THAT DN130: first order state variable filter

[emrr]

I stumbled on this awhile back, and can't say I've seen it implemented in any product I've personally encountered. Thought I'd throw it out for any discussion it inspires.

http://www.thatcorp.com/datashts/dn130.pdf

I don't think filter design enough to quickly grasp the various ways it might be modified. For example, would it be reasonable to run series implementations with parallel voltage control for 2nd order results, etc etc?

In what ways is this an elegant solution in the eyes of the experienced filter designer, and in what ways is it not?

[JR.]

An obvious benefit is voltage control of one pole filter, while these days I expect digital control to be more cost effective.

Back in the late '80s I designed a single ended noise reduction for AMR/Peavey (called SENR2 or something like that) using dbx/THAT VCAs. I combined a downward expander, with a sliding LPF, to retroactively clean up noisy audio tracks. By combining the downward expander with the sliding LPF you could use less of each, to realize a bunch of noise reduction with less audible deterioration than using either one alone .

As i recall I used a different topology than that app note to mitigate the less than wonderful VCA noise floor from SOTA VCAs 25-30 years ago. Instead of running the primary audio path through the VCA(s) I used a subtractive approach so the VCA created an opposite polarity signal that could be subtracted from the dry path. This delivered lower overall path noise with excellent transient response since the primary output signal didn't pass thought the VCAs.

I don't keep schematics of my old designs but IIRC the downward expander was inverted polarity and fed through a high resistance subtraction resistor for low noise contribution. The sliding LPF was an inverted version of the signal summed back in through a small cap. When the VCA was low gain the cap had no opposite signal swing so looked like a passive real pole at >20kHz. As the voltage gain on the cap increased, the effective pole frequency slid down into the passbacd. The only VCA noise was either cap coupled through the small cap, or large resistor back into the dry path. The subtractive approach traded a limited amount of attenuation or LPF in exchange for lower noise/distortion.

JR

PS: Note: I was not original with the concept of combining downward expander and sliding LPF, but IMO my execution was quite good. The noise floor of a studio noise reduction unit clearly matters. Wayne may have some observations about the Micmix dynafex who pioneered the concept AFAIK.

[JR.]

Yes, you could stack two in series to control the frequency for a SVF parametric.

On nice trick that you should be able to do, it offset the two pole frequencies to adjust the SVF bandpass filter Q.

Perhaps of interest for voltage control...

JR

[mediatechnology]

I did an 8 pole All-Pass Phaser with VCAs and it rocked: viewtopic.php?f=6&t=368&hilit=phaser&start=31

[emrr]

Ah right, the phaser sounded cool. Lost in the memory banks, here.

Thank you both.

[mediatechnology]

Doug -

Somewhere in the back of my mind I recall SSL doing a stereo programmable bus equalizer that was under automation control. It was the SSL 683...
I recall it having dBx 2150 VCAs in it and being a state-variable parametric design.
I might have drawings of it.

Theres' one on eBay right now for $4700.


SSL 683 Programmable Equalizer


JR's noise reduction sounds very much like the DynaFex that Micmix, SSM and CRL partnered on.

[emrr]

I like antiques, but that looks like one to avoid. : ) I suppose someone must be making a modernized variation.

Getting a better grasp here, thanks.

[mediatechnology]

When I was commissioning SSLs I was never very happy to see one of those even in their day.
I think GearSlutz has some threads on the SSL683.
I do think that for analog EQ, given the cost of high quality multi-section pots and switches, substituting VCAs or DPOTs seems to make sense.

As JR has been quick to point out both are more expensive than DSP.
I've been thinking we should open a DSP/microcontroller sub-forum here.
I'd also like to find someone who can write a plug-in.
Thoughts?

[emrr]

I agree it seems a luxury exercise to make an involved EQ with VCA's. Real world need diminishes by the day given all the affordable and decent DSP that's available. I see more desirable use in efx processors and in simpler filters used in the input chain before AD conversion. I also see the mechanical switch market shrinking dramatically over the last few years, so new EQ design with a pile of complex mechanicals seems like potential suicide from a repair shop POV. If I really needed it in analog domain, the VCA approach seems maybe smart.

[mediatechnology]

I think using VCAs for filter control make a lot of sense in stomp box effects and voltage-controlled synth applications.
Anywhere you want an "organic" sound controlled by knobs.
Or where there's a requirement for zero latency...
I noticed that Yamaha has just introduced a hybrid analog path/digitally-controlled mixer for under $2K. http://usa.yamaha.com/products/live_sou ... og-mixers/
You have to wonder how they're controlling the EQ or if they're simply using pots...