "filter" is throwing you off. Rather, think of it as phase modulation and consider the relationship between phase and frequency. This is how true vibrato circuits have been around since the 50's.
soundwave106
mojopin wrote:
Hmm..I will have to try it but I don't understand how filtering can change the harmonic structure.


All pass filters change the phase of a signal; if modulated with a LFO, yes, there would be an "apparent" slowdown and increase in the frequency. Hadn't thought of that application before.

Again, not so useful on your typical "standard" synth architecture (since you can just change pitch directly and depending on the synth, filter modulation options may not be so great), but useful in more modular applications or maybe even a more complex synth like the Matrix 12 (such as processing external signals or feeding the APF interesting shapes to modulate).
mojopin
Okay..trying to assimilate everything. Someone before likened APF to frequency shifting which would make sense since these are non-linear phase shifts. The linear phase shift of a delay line would give pitch shifting. Regardless of easier methods of vibrato, it is all food for thought and important to understand. Thanks guys.
mbartkow
mojopin wrote:
Okay..trying to assimilate everything. Someone before likened APF to frequency shifting which would make sense since these are non-linear phase shifts. The linear phase shift of a delay line would give pitch shifting. Regardless of easier methods of vibrato, it is all food for thought and important to understand. Thanks guys.

Nonlinear phase response means that the delay introduced by the unit is frequency dependend, or dispersive. In other words, different spectral components are delayed by a different amount of time.

Frequency shifting appears only when the phase response (or delay) changes over time, i.e. when the allpas filter is modulated in a continuous way.
mojopin
Yup, that all makes sense. Now I understand that I wasn't crazy for thinking certain sounds I put through a phaser sounded not quite out-of-tune but off somehow.
wsy
mbartkow wrote:
mojopin wrote:
Okay..trying to assimilate everything. Someone before likened APF to frequency shifting which would make sense since these are non-linear phase shifts. The linear phase shift of a delay line would give pitch shifting. Regardless of easier methods of vibrato, it is all food for thought and important to understand. Thanks guys.

Nonlinear phase response means that the delay introduced by the unit is frequency dependend, or dispersive. In other words, different spectral components are delayed by a different amount of time.

Frequency shifting appears only when the phase response (or delay) changes over time, i.e. when the allpas filter is modulated in a continuous way.


Uh, no. In many ways.

First off, delay line does not have linear phase shift. The phase shift varies with frequency but not linearly. Consider a 1000 Hz tone (1 cycle
per 1 ms) plus a 500 Hz tone (1 cycle per 2 ms) feeding into a 1 ms delay. The 1000 hz tone is delayed by 1 cycle = 360 degrees
of phase, and so adds constructively to itself in the final mix. The 500Hz tone is delayed by 1/2 cycle = 180 degrees of
phase, and cancels itself to zero in the final mix.

Adding another tone at 2000 Hz = 720 degrees phase shift, and that one adds constructively, but adding another tone at
1500 Hz shifts 540 degrees (1.5 times around the circle) and so is the same as 180 degrees of phase shift, and cancels itself to zero.

That's why delay lines are equivalent to comb filters in the steady state.

But delay lines are not dispersive - the delay of any particular wavelength through the delay line is constant. That is NOT true of
a filter network; those are usually dispersive and one frequency will come through faster or slower than another (though
there are filter designs with an inverted all-pass filter network and a low- or high-pass filter network combined; those are
phase-flat because the all-pass is designed to reverse the shift of the low- or high-pass).

As to frequency _shifting_ - no, a filter never shifts energy from one frequency to another. The output power spectrum of a
filter core (absent op-amps) is always less than the input power spectrum (not the voltage; impedance and resonance can
change the voltage arbitrarily by trading it off into current, so you need to look at
power spectrums, not just voltage.

The delay line will only shift frequency if the time delay is modulated; constant time delay lines don't shift frequency.

Finally nonlinearities in the chain (basically thresholds where the signal changes) pretty much always move energy from one
frequency to another, and usually (but not always) up in frequency. That's the typical way it's done (think distortion pedal) and,
interestingly, also how green laser pointers work.

There's an infrared diode laser, and a nonlinear crystal in the optical path that thresholds above a certain power level. The diode laser
"overloads" the crystal, producing harmonic distortion and the green light out is basically the 3rd order harmonic distortion of the
infrared laser input. A filter then removes the eye-damaging infrared and what you have left is a green laser.

So that's why a green laser pointer is like a guitar pedal.

- Bill
meatbeatz
^^^ and this is why I love muffs applause
mbartkow
I'm sorry, but you have it all mixed up.

wsy wrote:

Uh, no. In many ways.

First off, delay line does not have linear phase shift.


Pure delay lines do have linear phase response, because phase response by definition is the product of delay and frequency. If delay is constant for all frequencies (e.g. a non-dispersive delay), the phase response is linear.

But anyways, we are talking about phasers here, which are build from a chain of allpass sections, not delay lines. Analog allpass filters do have nonlinear phase response, and they are certainly dispersive.
See e.g.:
http://www.edn.com/electronics-blogs/living-analog/4375814/All-pass-fi lter-phase-shifter
or any textbook on analog filter design.

Most of your further explanation is based on false asumptions. Eg:
wsy wrote:

As to frequency _shifting_ - no, a filter never shifts energy from one frequency to another.

You are assuming a linear time invariant filter here, which is not the case for phasers that are modulated filters. Modulation is the key to varying group delay of the filter, and this yields frequency shift, quite simply demonstrated by Doppler.

Oh, and btw
wsy wrote:

That's why delay lines are equivalent to comb filters in the steady state.

Actually, it's not. A delay line has a flat amplitude response. In order to get a comb filter you need to mix the delayed signal with the original. And such a structure is not a delay line, but a filter.
moofi
I´m not too much into all that technical stuff here, then, of course you would have to mix the delayed signal with the original one, otherwise it would merely be a delayed signal, while the phasecancellation can only happen when the shifted phase signal is put against the original phase. Basically any simple delay with let´s say dry/wet at 50% and a few miliseconds delay time creates that basic flanging/phasing sound though without movement (combfilter), hence an LFO to modulate the delaytime for wandering.

mbartkow wrote:
[...]
wsy wrote:

That's why delay lines are equivalent to comb filters in the steady state.

Actually, it's not. A delay line has a flat amplitude response. In order to get a comb filter you need to mix the delayed signal with the original. And such a structure is not a delay line, but a filter.
wsy
mbartkow wrote:
I'm sorry, but you have it all mixed up.

wsy wrote:

Uh, no. In many ways.

First off, delay line does not have linear phase shift.


Pure delay lines do have linear phase response, because phase response by definition is the product of delay and frequency. If delay is constant for all frequencies (e.g. a non-dispersive delay), the phase response is linear.


But you can't hear the difference between 0 degrees and 360 degrees; the trick is that the signal out _always_ undergoes phase unwrapping and so the phase response looks like a sawtooth, not a linear ramp.

And yes, it really matters hugely. Try doing synthetic aperture radar without doing the phase unwrapping. People get patents
and PhDs in that today for how to do it with less loss of information and fewer phase unwrapping errors.

Quote:

But anyways, we are talking about phasers here, which are build from a chain of allpass sections, not delay lines. Analog allpass filters do have nonlinear phase response, and they are certainly dispersive.
See e.g.:
http://www.edn.com/electronics-blogs/living-analog/4375814/All-pass-fi lter-phase-shifter
or any textbook on analog filter design.


Didn't say that filters weren't usually dispersive; I said they were (and that delay lines weren't; "follow the power", as Horowitz
would say. You have to go way out of your way to make a filter that isn't dispersive (i.e. adding a post-allpass with the inverse phase response).

Quote:

Most of your further explanation is based on false asumptions. Eg:
wsy wrote:

As to frequency _shifting_ - no, a filter never shifts energy from one frequency to another.

You are assuming a linear time invariant filter here, which is not the case for phasers that are modulated filters. Modulation is the key to varying group delay of the filter, and this yields frequency shift, quite simply demonstrated by Doppler.


Depends on whether you can store energy in the filter or not - and how much. Heck, you could make a time-invariant 'filter' with a
synchronous motor turning a permanent-magnet-rotor alternator through a gearbox or pulley set and get "power" out at any frequency compared to any other frequency

Then hang a tuning fork or electrical resonance like a T filter in the front, and yeah, you
would technically "have" a frequency-shifting filter.

But I'd call shenanigans on that. smile

Quote:


Oh, and btw
wsy wrote:

That's why delay lines are equivalent to comb filters in the steady state.

Actually, it's not. A delay line has a flat amplitude response. In order to get a comb filter you need to mix the delayed signal with the original. And such a structure is not a delay line, but a filter.


Ah, that's true. Oversimplification; but then again you also have to mix original with
allpass-filtered output to get a phaser. So we're even. :-)

- Bill
mbartkow
wsy wrote:
mbartkow wrote:

Pure delay lines do have linear phase response, because phase response by definition is the product of delay and frequency. If delay is constant for all frequencies (e.g. a non-dispersive delay), the phase response is linear.


But you can't hear the difference between 0 degrees and 360 degrees; the trick is that the signal out _always_ undergoes phase unwrapping and so the phase response looks like a sawtooth, not a linear ramp.

I disagree. Phase unwrapping is a purely computational problem that is encountered mainly in processing of discrete (sampled) data (like in your SAR example). In the analog continuous world, group delay (the physical delay that may be observed on a scope using a transient signal as oposed to a purely theoretical sinusoid) is by definition the derivative of phase over frequency. If phase was not a continuous linear function, the derivative would have spikes and certain spectral components would have an infinite group delay. Nothing of that is observed for a normal delay.

Certainly, you can't hear a difference between 0 and 360 degrees. That is why you get a comb filter with multiple (infinite number of) notches by mixing the original signal with its delayed copy, because the cancellation ocurs at multiple frequencies (as oposed to a single notch you get from two pole analog allpass filter when mixed with the original, which explains why we need a whole chain of such filters to achieve the desirable efect of multiple notches in a phaser effect).

But this (not hearing the difference) does not invalidate the property of linear phase. Every textbook on signal and system theory I know treats a simple delay as a linear phase device. Plese, do not re-invent the whole discipline without a solid reason.


wsy wrote:
mbartkow wrote:

Most of your further explanation is based on false asumptions. Eg:
wsy wrote:

As to frequency _shifting_ - no, a filter never shifts energy from one frequency to another.

You are assuming a linear time invariant filter here, which is not the case for phasers that are modulated filters. Modulation is the key to varying group delay of the filter, and this yields frequency shift, quite simply demonstrated by Doppler.


Depends on whether you can store energy in the filter or not - and how much. Heck, you could make a time-invariant 'filter' with a
synchronous motor turning a permanent-magnet-rotor alternator through a gearbox or pulley set and get "power" out at any frequency compared to any other frequency

Then hang a tuning fork or electrical resonance like a T filter in the front, and yeah, you
would technically "have" a frequency-shifting filter.


I really can't get your examples involving green lasers, synchronous motors or tuning forks. I'm a classically trained electronic engineer, and we should talk about electronic circuits here, because synthesizer modules are electronic circuits.

I can barely imagine an analog electronic filter that would not use capacitors or inductors, and these elements are known from their capabilities of storing energy. Hence, yes all the allpass filters used for constructing a phaser effect store some energy in their internal state. In fact, it would not be possible to physically delay any signal without some form of energy storage. Still, I can't see your point in stating that energy storage is necessary for frequency change.

As I have already mentioned, there is some form of modulation taking place in a phaser driven by the LFO. From the properties of Fourier transform, modulating a signal often results in shifting in frequency domain. It shouldn't be surprising the result is an audible detuning. From my observation, the longer is the chain of allpass stages in a phaser, the more radical is the frequency shift.
wsy
mbartkow wrote:

Certainly, you can't hear a difference between 0 and 360 degrees. That is why you get a comb filter with multiple (infinite number of) notches by mixing the original signal with its delayed copy, because the cancellation ocurs at multiple frequencies (as oposed to a single notch you get from two pole analog allpass filter when mixed with the original, which explains why we need a whole chain of such filters to achieve the desirable efect of multiple notches in a phaser effect).

But this (not hearing the difference) does not invalidate the property of linear phase. Every textbook on signal and system theory I know treats a simple delay as a linear phase device. Plese, do not re-invent the whole discipline without a solid reason.

Quite true- if your listener is a ADC and some software, then you are entirely correct.

But unfortunately humans are not capable of hearing the phase component of a phaser in a monaural signal, and don't percieve
it in the tone domain in a stereo signal - it gets mixed in instead as position information, which is weird because your pan control
on a mix board has a much stronger effect, but some people (among them: me) find that panning without proper time delay and postfilter
to frankly suck because it does NOT achieve sound placement that sounds like real instruments. It's kind of like "sim sickness" that
bothers buddies of mine when they play too much Half-Life.
Quote:

I really can't get your examples involving green lasers, synchronous motors or tuning forks. I'm a classically trained electronic engineer, and we should talk about electronic circuits here, because synthesizer modules are electronic circuits.

I can barely imagine an analog electronic filter that would not use capacitors or inductors, and these elements are known from their capabilities of storing energy. Hence, yes all the allpass filters used for constructing a phaser effect store some energy in their internal state. In fact, it would not be possible to physically delay any signal without some form of energy storage. Still, I can't see your point in stating that energy storage is necessary for frequency change.

As I have already mentioned, there is some form of modulation taking place in a phaser driven by the LFO. From the properties of Fourier transform, modulating a signal often results in shifting in frequency domain. It shouldn't be surprising the result is an audible detuning. From my observation, the longer is the chain of allpass stages in a phaser, the more radical is the frequency shift.


Okay. Let's do the actual real-life experiment.

Here's a W-104 delivering a nice clean sine wave at about 5KHz into an OWON TDS8204 in FFT mode (blackmun window, 7.6K
record length, 50 Ksamples/second, 20 dB/division). The lower red blob is the actual signal at 5 KHz, the upper blue trace is the
Fourier transform, running from DC on the left and with the one big peak at the second division at 5 KHz:



No surprises there. Now, as a "control", run it through a classic Dotcom Q107 state variable filter, taking the lowpass output
(and yes, this filter is really rather old, and needs to final gain upped a bit, varying the cutoff frequency and resonance changes
the amplitude but doesn't add new frequencies:



No surprises; the output is still a clean 5 KHz sine wave.

Finally, the "test", of an Oakley Deep Equinoxe VC Phaser, 8 stages, on the local VCO.
Note that as postulated, it does NOT
add any new frequencies, nor shift the old frequency; the one peak in the FFT output stays exactly where it was, at 5 KHz:



Finally, for the "positive" result, running the sine wave through an STG Wavefolder (basically the "middle end" of a Serge
wavefolder) yields the expected array of new, somewhat-harmonically-related frequencies:



Conclusion: experimentally, neither the "normal" filter nor a phaser add or shift frequencies.

I agree that a time delay with varying clock rate -will- cause frequencies to shift, and that a nonlinear device such as a
wavefolder or Discontinuity or distortion pedal also -will- add frequencies.

EDIT: Looking at the last two plots above, I noticed that the peaks do not _precisely_ align on 5 KHz. So I carefully let things warm up a while more to thermally stabilize, and checked again. The peaks stayed put, right where they should be.

- Bill
mbartkow
wsy

First of all, you are measuring a wrong signal. It's the output of the allpass filter what should be measured, not the result of mixing this output with the original. After mixing, there is a cancellation effect in that every two spectral lines of close frequency (the original and shifted) are equivalent to a single sinusoid multiplied by another one of the differential frequency (the so called beating effect which we perceive as the notch in spectrum moving up and down).

Furthermore, you are using an inapropriate tool for this experiment. The modulation frequency is a few Hz only and this is the amount of frequency shift. The spectral resolution of your analyser is far too low to observe it. At this sample rate you would need at least a 64k FFT to see the difference, and also the screen resolution would have to be in the range of 50k pixels. It would be much more reasonable to use a computer and a hires spectrogram tool with zoom.
mbartkow
I don't have a phaser on my bench at this moment. But I can easily prove the frequency shift in a modulated filter. Below is a spectrogram analysis of the output of a 12db/oct lowpass filter processing a fixed frequency sawtooth signal. The cutoff frequency of this filter is modulated by a sinusoidal LFO with manually controlled rate, starting from slow oscillations (about 1Hz) and increasing up to 500Hz, then decreasing.

The spectrogram resolution was 16384 points (the limit od Adobe Audition) which is barely sufficient to see the effect at 1Hz (the sidebands of each harmonic partial are very clearly changing synchronously with the modulation). At the increased rate of the LFO there is apparent effect of frequency modulation and the spectrum is seamingly consisting of many products of this modulation. If you insist, I can perform the instantaneous frequency analysis in Matlab, to show exactly that the signal has a varying frequency in every harmonic partial.



PS. Chains of allpass filters were often used to create the vibrato effect (pitch modulation) in many electronic and electro-mechanical instruments long before CCD/BBD delay lines were invented. The effect of frequency change is clearly audible there.
wsy
mbartkow wrote:
wsy

First of all, you are measuring a wrong signal. It's the output of the allpass filter what should be measured, not the result of mixing this output with the original. After mixing, there is a cancellation effect in that every two spectral lines of close frequency (the original and shifted) are equivalent to a single sinusoid multiplied by another one of the differential frequency (the so called beating effect which we perceive as the notch in spectrum moving up and down).

Furthermore, you are using an inapropriate tool for this experiment. The modulation frequency is a few Hz only and this is the amount of frequency shift. The spectral resolution of your analyser is far too low to observe it. At this sample rate you would need at least a 64k FFT to see the difference, and also the screen resolution would have to be in the range of 50k pixels. It would be much more reasonable to use a computer and a hires spectrogram tool with zoom.


Well, I disagree... but may I suggest that you do the next round of experiments? You clearly have (or have access to) the proper equipment.

So do it, take photos, and post your results!

(I also _did_ do the experiment with a much longer record length and much faster modulation, which would have shown what
you claim, but all I saw were the normal AM sidebands of the notch moving up and down, same as what you'd get running the
signal into a VCA with the amplitude modulated with a VCO up in the low audio, which is just sum-and-difference as theory suggests,
not new frequency introduction and definitely not the aharmonic series as generated by the wavefolder shown above..

- Bill
mbartkow
I understand my previous example may be not that clear, because the filter cutoff frequency was deeply modulated and the resulting deep amplitude modulation may have masked the effects of frequency modulation.

Here is a second take. Now the signal to the VCO is almost a clean sinusoid, and the modulation depth is very shallow so that it almost does not affect the amplitude. Of course it also means that the FM depth is reduced, but the instantaneous frequency analysis is very capable of detecting even slight modulations.




Now the instantaneous frequency shows clearly the periodic pattern of frequency changing according to the LFO action. The right plot is a zoomed detail from the middle segment, where the LFO frequency was increased.
wsy
Ok, so then, do me a favor and do the exact same experiment but with a VCA in there instead of the phaser.

That way, we can see if the AM of a fast modulated VCA (sum and difference generation) is distinguishable from the passband / stopband motion in a phaser.

- Bill
mbartkow
wsy wrote:
Ok, so then, do me a favor and do the exact same experiment but with a VCA in there instead of the phaser.

That way, we can see if the AM of a fast modulated VCA (sum and difference generation) is distinguishable from the passband / stopband motion in a phaser.

- Bill


Please, note that I didn't do phaser analysis, but a simple 2-pole lowpass filter. As I said, I didn't have a long allpass filter on my bench at that moment.

Here are two examples: a deep amplitude modulation (roughly corresponds to my first take with a deeply modulated filter), and a shallow modulation (corresponds to my second take with the filter taking into account the depth of amplitude variation). Both are obtained from an exponential VCA






What can be observed from the comparison of the "deep" case (1st filter vs 1st VCA) is that in pure amplitude modulation the sidelobes of harmonic partials are purely symmetric, while in the filter output they were clearly balancing from one side of the main lobe to the other side.

What can be observed from the comparison of the "shallow" case (2nd filter example vs 2nd VCA example) is that the harmonic spectrum of the modulation product is much cleaner in the AM case. This is quite easily explained by the modulation theorem: the result of amplitude modulating a sinusoid by another sinusoid (the LFO waveform) is just the sum and difference frequencies. The spectra of FM signals are much more complex (recal FM synthesis). And indeed, in the modulated filter output we observe a whole harmonic series of Bessel spectra.


Once again: modulated allpass filters were commonly used for achieving vibrato (frequency modulation) in early electronic and electro-mechanical organs. The frequency modulation effect was clearly audible. Are you going to deny this fact?
mbartkow
(double post)
wsy
mbartkow wrote:
wsy wrote:
Ok, so then, do me a favor and do the exact same experiment but with a VCA in there instead of the phaser.

That way, we can see if the AM of a fast modulated VCA (sum and difference generation) is distinguishable from the passband / stopband motion in a phaser.

- Bill


Please, note that I didn't do phaser analysis, but a simple 2-pole lowpass filter. As I said, I didn't have a long allpass filter on my bench at that moment.

Here are two examples: a deep amplitude modulation (roughly corresponds to my first take with a deeply modulated filter), and a shallow modulation (corresponds to my second take with the filter taking into account the depth of amplitude variation). Both are obtained from an exponential VCA






What can be observed from the comparison of the "deep" case (1st filter vs 1st VCA) is that in pure amplitude modulation the sidelobes of harmonic partials are purely symmetric, while in the filter output they were clearly balancing from one side of the main lobe to the other side.

What can be observed from the comparison of the "shallow" case (2nd filter example vs 2nd VCA example) is that the harmonic spectrum of the modulation product is much cleaner in the AM case. This is quite easily explained by the modulation theorem: the result of amplitude modulating a sinusoid by another sinusoid (the LFO waveform) is just the sum and difference frequencies. The spectra of FM signals are much more complex (recal FM synthesis). And indeed, in the modulated filter output we observe a whole harmonic series of Bessel spectra.


Once again: modulated allpass filters were commonly used for achieving vibrato (frequency modulation) in early electronic and electro-mechanical organs. The frequency modulation effect was clearly audible. Are you going to deny this fact?


But what I'm seeing in the graphic above (sine wave carrier, sine wave modulator, into a VCA) is NOT what the AM modulation theorem
says should be there. I'm seeing {in the lower graphic} at least two harmonics on the upper and two on the lower in the second
plot, so I'm either hallucinating or it's not sine-on-sine AM (or Adobe is lying to you, which I wouldn't be surprised in the least.)

Now, what happens in a Hammond organ is nonlinearity on top of nonlinearity; I'm not deep into them so I won't say you're wrong
about Hammonds (or maybe that Hammonds are the exception; there's a pair of master oscillators listed there that I have no freakin'
clue how they were used except it was to make chords not "sound crappy").

But we agree that [sine, sine, VCA, 100% depth of modulation] should yield ONE plus and ONE minus harmonic for the second graphic,
so why am I seeing two each above?

Oh, wait... it's an EXPONENTIAL VCA. Okay, all bets are off then. You're taking the sine, exponentiating that (which
generates a whole harmonic series itself) and then using that as the modulator.

By any chance are you using the exponential (1V/octave, possibly labeled "pitch" or "keyboard tracking") input on the VCF as
well? That would explain a lot of things, as would accidentally using a sawtooth for either or both carrier or modulator; the first
plot above would make perfect AM sense if the carrier was a 330 Hz saw and the modulator was a varying-frequency
saw (stepping around but maxing out at about 100 Hz).

- Bill
mbartkow
wsy wrote:

Now, what happens in a Hammond organ is nonlinearity on top of nonlinearity; I'm not deep into them so I won't say you're wrong
about Hammonds (or maybe that Hammonds are the exception; there's a pair of master oscillators listed there that I have no freakin'
clue how they were used except it was to make chords not "sound crappy").

Google up "Hammond vibrato scanner", patents are readily available. It's basically a multitap allpass filter and a mechanical or optical multiplexer that scans the taps.


wsy wrote:
By any chance are you using the exponential (1V/octave, possibly labeled "pitch" or "keyboard tracking") input on the VCF as
well? That would explain a lot of things, as would accidentally using a sawtooth for either or both carrier or modulator; the first
plot above would make perfect AM sense if the carrier was a 330 Hz saw and the modulator was a varying-frequency
saw (stepping around but maxing out at about 100 Hz).

Yes, I was using an exponential control input for the filter, but not 1V/oct. I had to attenuate the LFO in order to reduce the amplitude modulation in my second example. A much better example would be to use a multi-stage allpas filter, because a flat magnitude response would allow to get rid of the amplitude modulation effects and focus on phase modulation. As I said, I don't have one at the bench right now.

My LFO in all experiments was a sinsusoidal, not sawtooth.
wsy
mbartkow wrote:
wsy wrote:

Now, what happens in a Hammond organ is nonlinearity on top of nonlinearity; I'm not deep into them so I won't say you're wrong
about Hammonds (or maybe that Hammonds are the exception; there's a pair of master oscillators listed there that I have no freakin'
clue how they were used except it was to make chords not "sound crappy").

Google up "Hammond vibrato scanner", patents are readily available. It's basically a multitap allpass filter and a mechanical or optical multiplexer that scans the taps.


wsy wrote:
By any chance are you using the exponential (1V/octave, possibly labeled "pitch" or "keyboard tracking") input on the VCF as
well? That would explain a lot of things, as would accidentally using a sawtooth for either or both carrier or modulator; the first
plot above would make perfect AM sense if the carrier was a 330 Hz saw and the modulator was a varying-frequency
saw (stepping around but maxing out at about 100 Hz).

Yes, I was using an exponential control input for the filter, but not 1V/oct. I had to attenuate the LFO in order to reduce the amplitude modulation in my second example. A much better example would be to use a multi-stage allpas filter, because a flat magnitude response would allow to get rid of the amplitude modulation effects and focus on phase modulation. As I said, I don't have one at the bench right now.

My LFO in all experiments was a sinsusoidal, not sawtooth.


Yeah, but it's going in through an exponential input, so it'll end up having a lot of other frequency components as well.

Interesting point - phase modulation (which is definitely happening in an all-pass filter) is equivalent to FMing with the derivative of
the control signal. Of course, if the control signal is constant, then the derivative is zero, which means "no FM, carrier out = carrier in".

Which then explains why phasers have LFOs (maybe going from .001 Hz to 1 Hz) and _not_ VCOs (because the added FM
doesn't sound good? Testing it by running a VCO into my Deep Equinoxe external.... and yeah, it sounds FMey in a bad way,
and not particularly sweet to the ear. Maybe that's because there are too many stages in the Deep Equinoxe (min 4, max eight, I believe)

Kinda like a bandsaw cutting a banjo....

- Bill
mojopin
wsy wrote:

Which then explains why phasers have LFOs (maybe going from .001 Hz to 1 Hz) and _not_ VCOs (because the added FM
doesn't sound good? Testing it by running a VCO into my Deep Equinoxe external.... and yeah, it sounds FMey in a bad way,
and not particularly sweet to the ear. Maybe that's because there are too many stages in the Deep Equinoxe (min 4, max eight, I believe)

Kinda like a bandsaw cutting a banjo....

- Bill


Now I can see the usefulness of the Schippmann phaser. You can select from 2-16 stages. I imagine the 2-stage would sound great with audiorate modulation.
C14ru5
I don't have a Schippmann phaser, but the Doepfer A-101-3 does provide similar options to choose the number of stages, up to 12. I recently did a shootout between all my Doepfer phasers / all pass filters. You should clearly be able to hear the difference in sound between the 3-stage-only A-106-6 and the other phasers which were all recorded in 6-stage mode:

Doepfer phaser shoot-out
mbartkow
C14ru5:
Yes, that was a great comparison, thank you. By listening to the "clean" examples there is definitely a pitch modulation going on, and this is also clearly visible in a spectrogram:



I believe that this definitely proves my earlier point: it is possible to change the frequency of a signal by modulating a filter.

wsy:
There are phasers with voltage controlled LFOs which counts as VCO. The notion of "bad sounding" is a very ambigous thing, especially in the context of modular environment. I very much like the sound of exponential audio rate modulation in a phaser.
MUFF WIGGLER Forum Index -> Eurorack Modules  
Page 1 of 3
Powered by phpBB © phpBB Group