Envelope controlled filter question

[Kipper4]

Got it.

http://www.idea2ic.com/LM13600/UsingOTAs2.pdf

[Fancy Lime]

@ Mark

We're glad you found it, and glad we found you, too.

The pleasure is all mine, I feel very welcome. This is certainly one of the nicest and most collaborative forums I know. Strange how forums have their own dynamics. Here I feel like I can ask the stupidest questions (although I try not to but that's not the point here) and would still get a polite, helpful and sincere answer. There are other forums where that sort of thing leads to, uhm, distinctly not polite "conversation". Good to be here.

@ Kipper4
Thanks! The article looks very helpful, although I have not read the whole thing yet. R.G.'s Auto-Wah article on Geofex is where I started. I learned a long time ago to first check the usual suspects for "technology of the ..."-type article before diving into a new topic. As I would recommend to anybody. Had that same discussion with R.G. a few days ago on a polarity protection thread.

Back to the topic: I've been thinking about the Attack control problem with the voltage divider a bit. Where did I read about that same problem in a different context? Then it hit me like a ... Hammer. Sorry for the pun. A certain Mark Hammer posted a little piece of circuitry that solves exactly that problem and called it the Stupidly Wonderful Tone Control. By doing the same thing with the envelope network, replacing Rs with a pot (say 500Ω) and dangling C from the wiper should at least lessen the effect that the Attack setting has on the Sensitivity while also keeping the current limiting for the filter control unit constant. has anyone tried this? If not, I will. If it works, the corresponding knob shall be labeled "SWAC".

Progress report on the filter section research: Seems to get lowpass, bandpass and bandstop characteristics, a state variable filter topology by the name of Kerwin-Huelsman-Newcomb (KHN) Biquad Filter is what we want, which also gets us a highpass output thrown in for free. Thats what the Mutron uses, implemented with opamps. I have not been able to find that particular design realized with OTAs. All actual designs I found (like the DOD FX-25 and Mad professor Snow White) all seem to use a different topology. Specifically the one that is mentioned in the application notes of many OTAs. But that only has band- and lowpass outputs, so no way of getting a notchfilter. It should be fairly easy to convert the Mutron filter for use of an OTA by simply replacing the two inverting integrators with respective inverting integrator OTA stages. That must have been done before but I can't find it, so I'll draw up the schematic and try if it works as soon as I find  quiet minute. Maybe a "non-optical Mutron" schematic is something some people might like.

Cheers,
Andy

[Transmogrifox]

Here's an ASCII schematic for Attack and Release control:
                                                     POT
                                  1N914       ATTACK
[OP AMP OUTPUT]-x------>|----------/\/\/\/--------x
                            |------|<----------/\/\/\/--------|
                                   1N914       RELEASE         |
                                                     POT               |+
                                                                       -----  Capacitor
                                                                       -----
                                                                          |
                                                                          |
                                                                        GND
The forward diode forces the resistance value on the attack pot to be used when the output voltage is higher than the capacitor voltage, setting its charge time.

When the op amp output gets lower than the capacitor voltage then it must discharge through the release pot, which allows you to set the discharge time.

This is a bit simplified because you would usually want a current limiting resistor in the op amp output to keep things happy when one of the pots is set to 0 resistance.

This circuit also assumes the voltage coming from the op amp output is already rectified and reasonably filtered (to a minimum attack/release time).  A peak detector with minimum release time would be appropriate pre-conditioning.

In either case this solves the voltage divider problem.

[Fancy Lime]

@Transmogrifox
Hmm, interesting idea. Certainly worth a try.


As for the idea to use a Stupidly Wonderful Tone Control like arrangement for the Attack: OF COURSE someone had thought of that already. The Lovetone Metball seems to do exactly that, according to some cryptic forum descriptions, although I could not find a schematic of it. From what I understand the Meatball is pretty much a Mutron that had the living bejeezus modded out of it.

[Transmogrifox]

From what I understand the Meatball is pretty much a Mutron that had the living bejeezus modded out of it.

In other words the Meatball uses a state variable filter...and that's where the similarity to the Mutron ends.

[Fancy Lime]

In other words the Meatball uses a state variable filter...and that's where the similarity to the Mutron ends.

It also uses a precision rectifier for envelope tracking and optocouplers to control the SVF. I found the schematic in the meantime. So the concept is identical with that of the Mutron. One could even mod an existing Mutron to be functionally almost identical to the Meatball by adding a bunch of switches to switch capacitors and connections and some pots to replace resistors. Don't get me wrong, I'm not trying to downplay the design of the Meatball or anything. It seems to be a great sounding very versatile envelope filter from what I can tell (only watched some demos, never had a chance to play it myself) and that's what design is about (as opposed to invention). Adding knobs and switches in all the obvious places is one thing but to make the whole thing and the additional settings sound good and have useful ranges is what distinguishes a good design from an unusable Mutation by Blind Urge to Mod (BUM). Just saying, from a concept perspective, it is a descendant of the Mutron, isn't it? One could say it is a member of the wider Mutron family. May also be convergent evolution, of course, although that seems a bit unlikely considering how well known the Mutron is.

Cheers,
Andy

[Mark Hammer]

The number of descendants of swept SVFs is substantial, so I wouldn't take anything away from the Meatball.  And if you look up the Mu-tron III on the effectsdatabase, you'll see that the "family tree" is quite extensive.  One of my first 3 pedal purchases, in 1977 or so, was a Fernandes Funky Filter ($25, new) that was a repackaged and rebranded Mu-Tron.  That said, the Meatball is a hyper-modded Mu-tron III....which is not a bad thing.

If you look in Craig Anderton's Electronic Projects for Musicians book from the early 80's, it also shows a project to adapt a state-variable filter to envelope control.  The Ibanez AF9 Autofilter is also a Mu-Tron clone.

[Transmogrifox]

The SVF envelope controlled filters are all in the same family of effects in the sense that the designs use a similar principle.

It is unclear to me whether the Mutron was the reference design and these others are variants, or if it's just because the SVF is a well-understood (well-studied) filter topology it stands on its own as the origin for this type of design.

In other words, designers could independently come to a very similar circuit topology without ever having reverse-engineered the competition.

Case in point : When I was in college learning about the biquad filter designs I was envisioning an envelope-controlled filter without ever having known there was ever such a thing as a "Mutron".  At the time I didn't know much of anything about guitar effects except that I had a Digitech RP-20 Valve that made some really cool sounds, and also a few distortion pedals.  The rest was pure invention as far as I was concerned -- only to find out the same things were invented before I was even born.  Ignorance really is bliss.  It was a nice feeling to think I was really creating some new things before I discovered my schematics would have been considered a (poorly designed) knock-off of some well-known guitar effect.

If I were to make a guess I would tend to think there is an analog synthesizer circuit that might be credited as the pioneer in using a state-variable topology in a VCF...from there it is a natural extension of the concept to apply the same type of filter to other types of musical instruments.  More likely than not, the guitar-oriented variants would be a stripped-down implementation of a more full-featured implementation (synth players like LOTS of knobs).  Several makers in the pool of competition could have easily come to the same conclusions independently.

I'm sorry if I have belabored this point beyond what is due.  I got to going and couldn't stop myself

The main point I think we all agree upon is that the Mutron designs and many others share a lot in common so you can start with about any one of them as a reference design and end up in the same place by "modding the bejeezus out of it".

[PRR]

> the same things were invented before I was even born.

"The ancients are stealing our inventions!"

If you live long enough, you will realize that there is very little new in audio.

The State Variable was studied in the 1940s or before. At that time it was expensive; practical analog computers used simpler less-universal filter schemes. $50 op-amps were a revolution, and $2 op-amps the sky exploded with SVF ideas.

Many old ideas are "ahead of their time"; you can do that, but why would you want to?? The concepts and techniques of musical sound-bending lagged behind filter theory.

[ElectricDruid]

Progress report on the filter section research: Seems to get lowpass, bandpass and bandstop characteristics, a state variable filter topology by the name of Kerwin-Huelsman-Newcomb (KHN) Biquad Filter is what we want, which also gets us a highpass output thrown in for free. Thats what the Mutron uses, implemented with opamps. I have not been able to find that particular design realized with OTAs. All actual designs I found (like the DOD FX-25 and Mad professor Snow White) all seem to use a different topology. Specifically the one that is mentioned in the application notes of many OTAs. But that only has band- and lowpass outputs, so no way of getting a notchfilter. It should be fairly easy to convert the Mutron filter for use of an OTA by simply replacing the two inverting integrators with respective inverting integrator OTA stages. That must have been done before but I can't find it, so I'll draw up the schematic and try if it works as soon as I find  quiet minute. Maybe a "non-optical Mutron" schematic is something some people might like.

Often the "data sheet" version of the filter uses the two inputs of the OTA to provide the inverting action that is needed for some of the feedback paths. This saves an op-amp, but means you lose the highness output (and therefore the possibility of a notch). However, you can always add it back in. If you look at the ESP pages SVF layout, the differential mixer stage can be added ahead of the two integrators, and then you get a highness output.

It's just tempting, if you're using OTAs, to use the +/-ve inputs and save the op-amp. Especially if you're a 1980's pedal designer with your boss breathing down your neck and muttering about costs!

HTH,
Tom

[Fancy Lime]

Alright, I adapted the Mu-trons SVF for use with a LM13700 OTA.



I'll try that with manual frequency control first before hooking it up to an envelop follower.

If someone spots errors, please let me know. the buffers are simplified in the schematic, in real life they need a 5.1k or something to V-, but I wanted to keep the schematic readable.

Cheers,
Andy

[ElectricDruid]

I was messing with a filter very much like this the other day. It worked fine, except I was getting terrible thumping at the low end (control feedthrough?). It was ok if you pushed the filter frequency higher up into the upper audio range, but then you don't get that nice low-end "quack" that you want from these type of things.
Does anyone have any experience to share with controlling control feedthrough with these LM13700 filters?

Thanks,
Tom

[Fancy Lime]

Could that have to do with AC in the control voltage? Have you played with the values of the resistors that go into the control terminals? I was wondering why it is that all practical designs seem to have one resistor before each control terminal and after they split from the control voltage source (thus separating the control terminals), whereas the application notes all have one resistor before the split (thus having the control terminals directly connected). Which way did you go here? Have you tried the respective other? I would imagine that allowing the control terminals to talk to each other may cause unwanted side effects in a real-world circuit.

Andy

[Mark Hammer]

Swept filters generally become most interesting and attractive when there is some resonance added.  If the sweep starts low, especially for a lowpass filter, the additional gain provided by that resonance can make the bass particularly obnoxious.  FL is correct in that starting the sweep up higher avoids the thump, but loses the appeal.  I adapted the DOD FX-25 for a bass-player friend, so that it was lowpass, rather than the stock bandpass.  It also had an annoying thump at the bottom of the sweep.  I wonder if there was some way to have the resonance co-vary with the sweep.  That is, at the low end of a lowpass sweep, the resonance is reduced, only increasing as the corner frequency does.  In some respects, that would be a job for a SVF that has both voltage controlled corner/center frequency AND resonance.

[Fancy Lime]

Hmm, interesting thought. The resonance pot on the FX-25 and its cousins controls the negative feedback of the bandpass stage. So by replacing this negative feedback, which is allpass as things are now, with a lowpass negative feedback in such a way that the resonance pot only affects the high frequencies but lows are always passed through to the inverting input, should have the desired effect. Might be worth a try. An appropriately sized cap between the 22k feedback resistor and the resonance pot might be all it takes, if I'm not mistaken. Maybe a resistor in series with the cap, too.

Andy

[Fancy Lime]

No, now I see it. What I just proposed won't work. One would need to replace the resonance pot with a fixed resistor (10k or so) and put the resonance pot (100k log) from the new cap to ground.

[Fancy Lime]

There is another thing I was wondering about in the FX-25:

It uses a half-wave rectifier to fill a 22µ cap (which goes to ground) with the positive half wave. So one would assume that with no signal, the cap is empty and the control line to the OTA should sit close to ground. Yet the cap does not have a resistor parallel to ground that would do that but has a 1M going to V+. Is that correct? I've been wrecking my brain but I cannot figure out how this would possibly work. Seems to me that with no signal the control line (as per the schematic) sits at V+ and when there is signal it goes more positive. That can't be right. I must be looking at this the wrong way. Can someone explain?

I'm trying to modify the envelope follower such that it can be switched from up to down mode, which should be possible by turning around the diodes and adjusting the idle voltage at the 22µ cap. But for that I should probably understand how the whole thing works. If that 1M resistor went to ground I would think I understand but as it is...

Thanks,
Andy

[Transmogrifox]

The OTA control inputs sink current, which discharges the cap through the respective 10k resistors. 

The 1M resistor sets the minimum cut-off frequency.  If it weren't there the filter would go almost silent on low signal levels.

In the end that node with the 22uF cap probable sits at around 0.6V when there is no input signal.

[Fancy Lime]

Thanks! I see, that complicates any attempt to make a down mode. May explain why most OTA based ECFs have no down mode and the Boss FT-2 has such a complicated envelope section.

If the resistance between the cap and the OTA works as the drain, adding a pot there would be a decay control, would it not? That would probably also change other thing, wouldn't it?

Replacing the 1M resistor with a pot would make a Range control, right. That would definitely be interactive with a potential decay control. But hey, you need a pilot license for a dynamic filter of a certain complexity anyway... 

BTW there is a youtube video with a FX-25 with a "synth mode". By the sound of it I would assume that that simply disconnects the 22µ cap, making the filter follow the dynamics instantly for the full-on ripple effect, we normally want to avoid. However, I could not find anything to verify or falsify this. Does anyone know. Great option for some space funk.

Cheers,
Andy

[Transmogrifox]

It would be a little bit complicated to modify an existing FX-25B to use a reverse mode, but when designing from the ground-up you can add a second rectifier path that works in reverse -- just reverse the diodes.  The you can switch between the rectifier that pulls down, or the one that pushes up.

The switch function would switch to the second rectifier.  The second rectifier output side would have another lower-valued resistor to Vcc to pull up to your max range, or maybe a transistor configured as a constant-current source so you still get the same time constant against the capacitor for attack and release times (the 10k resistors going to the CV inputs dominate the release time constant).

I don't see it as being particularly more complicated than what you have to do with a Vactrol.

As for your comment about range control, replacing the 1M resistor with a pot wouldn't change the decay time very much until it got close to 10k.

About decay control, you would get a change in decay time by putting a pot in series with the 10k resistors to the OTA CV inputs, but you would also change the max range of the filter.  This is the control that would be notably interactive.