[sdiy] Different way of controlling resonance in a State Variable Filter

[Richie Burnett]

I've used this technique with digital state variable filters before, Tom, but not specifically with the aim of controlling the resonance. For instance, I've halved the gain of one integrator and doubled the gain of the other in order to either prevent (or intentionally provoke!) clipping in one of the integrators. Then corrected the Q factor by adjusting the q feedback from after the first integrator.

As it shows in that paper, you can fix the q feedback level from between the integrators and then manipulate the svf's Q factor by inversely scaling the two integrator gains but there are a few issues:

1. The total available f and Q control range is not very wide as MR pointed out, because the dynamic range required for the frequency control adds to that required for the Q control, potentially challenging the gain range of the two VCAs.

2. As the integrator gains become increasingly mismatched, then the BP and LP output levels become more mismatched. This might be an issue if you want the user to be able to switch responses on the fly.

3. Mismatched integrator gains can mean that one integrator will be working with a large signal (risking clipping) whilst the other is working with a tiny signal (potentially compromising SNR).

You can easily see why making the integrator gains mismatched alters the Q by looking at the transfer function of the SVF. The two integrator gains multiplied together give you the s^2 term in the denominator that is square-rooted to determine the resonant frequency of a second order filter. And the feedback from between the two integrators results in the s term that determines the damping (inverse of Q.)

In a sense there is a bit of this behaviour going on in every real analogue state variable filter because of the slight mismatch in the integrator capacitor values and the tolerance variation in the OTA transconductance values.

-Richie,

Sent from my Xperia SP on O2

---- Tom Wiltshire wrote ----

>Hi All,
>
>I came across the following interesting ap note form THAT Corp:
>
>	http://www.thatcorp.com/datashts/AES13-031_Digitally_Controllable_Audio_Filters.pdf
>
>Despite the title, it's mostly VCA applications (as might be expected) so it could just as well have been called "Voltage Controllable Audio Filters".
>
>Anyway, there's a particularly interesting circuit in Figure 10, at the bottom of the page numbered 201.
>
>I've usually seen SVFs with a separate VCA to control the resonance, sometimes in the path from the LP output (Roland System 100M, I think) but more commonly in the path from the BP output. This one dispenses with that and uses the Resonance CV to push the two integrators frequencies apart. This reduces the resonance in the circuit, I'm guessing.
>
> Seems like a pretty good idea for various reasons, not least of which is that it saves a VCA (or swaps a VCA for an op-amp anyway). I'd be interested to know how the circuit affects the resonant behaviour of an SVF and if this method of control makes it more stable across the range (athough I don't see on the face of it why it should). If you had two DAC channels controlling the VCAs, you could do the Freq CV/Res CV maths in the digital domain and just output the voltages you need, saving even more parts.
>
>Has anyone tried this? Has anyone seen it used anywhere?
>
>Thanks,
>Tom
>
>
>
>_______________________________________________
>Synth-diy mailing list
>Synth-diy at synth-diy.org
>http://synth-diy.org/mailman/listinfo/synth-diy