Playing only the filter

[Ross21]

Hi,

I'm playing around with a vintage Roland SH2 synth and it is playing the filter only. Want to sk about that.

I've built a patch in which all oscilators are muted (OSC1, OSC2 and the sub-osc are muted by the mixer), envelope is all zero'd besides the DECAY which is set to ~4. Setting the VCF RESONANCE to maximum and VCF ENV MODULATION to ~5 plays a very decen 'kick' like tone.
So, obviously only the filter is playing.

Why is that? Is it due to self-oscilation of the filter? Or just it speed of 'opening' and 'closing' which happens in audio rate? What actually creates the tone in this setup ?

[BertKoor]

Yep, upping the filter resonance will make it self-oscillate. Essentially the output feeds back into the input: feedback!

[BlackWinny]

Watch these short yet very pedagogic videos:

https://www.youtube.com/watch?v=1b6-zsKR8a8
https://vimeo.com/157757983
https://www.youtube.com/watch?v=dVgIf71uWB4

And here are some readings :

• Of Responses & Resonance (Sound On Sound - A chapter of "Synth Secrets", the famous Gordon Reid's book)

• The chapter 3 of the Nord Modular Book

• How to control a self-oscillating filter | MusicRadar (using Diva as example)

[No_Use]

Yep, upping the filter resonance will make it self-oscillate. Essentially the output feeds back into the input: feedback!

But where does the initial output come from when the oscillators are turned off as in OP's case?

[vurt]

Yep, upping the filter resonance will make it self-oscillate. Essentially the output feeds back into the input: feedback!

But where does the initial output come from when the oscillators are turned off as in OP's case?

the filters output in to its input.
the "signal" comes from electrons firing round and being filtered and resonated and so on until feeding back.

[Gamma-UT]

But where does the initial output come from when the oscillators are turned off as in OP's case?

Noise. There's generally enough noise in a hardware analogue synth to provide an initial input to the feedback loop.

[Ross21]

Yep, thanks for the clear answers, vids and written material.

Regarding the noise being the source to generate the output sound...
So, I took a VST emulation of a vintage and famous Roland synth, very similar to the SH2 and played with it. I must admit that though it sound quite close in many aspects, it is very NOT CLOSE in imitating the self-oscilation thing. I guess it is not noisy enough...

Say I would like to immitate this sound in a modular VST synth (say) ZEBRA? How could I add both noise and feedback to a filter only patch?

[cthonophonic]

Zebra's filters won't produce any noise on their own, but if you also add a noise module with very low volume (like 0.01), that's enough to get some of the filters to self-oscillate at high resonance. Try different filter types to see which you like best. I find that the MS20 filters self-oscillate the most easily. If you are using Zebra, you will need to be feeding it a note, since the output volume is always controlled by an envelope or gate. If you are using Zebrify, the gate is open by default.

[BlackWinny]

The best self-oscillating filters are among the ladder and 4-pole filters. At least in the true analog circuits.

Because it is in these analog and complexly structured filters that there was (and is again today) the biggest presence of nonlinear events. It is these nonlinear events (due to the heat and to the stochastic movement of the atoms depending the electrical resistivities of the components) which generate an electrical "noise", and the management of feedbacks of physical or electrical noises creates very quickly all kinds of self-oscillations (we know also these phenomena in my job, the biology). This physical/electrical noise, when itself very soft, was the origin of the wonderful "color" given by these analog filters on the sound, particularly in frequencies close to the resonance. And when it is increased the resonance selects some frequencies (a rather narrow range around the cutoff frequency) within the innumerable frequencies of the noise... and enhances them simply by periodical additions until reaching a threshold where begins the self-oscillation. It is from this colored resonance that emerges the self-oscillation when the threshold is reached. The source is really the internal noise, and the self-oscillation is an effect of the periodical addition within the narrow range of frequencies taken in the noise around the cutoff frequency.
That kind of physical effect of self-oscillation is well known in electronics but also in many other scientific disciplines : in biology, in geology, in materials science (everybody knows the case of the glasses exploding as some frequencies), in architecture (the design of suspension bridges for example, where cables can be subject to extremely critical self-oscillations)... And in these domains the "cutoff" frequency is determined by the material which is used (not only its composition but also the pressures and the heat which together affect the crystal structure of the materials) or which is analyzed.

[justin3am]

Some filters are designed to oscillate when triggered with a very short DC offset. This is often called pinging the filter. A good example of this is the Serge VCFQ (Variable Resonance Filter). This filter will not resonate on it's own but if you sends trigger to the filter, it will ring for a period of time determined by the level of resonance. This makes the VCFQ ideal for creating plucked/percussive sounds, as the decay of the ringing takes on a very natural slope.

The filters in u-he's ACE and Bazille can be pinged this way but they will also oscillate without any signal input (they just require a voice/note to be tiggered).

I find that Band Pass filters are ideal for self-oscillation but lots of filters will do it and each with their own character. In multi-mode/state variable filters, the each output of the hi pass output will be 90° out of phase from the low pass output. This can be very useful for generating signals in quadrature.