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

Low frequencies when using a filter as an oscillator
MUFF WIGGLER Forum Index -> Modular Synth General Discussion  
Author Low frequencies when using a filter as an oscillator
krell
I've noticed that when using a filter as an oscillator, some filters are capable of going lower in tune than others. What determines this, and can this be changed? (either through hardware modification or clever patching)?
EATyourGUITAR
It depends on the filter. Some of them are AC coupled everywhere to block DC. Some of them are voltage controlled but really they are current controlled on the inside. All of the components together set the range of operation. Some filters will never work on DC. It depends on what you consider a usable low frequency. Is it 10Hz, 1Hz, 0.001Hz? maybe your capacitors would be $10 each in polyfilm.
R.U.Nuts
I think the OP means that some VCFs won't self oscillate at lower audio frequencies... But I have no idea why, too...
krell
EATyourGUITAR wrote:
It depends on the filter. Some of them are AC coupled everywhere to block DC. Some of them are voltage controlled but really they are current controlled on the inside. All of the components together set the range of operation. Some filters will never work on DC. It depends on what you consider a usable low frequency. Is it 10Hz, 1Hz, 0.001Hz? maybe your capacitors would be $10 each in polyfilm.


Oh interesting, I didn't know it was that involved, but I guess it doesn't surprise me either. Basically I'm using a filter as the sine source for a 909-style kick and I wish it'd go a bit lower (the Manhattan Analog MA35). It does go quite low, but I'd like just a tad lower (I am attenuating the envelope).

R.U.Nuts wrote:
I think the OP means that some VCFs won't self oscillate at lower audio frequencies... But I have no idea why, too...


Maybe, in practice it does seem as though different filters oscillate at different resonance levels, and that this impacts how low frequency it will go when tracked. But that might be an erroneous impression since I'm comparing such different filters?
EATyourGUITAR
some filters have not enough resonance to self oscillate. some filters self oscillate with a gain of 1. some other filters self oscillate at different gains. the ability to self oscillate when the resonance is increased is not affecting nor is it affected by the lowest frequency that the filter will pass. the ability of a filter to self oscillate depends on the filter type and also if the designer has removed the ability of the filter to self oscillate as an intentional design choice. there is no practical value in selling someone a home stereo that sounds like a devilfish TB-303 when some adjustments are made to the equalizer. likewise the lowest frequency that a filter will pass or boost is the decision of the designer. sometimes the designer is just reproducing an old circuit with little understanding of how it works. in those cases, the designer will carry over the resistor and capacitor values completely unchanged from the previous filter designer. this means that the corner frequency, resonant frequency, bandwidth etc.. are all the same as in the previous design. occasionally components will go up or down slightly in value to conform to the cheapest option for mass produced products. as a result, the lowest resonant frequency will also be changed slightly as a side affect.
mskala
In order to oscillate, the gain around the feedback loop has to be exactly 1. An oscillating filter normally is designed for gain slightly greater than 1 on low-level signals, and then when the oscillations get strong enough they will drive it into distortion, which has the effect of reducing the gain, so it will balance out and oscillate stably at some amplitude level. To get a purer sine wave output, the filter will be designed for gain as small as possible while still exceeding 1, which means it will need less distortion to bring the gain down to exactly 1. Special-purpose oscillators designed for very pure sine output will have the gain very carefully adjusted.

But the gain around the filter can depend on a lot of things - like temperature, component tolerances, and frequency. As the filter gets further away from the centre of its operating range, those effects usually tend to reduce the gain; until at some point it's no longer able to oscillate. As a result, there's a limited range of frequencies over which it can oscillate.

My own filter designs usually involve OTAs as the frequency-controlling elements, and plastic-film capacitors (the inductors in the Coiler are not so relevant at the lowest frequencies) and I think the effects limiting the lowest oscillation frequencies come from those things. There's a limit to how accurately OTAs can track at very low frequency levels; stuff can happen like different OTAs ending up with different time constants when they're all supposed to be the same, so that the filter no longer has exactly the right response curve, and its overall gain is reduced. Then the film capacitors, which are preferred for audio frequencies, can have effects like "soakage" which cause their response to be non-ideal at very low frequencies (more loss, wrong overall effective capacitance); that can cause gain loss as well. And stuff like op amp offsets can affect the maximum amplitude available at different points in the circuit, which in turn changes the way the distortion will interact with the gain. The result is that the safety margin in the filter's small-signal gain, which needs to be slightly more than 1 for oscillation to be reliable, gets eroded. At low enough frequencies the small-signal gain actually dips below 1 and then the filter won't oscillate.

My Coiler and Leapfrog VCFs were both designed with a goal of oscillating down to at least the lower end of audio - about 20 or 30Hz, and nearly all individual units built to those designs really will do that. Most will go a little further; a few individual units can oscillate well into the sub-1Hz range. But I've found a fair bit of variation in exactly how low individual units can go. Some can oscillate deep into LFO range and others fade out just below 20Hz. That's down to individual tolerance variations in the components, and with the Coiler in particular, it's expected that there will be a lot of individual variation because that's part of the design concept.

In general, I would not expect a synth VCF (mine or anyone else's) to necessarily be able to oscillate at very low frequencies unless it's specifically meant to function as an LFO too - like the Intellijel Dr. Octature, which is labelled as a "VCF/VCO/LFO." Synth VCFs in general are usually only really designed to operate normally at audio frequencies. Someone who wanted to design for a wide oscillation range would probably do it by making sure the maximum gain around the loop can be adjusted to significantly more than 1, to make sure it will stay above 1 even when reduced by these other effects. But that is a trade-off against distortion and the purity of the signal, because the higher small-signal gain implies more distortion needed to bring it down to 1 at less extreme frequencies.
krell
mskala wrote:
In order to oscillate, the gain around the feedback loop has to be exactly 1. An oscillating filter normally is designed for gain slightly greater than 1 on low-level signals, and then when the oscillations get strong enough they will drive it into distortion, which has the effect of reducing the gain, so it will balance out and oscillate stably at some amplitude level. To get a purer sine wave output, the filter will be designed for gain as small as possible while still exceeding 1, which means it will need less distortion to bring the gain down to exactly 1. Special-purpose oscillators designed for very pure sine output will have the gain very carefully adjusted.

But the gain around the filter can depend on a lot of things - like temperature, component tolerances, and frequency. As the filter gets further away from the centre of its operating range, those effects usually tend to reduce the gain; until at some point it's no longer able to oscillate. As a result, there's a limited range of frequencies over which it can oscillate.

My own filter designs usually involve OTAs as the frequency-controlling elements, and plastic-film capacitors (the inductors in the Coiler are not so relevant at the lowest frequencies) and I think the effects limiting the lowest oscillation frequencies come from those things. There's a limit to how accurately OTAs can track at very low gain levels; stuff can happen like different OTAs ending up with different gains when they're all supposed to be the same, so that the filter no longer has exactly the right response curve, and its overall gain is reduced. Then the film capacitors, which are preferred for audio frequencies, can have effects like "soakage" which cause their response to be non-ideal at very low frequencies (more loss, wrong overall effective capacitance); that can cause gain loss as well. And stuff like op amp offsets can affect the maximum amplitude available at different points in the circuit, which in turn changes the way the distortion will interact with the gain. The result is that the safety margin in the filter's small-signal gain, which needs to be slightly more than 1 for oscillation to be reliable, gets eroded. At low enough frequencies the small-signal gain actually dips below 1 and then the filter won't oscillate.

My Coiler and Leapfrog VCFs were both designed with a goal of oscillating down to at least the lower end of audio - about 20 or 30Hz, and nearly all individual units built to those designs really will do that. Most will go a little further; a few individual units can oscillate well into the sub-1Hz range. But I've found a fair bit of variation in exactly how low individual units can go. Some can oscillate deep into LFO range and others fade out just below 20Hz. That's down to individual tolerance variations in the components, and with the Coiler in particular, it's expected that there will be a lot of individual variation because that's part of the design concept.

In general, I would not expect a synth VCF (mine or anyone else's) to necessarily be able to oscillate at very low frequencies unless it's specifically meant to function as an LFO too - like the Intellijel Dr. Octature, which is labelled as a "VCF/VCO/LFO." Synth VCFs in general are usually only really designed to operate normally at audio frequencies. Someone who wanted to design for a wide oscillation range would probably do it by making sure the maximum gain around the loop can be adjusted to significantly more than 1, to make sure it will stay above 1 even when reduced by these other effects. But that is a trade-off against distortion and the purity of the signal, because the higher small-signal gain implies more distortion needed to bring it down to 1 at less extreme frequencies.


Thanks, that was a great explanation!
The Grump
My Mopho's filter does it well, Evolver... not quite as well. My MW1 is fantastic for this, and when I had a Model D, it would stop self oscillating before hitting the lower frequencies. I'm assuming we're referring to analog and not digital filters?
krell
The Grump wrote:
My Mopho's filter does it well, Evolver... not quite as well. My MW1 is fantastic for this, and when I had a Model D, it would stop self oscillating before hitting the lower frequencies. I'm assuming we're referring to analog and not digital filters?


Yes, analog filters
MUFF WIGGLER Forum Index -> Modular Synth General Discussion  
Page 1 of 1
Powered by phpBB © phpBB Group