Notch yer average SVF

[atavacron]

https://www.edn.com/three-op-amp-state-variable-filter-perfects-the-notch/

The author notes that either the V1 or V2 node can be used, but that V2 is lower noise. I’m impressed by the virtual earth being used as an output - never seen that before.

Also there’s this version, I’m not sure how you can change just one pole of a two pole filter and get the same bandwidth, must be particular to the band-stop form. Any chance this could work over a wider frequency range?



^ from Notch Filter Allows Single Pot Adjustment of Notch Frequency With Little Change In Its High Q Value

 

[JohnRoberts]

The author notes that either the V1 or V2 node can be used, but that V2 is lower noise. I’m impressed by the virtual earth being used as an output - never seen that before.

not actually a virtual earth, just a - input, buffered by an op amp.

JR

View attachment 102398

^ from Notch Filter Allows Single Pot Adjustment of Notch Frequency With Little Change In Its High Q Value

 

[Bo Deadly]

I don't think there's anything super extraordinary going on here. The Rf resistors on the integrators do not have to be perfectly matched. So it seems to me they're just using a limited adjustment range to allow for one pot.

SVF circuits are brilliant and provide a lot of functionality and precision but I think I would usually prefer the slightly lower noise of a simpler circuit.

 

[atavacron]

it seems to me they're just using a limited adjustment range to allow for one pot.

Got it.

 

[atavacron]

What I like about this circuit is that it makes the notch a wee bit more practical to integrate into a design. Noise gain of less than 2, as opposed to 3 minimum with the typical HP+LP inverting sum. No additional load on any of the amplifiers.

Now if I could only grok how to best integrate this into a Net EQ style parallel-filters design, without taking the SVF entirely out of the mixer section and placing it in series on the way to the output. Best I've come up with is: 1) Buffer and amplify the summing node non-inverting, as drawn, 2) Drop that into the inverting input of an equal-value diff amp, while bringing the input signal into the non-inverting input, and 3) bring that into the Cut side of the cut/boost network at unity with the input signal. In other words, subtract everything but the notch from the original signal, then subtract the resultant sharp peak from the original signal at the Cut amp.

 

[JohnRoberts]

Speaking of notch filters I designed a monitor mixer/console late last century and put sweepable notch filters into each of the the sub masters. I based the notch filters on a SVF topology and incorporated a variant on FLS (the LED over each slider in a GEC that shows where feedback is happening). In these notch filters some goldilocks LEDs indicated if the notch was tuned too high or too low by comparing the HP and LP output for which was stronger to assist in tuning the notch.

That mixer also used another FLS variant where a single LED in each input strip would light only for the loudest input channel (presumable the input that was feeding back.).

JR

PS; It may have been a bit too cool for the room (peavey distribution)...

 

[atavacron]

PS; It may have been a bit too cool for the room (peavey distribution)...

That's pretty next level man. That visual feedback must have been a godsend for people running mons on loud stages. Which mixer? I'll find it.

 

[JohnRoberts]

IIRC it was called SRM2410, built into a light weight flight case with universal switching power supply.

JR

 

[atavacron]

Looks something like 1/6 octave

 

[JohnRoberts]

That was decades ago, so at least I wasn't imagining it. The "goldilocks" tuning LEDs were cute (IMO). I did something like that in my electronic drum tuner with LEDs indicating what direction to turn the lugs for optimal clear...
==
The notch filters were not made super narrow because the ambient conditions inside live performance spaces can change between empty and full of audience members. Using too narrow notch filters might lose capture of the feedback nodes after heat and humidity causes them to drift in pitch.

JR

 

[abbey road d enfer]

Now if I could only grok how to best integrate this into a Net EQ style parallel-filters design,

In an additive EQ, the only way to make a notch is to flip the polarity of a bandpass filter.

 

[atavacron]

In an additive EQ, the only way to make a notch is to flip the polarity of a bandpass filter.

That’s a heck of a lot easier than making the notch, then summing the polarity-flipped notch in with the original signal. After looking at some four-amp SVFs with notch outputs, I’m still mystified: What is the mechanism by which summing the BP (-) and Input (+) signals results in a notch, rather than a shape that looks like an upside down BP?

 

[JohnRoberts]

If you subtract the bandpass from the wideband input signal that bandpass creates a hole in the audio due to missing signal. To look like a notch filter the BP needs to be high Q narrow bandwidth.

JR

 

[atavacron]

D & JR, I appreciate your replies. Maybe this will get at what I’m trying to figure out:
Result #1 is a more conventional version of what I described in post #5. I’m thinking that’s a valid signal to drop into the cut or boost network the same way you would a BP signal.

Result #2 is what I initially assumed D described in post #11, but it would just boost BP in cut and cut BP in boost.

Result #3 is i think what JR described in post #13 (and probably is what D meant), but it would also boost BP in cut and cut BP in boost.

Maybe there’s something I’m missing? Or is the message simply, “Make a high Q band pass and make it cut-only”?

 

[abbey road d enfer]

See how it's done in the dbx905.
Note how it disables the boost/cut pot and uses a different path for the BP signal.

 

[atavacron]

See how it's done in the dbx905.
Note how it disables the boost/cut pot and uses a different path for the BP signal.

Bandpass, sans phase correction, unity gain into the boost amp. Check. Thanks much.

I’ll leave the manual here for people in the future - the schematic is easier when you reference the frontpanel and curves.

 

[abbey road d enfer]

Better if you post it in Technical Documents