Variable Frequency Notch filters?

[Bill Mountain]

I am thinking my bass could benefit form a notch filter.  I know there are plenty of circuits available for simple active and passive filters but I'm looking for something that will allow me adjust the notch frequency but won't effect the predetermined Q or depth.

I plan to use a bridge-T filter and a stereo pot to adjust the resistors at the same time therefore adjusting the notch frequency but I'm really limited to what notch q and depth I can achieve.

This amp has something similar to what I'm looking for:

http://acousticamplification.com/products/index.cfm?catid=3&subcatid=3&modid=13&product=B800H

But the schematics aren't available so I don't know how the notch control works.

Are there any other circuits that would give me more flexibility in setting notch Q and depth but still allowing me to adjust frequency while leaving q and depth unaffected?

[R.G.]

A state variable filter is what you want.

Two dual opamps, independent Q, gain, and frequency. Actually, you get your choice of high pass, low pass, band pass or notch as well, all at the same time.

Twin T's will do it, but you need some active stuff to bootstrap the filter to a Q higher than about 0.7, and they're not all that easy to tune without changing Q.

[Mike Burgundy]

Both TwinT and gyrator circuits tend to change Q with changing f. they do work quite well though. if you really want, google constant Q. Rane has some nice papers on it online, there's a good project at ESP (http://sound.westhost.com/project75.htm), etc.
I think that Acosutic's notch is comparable to several other manufacturer's "shape" "contour"  or "character" controls (SWR, Hartke) which are also basically mid-notch, "loudness"  controls - rather low-Q.

[Bill Mountain]

Both TwinT and gyrator circuits tend to change Q with changing f. they do work quite well though. if you really want, google constant Q. Rane has some nice papers on it online, there's a good project at ESP (http://sound.westhost.com/project75.htm), etc.
I think that Acosutic's notch is comparable to several other manufacturer's "shape" "contour"  or "character" controls (SWR, Hartke) which are also basically mid-notch, "loudness"  controls - rather low-Q.

I've played the Acoustic amp and it is a 10dB notch you can adjust from 50Hz to 1000Hz.  It works quite well (in the store at least).

[Bill Mountain]

A state variable filter is what you want.

Two dual opamps, independent Q, gain, and frequency. Actually, you get your choice of high pass, low pass, band pass or notch as well, all at the same time.

Twin T's will do it, but you need some active stuff to bootstrap the filter to a Q higher than about 0.7, and they're not all that easy to tune without changing Q.

Thanks RG.  I'll look at State Variable Filters.  I've seen them before but didn't quite understand what they were and how to use them.  Can I use a single pot to adjust the frequency on the fly for different basses and such?

[R.G.]

State variable filters get their name from the fact that they implement in an analog computer fashion the governing equations for a second-order resonance. The analogy that's always used is a pendulum. The pendulum has some mass for inertia, gravity causes a restoring "spring" force and the pivot point has some friction. The length of the string interacts with the mass and gravity to form a simple resonance that always swings at the same frequency no matter what the amplitude of the swing is. This used to be the primary timekeeping principle in the world, once people figured out how to give the pendulum a "kick" every so often to make up for friction losses.

They consist of two cascaded integrators and a network of feedback and gain to take care of the friction and length-of-string parts as well as gravity.

It turns out that the parts of the equation that describe this have outputs at different opamps in the simulation that give high pass, low pass, and band pass outputs. Doing a simple analog subtraction on two of the outputs gives a notch. All of the parameters for gain and Q can be set with a single pot, frequency can be set with a double pot, and the settings don't interact.

You can use a dual-section pot to vary frequency independent of other parameters, and on the fly. Or you can use an LED/dual LDR, or a dual OTA.

One very interesting thing about SVFs is that you can use one LM13700 OTA to implement an SVF with a 1000:1 frequency sweep range if you like. They're popular with synth guys for that.

The Mutron/Neutron envelope filter uses an SVF, as does Anderton's "Super Tone Control".

[tca]

The length of the string interacts with the mass and gravity to form a simple resonance that always swings at the same frequency no matter what the amplitude of the swing is. This used to be the primary timekeeping principle in the world, once people figured out how to give the pendulum a "kick" every so often to make up for friction losses.

Not quite! Only for a small initial angle amplitude where the sin of the angle is almost equal to the angle. For large amplitudes of oscillation the oscillation period depends also on the mass of the pendulum (simple pendulum), besides, length of the string and mean g where it is located at the surface of the earth.

[Blitz Krieg]

LM13700 OTA to implement an SVF with a 1000:1 frequency sweep range

this could be controlled via a single potentiometer?

[DDD]

I oftenly use Twin-T circuits with variable "lower" resistor or with variable resistor in series with the "first" capacitor.
Of course, Q-factor varies with F, but it's not so bad; sometimes it sounds decently or quite good.
Advantage: very simple effective circuit without extra Op-Amps or trannies. Volume loss is quite acceptable.

[R.G.]

Yes, with a single pot setting two currents into the two Iabc pins, which is not terribly difficult. With the 13700, the problem is keeping the width of sweep *down* to a small range of frequencies, only an octave or two.

And yes, pendulum frequencies do change for large swings.  

[Mark Hammer]

What about implementing a standard multiple feedback bandpass filter, and feeding its input to a summing node such that it subtracts from the original full bandwidth signal?  That's the sort of thing whose centre-frequency can be swept via a single variable resistance (although I gather the Q can change a little as one sweeps).



Alternatively, one can use a pair of manually-tuned allpass stages, mixed with clean, to produce a notch; essentially a non-modulated phase-shifter.

[deafbutpicky]

@Mark Hammer
What would be the benefit of using allpass stages?

There's an interesting notch filter idea in the Mountainking Megalith Fuzz, where the hp-filtered
emitter signal is fed back to the collector side, adjustable frequency and notch depth, blendable highs...

[R.G.]

@Mark: Yep, you're right. That's an option and it gets better the smaller the frequency adjustment is. But the Q does vary, although the less change in frequency, the less the Q changes too. The active filter cookbook probably has some equations somewhere relating the sensitivity of Q to that resistor, but I'd have to go dig it out.

[Bill Mountain]

Thanks for all of the ideas guys.

What type of Q or depth should I be looking for?  I want to removed remove clacky/honky/farty frequencies but still sound musical.

Either I want something transparent where I won't notice what is missing or I want something with a pleasant curve that makes a decent scooped sound.  I'm not sure yet.

The impetus for all of this is loving 90% of my clean tone but hating the clunky mids and sometimes huge/annoying lows mids.  I was going to tweak until it sounded good but maybe there is a place I should be starting.

[lietuvis]

Hi, I recently built Yamaha NE-1 parametric eq which is variable mid scoop circuit, so it's some kind notch filter and uses easy obtainable dual A50K potentiometer.
After I built it for my friend I liked it, then used yamaha BB604 preamp circuit with NE parametric implemented and built it in to my bass, so far I do not regret.

[DDD]

Thanks for all of the ideas guys.

What type of Q or depth should I be looking for?  I want to removed remove clacky/honky/farty frequencies but still sound musical.

Either I want something transparent where I won't notice what is missing or I want something with a pleasant curve that makes a decent scooped sound.  I'm not sure yet.

The impetus for all of this is loving 90% of my clean tone but hating the clunky mids and sometimes huge/annoying lows mids.  I was going to tweak until it sounded good but maybe there is a place I should be starting.

Usually minus 10-12 dB is enough, sometimes -6dB is enough, too.

[Bill Mountain]

Hi, I recently built Yamaha NE-1 parametric eq which is variable mid scoop circuit, so it's some kind notch filter and uses easy obtainable dual A50K potentiometer.
After I built it for my friend I liked it, then used yamaha BB604 preamp circuit with NE parametric implemented and built it in to my bass, so far I do not regret.

I think you're on to something.  I was playing on my amp yesterday which has a sweepable midrange.  I set the mid to varying levels of cut and I adjusted the frequency.  I imagine this would be the same as a variable frequency notch filter???

I found some settings I liked but I don't have numbers to back them up so I still want to experiment with my own filters and I will look at parametric midrange designs as well.

I do know that on my SWR amp the q changes with frequency.  Is that a characteristic of all parametric designs?

[R.G.]

I do know that on my SWR amp the q changes with frequency.  Is that a characteristic of all parametric designs?

No. In fact, the term, "parametric", originally implied that the parameters of the filter, like frequency, gain, steepness/order, and Q were independently adjustable.