Will the real MS-20 filter please step forward

[Cliff Schecht]

If you can hand me a full schematic I can knock out a layout pretty quickly. Since I have a soft spot in my heart for synth filters of any kind I would enjoy doing the board. I usually don't offer to do boards for others anymore because I've had quite a few bad experiences with people getting pushy even though I'm doing work for FREE but again, I'm a sucker for any VCF topology I haven't played with.

I also think we should look at adapting the Steiner filter (diode/cap based VCF) for use with guitar. When I was at PAiA I actually adapted and modernized the design a bit and REALLY enjoyed playing with that filter. They never released it as a product but I have a bunch of working prototype boards that work off of +/-12V or +/-15V (part of the circuits biasing has to be modified depending on the supply voltage IIRC). If there is interest I can dig up my old project files and post them up.

Also FWIW I think it's worth the effort to try to redesign everything to run off of a single 9V neg center power supply. While this really means adding a boost SMPS for +12V/+15V and an inverting SMPS for -12V/-15V, neither of this circuits are hard to build or troublesome! The parts needed for these supplies rival the cost of a linear regulator solution and you can eliminate the need for extra wall warts (a big plus in my book). The flyback option is also there but I will have to look and see if a suitable off-the-shelf thru-hole coupled inductor is available for cheap.

[Strategy]

that rules Cliff. thank you!!! WE HAVE A VOLUNTEER! 

What about some kind of MAX1044 based power solution? +9V boosted up to 12 or 15 bipolar? The Gristleizer project and others use this but I can't remember it boosts power and produces the negative power as well. Smallbear and popular suppliers carry the chip and this forum is pretty comfortable with projects that use it.

Consider this work not for me, but for the forum community. If the test case goes well I'll do a run of PCBs, an organized project document, a build thread etc. and we can get some people going on a classic filter in an organized fashion.

I have heard similarly good things about the Steiner. The MS-20 takes the cake for me and it does have its place in FX/guitar/history, via Frostwave and others - it's a good filter for this particular group. We'll consider it the "gateway drug" to the forum's future synth filter addiction!

the Stinchcombe or Rene Schmitz schematics Rick posted seem to be the best starting points although the Schmitz one ought to have the CA3080's replaced with the LM13700.

the input stage (accommodate input levels) could be a typical LPB, correct? Or would this even be necessary?
seems like the modulation theme that keeps popping up is one LFO per filter but I'm still stumped on a circuit that allows syncing or inverting them. My feeling is, simple is good and MODDABLE is good.

Maybe for a layout we start with the most basic building block, the filter itself and do the modulation as a companion board once the elemental part of it is verified? or knit the two layouts together...

Strategy

If you can hand me a full schematic I can knock out a layout pretty quickly. Since I have a soft spot in my heart for synth filters of any kind I would enjoy doing the board. I usually don't offer to do boards for others anymore because I've had quite a few bad experiences with people getting pushy even though I'm doing work for FREE but again, I'm a sucker for any VCF topology I haven't played with.

I also think we should look at adapting the Steiner filter (diode/cap based VCF) for use with guitar. When I was at PAiA I actually adapted and modernized the design a bit and REALLY enjoyed playing with that filter. They never released it as a product but I have a bunch of working prototype boards that work off of +/-12V or +/-15V (part of the circuits biasing has to be modified depending on the supply voltage IIRC). If there is interest I can dig up my old project files and post them up.

Also FWIW I think it's worth the effort to try to redesign everything to run off of a single 9V neg center power supply. While this really means adding a boost SMPS for +12V/+15V and an inverting SMPS for -12V/-15V, neither of this circuits are hard to build or troublesome! The parts needed for these supplies rival the cost of a linear regulator solution and you can eliminate the need for extra wall warts (a big plus in my book). The flyback option is also there but I will have to look and see if a suitable off-the-shelf thru-hole coupled inductor is available for cheap.

[Cliff Schecht]

that rules Cliff. thank you!!! WE HAVE A VOLUNTEER! 

What about some kind of MAX1044 based power solution? +9V boosted up to 12 or 15 bipolar? The Gristleizer project and others use this but I can't remember it boosts power and produces the negative power as well. Smallbear and popular suppliers carry the chip and this forum is pretty comfortable with projects that use it.

Consider this work not for me, but for the forum community. If the test case goes well I'll do a run of PCBs, an organized project document, a build thread etc. and we can get some people going on a classic filter in an organized fashion.

I have heard similarly good things about the Steiner. The MS-20 takes the cake for me and it does have its place in FX/guitar/history, via Frostwave and others - it's a good filter for this particular group. We'll consider it the "gateway drug" to the forum's future synth filter addiction!

the Stinchcombe or Rene Schmitz schematics Rick posted seem to be the best starting points although the Schmitz one ought to have the CA3080's replaced with the LM13700.

the input stage (accommodate input levels) could be a typical LPB, correct? Or would this even be necessary?
seems like the modulation theme that keeps popping up is one LFO per filter but I'm still stumped on a circuit that allows syncing or inverting them. My feeling is, simple is good and MODDABLE is good.

Maybe for a layout we start with the most basic building block, the filter itself and do the modulation as a companion board once the elemental part of it is verified? or knit the two layouts together...

Strategy

If you can hand me a full schematic I can knock out a layout pretty quickly. Since I have a soft spot in my heart for synth filters of any kind I would enjoy doing the board. I usually don't offer to do boards for others anymore because I've had quite a few bad experiences with people getting pushy even though I'm doing work for FREE but again, I'm a sucker for any VCF topology I haven't played with.

I also think we should look at adapting the Steiner filter (diode/cap based VCF) for use with guitar. When I was at PAiA I actually adapted and modernized the design a bit and REALLY enjoyed playing with that filter. They never released it as a product but I have a bunch of working prototype boards that work off of +/-12V or +/-15V (part of the circuits biasing has to be modified depending on the supply voltage IIRC). If there is interest I can dig up my old project files and post them up.

Also FWIW I think it's worth the effort to try to redesign everything to run off of a single 9V neg center power supply. While this really means adding a boost SMPS for +12V/+15V and an inverting SMPS for -12V/-15V, neither of this circuits are hard to build or troublesome! The parts needed for these supplies rival the cost of a linear regulator solution and you can eliminate the need for extra wall warts (a big plus in my book). The flyback option is also there but I will have to look and see if a suitable off-the-shelf thru-hole coupled inductor is available for cheap.

I'm just not a big fan of charge pumps. They are quick and dirty power solutions meant for low current applications. They don't provide regulated voltage either, the output voltage is a function of the input voltage and load resistance. With an SMPS solution the output voltage is stable regardless of input voltage and load resistance (assuming you stay within the specs). For designs where there are bias points based on the supply voltage (like the linear-to-expo converter circuit) you want the supply to be pretty steady. Since the OTA's supply current varies with the bias pin current you will be causing the supply voltage to droop a bit as you go towards the higher octaves. This is mainly why I'd like to avoid charge pumps, especially when an SMPS solution is really just as cheap and simple.. People should not be intimidated by switching power supplies, in just about any situation we encounter they are easy to design and use, plus there are many people on this forum who can help others figure out any issues they are having..

Going from the 3080 to the 13700 isn't a big deal as long as you adjust the bias circuit accordingly to properly feed current into the Iabc pin (I have to look into this). The rest of the circuit has essentially the same drive requirements. FWIW I think it's worthwhile to use the linearizing diodes in the 13700, I use them in all of my filter designs to get a bit more headroom before distortion. Also since you have a larger input range you can get a better signal to noise ratio (i.e. a quieter circuit in operation).

I don't see why a BJT-based input (LPB) would be needed when an op amp circuit is simple and just as effective. We aren't driving long capacitive cables, we are driving the inputs to an OTA (relatively high impedance) which an op amp is fine and dandy for. The veritable TL07x op amps are great in filter circuits IME as they have super high input impedance, decent drive capability and decent DC performance (important for DC coupled circuits mostly).

I'll look at the design tomorrow but will leave it up to you to choose the LFO options you want. It's really the OTA linear-to-expo converter that is the tricky part, and I have a simple yet effective circuit already designed to do this with the 13700's.

[Jordan A.]

Cliff, it's great to see you involved in this one, I have always done linear power supplies but it would be fun to mess with whatever you come up with.  I'm pretty useless for PCB layouts but I can breadboard and test various options as they come up, as well as building a hard version on perfboard when it comes to proof of concept time.

Strategy, is it correct that for now you want to focus on a hardwired HP into LP series setup?  I appreciate you reigning things back into sanity territory (but I do think we should make it easy to access the parallel and stereo modes, they come for free when you build it!) and I agree that getting the filter itself built up first is a good idea, and afterwards modulation options can be worked out.

I think I'll breadboard a second VCF and put it in series with the one I have and see how it sounds, and then spend some time modulating them.  It would be cool if someone else did this as well so we could compare findings.

What do you guys think about the frequency control setup I proposed (one master frequency knob and one "separation" knob)?  It's either that or two independent frequency knobs, but that way you need two hands to do a manual frequency sweep.  The "separation" control does add one opamp and four resistors, but I like it better in practice.  The separation control could also be considered a bandwidth control.

As far as levels, it seems like there are two ways to go- boost the guitar signal up so that it is not swamped by the resonance when you crank it up, or make the clipping diodes smaller so that the resonance signal doesn't get so big in the first place.  If you use an opamp first and add a volume control at the end you can set the input gain to be whatever you want and then keep the volume level reasonable on the way out, kind of a boost it up and then knock it back down approach.  This is another thing that I would recommend breadbaording before deciding one way or another, since I have one built and working already let me know if you want me to try anything.

ok!

jordan

[Strategy]

@ Jordan, its probably a little bit dogmatic of me but I'm definitely inclined to follow the original MS-20 layout as the essential building block, so two separate frequency controls is ideal. The greatest thing about the filter I think is that you can control the two separately and dial in much more complex voicings.

The reason I keep leaning towards "simple" is maybe personal and practical in nature. I just find that with the more complex projects, it can take me literally years to finish them. Super fast projects, on the other hand, can be a little limiting -- and the 'cut corners' can be apparent (like in the Dutch MS-20 filter I just built.) There's kind of a medium zone that has a) easy to find parts b) moderate cost c) not-too-crazy wiring that makes a project "just right". Where you can sit down, do it in a couple of days and go straight to music making.

On the other hand I think all the mod ideas are crucial, so the project/layout/pcb has to leave "doors open" to do those - extra ground pads, maybe even a grid of extra pads somewhere? An objective would be to develop specifically docuemented expansions and mods here in the forum. Given that it's a synth module with pretty easy CV connections, I can think that someone with pretty basic knowledge could customize the circuit in a lot of different ways. I would rather open the doors to that than develop a project where the "optional" configurations are prescribed for the builder and might scare off people with less experience (people like me!)

Strategy

Cliff, it's great to see you involved in this one, I have always done linear power supplies but it would be fun to mess with whatever you come up with.  I'm pretty useless for PCB layouts but I can breadboard and test various options as they come up, as well as building a hard version on perfboard when it comes to proof of concept time.

Strategy, is it correct that for now you want to focus on a hardwired HP into LP series setup?  I appreciate you reigning things back into sanity territory (but I do think we should make it easy to access the parallel and stereo modes, they come for free when you build it!) and I agree that getting the filter itself built up first is a good idea, and afterwards modulation options can be worked out.

I think I'll breadboard a second VCF and put it in series with the one I have and see how it sounds, and then spend some time modulating them.  It would be cool if someone else did this as well so we could compare findings.

What do you guys think about the frequency control setup I proposed (one master frequency knob and one "separation" knob)?  It's either that or two independent frequency knobs, but that way you need two hands to do a manual frequency sweep.  The "separation" control does add one opamp and four resistors, but I like it better in practice.  The separation control could also be considered a bandwidth control.

As far as levels, it seems like there are two ways to go- boost the guitar signal up so that it is not swamped by the resonance when you crank it up, or make the clipping diodes smaller so that the resonance signal doesn't get so big in the first place.  If you use an opamp first and add a volume control at the end you can set the input gain to be whatever you want and then keep the volume level reasonable on the way out, kind of a boost it up and then knock it back down approach.  This is another thing that I would recommend breadbaording before deciding one way or another, since I have one built and working already let me know if you want me to try anything.

ok!

jordan

[Strategy]

I'm just not a big fan of charge pumps. They are quick and dirty power solutions meant for low current applications. They don't provide regulated voltage either, the output voltage is a function of the input voltage and load resistance. With an SMPS solution the output voltage is stable regardless of input voltage and load resistance (assuming you stay within the specs). For designs where there are bias points based on the supply voltage (like the linear-to-expo converter circuit) you want the supply to be pretty steady. Since the OTA's supply current varies with the bias pin current you will be causing the supply voltage to droop a bit as you go towards the higher octaves. This is mainly why I'd like to avoid charge pumps, especially when an SMPS solution is really just as cheap and simple.. People should not be intimidated by switching power supplies, in just about any situation we encounter they are easy to design and use, plus there are many people on this forum who can help others figure out any issues they are having..

Going from the 3080 to the 13700 isn't a big deal as long as you adjust the bias circuit accordingly to properly feed current into the Iabc pin (I have to look into this). The rest of the circuit has essentially the same drive requirements. FWIW I think it's worthwhile to use the linearizing diodes in the 13700, I use them in all of my filter designs to get a bit more headroom before distortion. Also since you have a larger input range you can get a better signal to noise ratio (i.e. a quieter circuit in operation).

I don't see why a BJT-based input (LPB) would be needed when an op amp circuit is simple and just as effective. We aren't driving long capacitive cables, we are driving the inputs to an OTA (relatively high impedance) which an op amp is fine and dandy for. The veritable TL07x op amps are great in filter circuits IME as they have super high input impedance, decent drive capability and decent DC performance (important for DC coupled circuits mostly).

I'll look at the design tomorrow but will leave it up to you to choose the LFO options you want. It's really the OTA linear-to-expo converter that is the tricky part, and I have a simple yet effective circuit already designed to do this with the 13700's.

@ Cliff, I did not know that about the charge pump option, so that is really helpful information. And the input circuit as an op amp also sounds good. In the meantime I have been looking in the Gallery section to see what people have done for LFO's. The TAPLFO is extremely tempting but I keep thinking "what do you do when it's gone?"

Strategy

[Jordan A.]

@Strategy, I appreciate the dogmatic approach, and think the goal of staying close to the original is laudable.  With that in mind...

With the original MS20, for each VCF you have frequency and peak controls, plus a level control for "modulation generator" and level control for "envelope generator 2".  So eight knobs total.  It seems like a pretty nice setup.

The modulation generator looks to be a triangle/square LFO with a "skew" control to tilt the triangle into ramps and alter the pulse width, we can whip up something like that easily with a dual opamp.  Two knobs here if you want to keep the skew function the way they did it, plus a switch if we want to switch between triangle and square.  

If we want to invert the LFO going to VCF B we could add a switch for reverse or normal sweep, or make the VCF B LFO level control an attenuverter.  Either way takes one opamp, the attenuverter option has one less switch but it's a little weird to have one knob go from 0 to 5 to 10 and the other go from -10 to 0 to +10.  I should add that a reverse sweep mode makes more sense to me in parallel mode, which sounds like it may be abandoned anyways, so this may be a non issue.

Instead of "envelope generator 2" I imagine an envelope follower will be easier and more flexible.  I personally get by with just sensitivity and attack controls, everybody feel free to suggest EF circuits if you have favorites.  I think attenuverters for the EF level controls make sense, it offers maximum flexibility with minimum panel controls.  I imagine the whole EF circuit plus attenuverters would be between 3 and 5 opamps.

So doing it pretty much like korg, I come up with 12 knobs, kind of a lot.  If you feel like that is getting out of hand the knob count could be thinned out some.  You may only need one control for the LFO level, for instance, and one control for the EF level.  Scraping the LFO skew function saves a knob, losing the EF attack control saves one.  The EF sensitivity control could also serve as the EF level control.  Get rid of all those and you are down to 7 knobs.  

I think I'm starting to see why you don't see to many commercial versions of this with modulation on board, it gets knob heavy really quick...

For those of us not scared of panel wiring, or those of us skilled at making layouts utilizing board mounted pots, it's still definitely worth while.

jordan

:edit: oh yeah, strategy, with the master frequency + separation control setup I was suggesting, you can still get the entire range of sounds.  You detune the filters with the separation/bandwidth control and sweep them both together with the frequency control, all variations are available.  That said, I'm happy to keep things towards the authentic side if you like, but for my own I know I prefer the one where you don't need both hands to sweep it.

cheers!

[Strategy]

So this is where my dogmatic nature is quirky/unpredictable...I really want to remain true to the MS-20 filter core but I'm not as attached to the MS-20's modulation section. For one thing the native LFO does not go slow enough for my taste (maybe my original MS-20 needs a calibration or some mods!)

So we could have 2x LFOs and an EF but should consider using circuits other than or smaller than MS-20's. 12 knobs is pretty intense. I've done my fair share of mega projects, but none that big yet. (anyone following the String Ringer thread?? holy crap - I've been working on one for about 2 years!)

I am more and more starting to feel like we could do the modulation section as a daughter board. Would be inexpensive and give some options for people. There is a side benefit that people could build standalone 9V modulation generators for use with projects that accept CV's (like Ring Stinger, some Tim Escobedo project versions, etc.) or commercial pedals that are expandable with CV's (like Moogerfoogers.) The filter core should have a 9V out for chaining sub boards and add on's. There are a number of minimal and low-ripple EF's in this forum perhaps like the ones used in Clari Not and others. It might be fun to see if the Meatball's Vactrol EF is applicable in some manner but vactrol's ROHS certification is twilighting and the manufacturer has been bought out. I don't believe it's discontinued line, but, they have gone up in price at least 150%.

I found a couple LFO layouts in the Gallery, but they  are untested by me. Will post links.

@Cliff: perhaps the filter core layout can proceed without any modulation at all?

@Jordan: aside from TAPLFO and the MS-20's native modulation, are there any "minimal" LFO circuits that jump out as good options?

Strategy

@Strategy, I appreciate the dogmatic approach, and think the goal of staying close to the original is laudable.  With that in mind...

With the original MS20, for each VCF you have frequency and peak controls, plus a level control for "modulation generator" and level control for "envelope generator 2".  So eight knobs total.  It seems like a pretty nice setup.

The modulation generator looks to be a triangle/square LFO with a "skew" control to tilt the triangle into ramps and alter the pulse width, we can whip up something like that easily with a dual opamp.  Two knobs here if you want to keep the skew function the way they did it, plus a switch if we want to switch between triangle and square.  

If we want to invert the LFO going to VCF B we could add a switch for reverse or normal sweep, or make the VCF B LFO level control an attenuverter.  Either way takes one opamp, the attenuverter option has one less switch but it's a little weird to have one knob go from 0 to 5 to 10 and the other go from -10 to 0 to +10.  I should add that a reverse sweep mode makes more sense to me in parallel mode, which sounds like it may be abandoned anyways, so this may be a non issue.

Instead of "envelope generator 2" I imagine an envelope follower will be easier and more flexible.  I personally get by with just sensitivity and attack controls, everybody feel free to suggest EF circuits if you have favorites.  I think attenuverters for the EF level controls make sense, it offers maximum flexibility with minimum panel controls.  I imagine the whole EF circuit plus attenuverters would be between 3 and 5 opamps.

So doing it pretty much like korg, I come up with 12 knobs, kind of a lot.  If you feel like that is getting out of hand the knob count could be thinned out some.  You may only need one control for the LFO level, for instance, and one control for the EF level.  Scraping the LFO skew function saves a knob, losing the EF attack control saves one.  The EF sensitivity control could also serve as the EF level control.  Get rid of all those and you are down to 7 knobs.  

I think I'm starting to see why you don't see to many commercial versions of this with modulation on board, it gets knob heavy really quick...

For those of us not scared of panel wiring, or those of us skilled at making layouts utilizing board mounted pots, it's still definitely worth while.

jordan

:edit: oh yeah, strategy, with the master frequency + separation control setup I was suggesting, you can still get the entire range of sounds.  You detune the filters with the separation/bandwidth control and sweep them both together with the frequency control, all variations are available.  That said, I'm happy to keep things towards the authentic side if you like, but for my own I know I prefer the one where you don't need both hands to sweep it.

cheers!

[Strategy]

Frostwave Resonator panel layout so you can get a sense of the elegant simplicity!

EDIT: I forgot that it had VC for resonance! I never found a completely familiar musical use for that but it is a popular bonus feature for filters in the modular synth universe

[frequencycentral]

VC resonance: just another OTA, see P5 rev2 VCF schematic.

[Cliff Schecht]

This control is especially easy to implement because the V-to-I converter for the OTA is linear instead of exponential (typically). I like the option for either linear or expo for some controls but haven't played around much even with the voltage controlled resonance I've implemented in designs.

[Jordan A.]

Not much time this morning, but yes, a classic integrator/comparator LFO is super easy to build and only uses two opamps.  It's very easy to set those up for whatever speed you like, and it's also easy to set them up with a very large range of speeds available.  If you have time check out the LFO in Jurgens SOST flanger, it's super cool- triangle, slewed square and sample + hold, it can be built with six opamps (no OTAs) and one single pole on-off-on switch.  I think a very nice setup is one LFO like that + one simple triangle only LFO, for eight opamps total you get a LOT of LFO action.

I don't personally see VC resonance being all that useful, it's not too hard to implement but it does add parts.

cheers,

jordan

[Jordan A.]

Spent some time with two filters in series today, big fun.  An reverse sync option for the LFO will be quite useful I think, lots of "yoy" in there with the two cutoff frequencies dialed in.

My opinion after playing with it, if this is going to be a strictly series affair, is that one LFO will do.  It's still definitely cool with two, but if we are counting features and trying to keep it trim I think one LFO (with the option to reverse the sweep to one of the VCFs) will suffice.  I'm a fan of wild modulation and I could change my mind about that at any time, but after playing with it I still think separate LFOs make more sense when the VCFs are in parallel.  You can always add jacks for CV input.

Another issue is the feedback, Rene's version as drawn will self oscillate over a large portion of the resonance control, and the oscillation signal gets up to around 15 volts peak to peak at certain frequencies with the control maxed, you probably don't want to send that to a guitar amp.  I think the easiest and possible best solution will be to simply change the clipping diodes in the feedback amp, I'll mess with this some more.  I also added a 5k trimmer wired as a variable resistor in between the 1k8 feedback resistor and ground, so I could tailor the feel of the feedback control.  Since the whole point of an MS20 VCF is wild and nasty resonance we don't want to do anything to kill off it's character, just ensure that we don't blow up our speakers.

One of you guys has got to breadboard one so we can compare notes!  C'mon, do it!

jordan

[Cliff Schecht]

You can always design less gain into the resonance circuit as well. IME a poorly set range on any resonance pots can really make some ears bleed. Maybe some sort of simple compressor would be of use to prevent volume spikes from overwhelming an amp. This could also help counteract the relatively large amplitude variations between the highest point of the passband and the lower portions (especially when Q is set high).

[Jordan A.]

You can always design less gain into the resonance circuit as well. IME a poorly set range on any resonance pots can really make some ears bleed. Maybe some sort of simple compressor would be of use to prevent volume spikes from overwhelming an amp. This could also help counteract the relatively large amplitude variations between the highest point of the passband and the lower portions (especially when Q is set high).

Yes, that's what i did with the trimmer in series with the lower leg of the feedback divider.  The thing is the out of control way beyond oscillation stuff is sort of what makes this VCF special in the first place, so too much "improvement" might end up neutering it.  I absolutely like your idea of a compressor at the end, when set up for a narrow band pass sound the volume is obviously reduced, a compressor would help with that as well.

There are really two distinct parts to this i guess,

-the gain of the feedback amp as set by the resistors, this largely controls how the resonance control feels and how soon oscillation sets in

-the clipping threshold as set by the diodes, which largely controls how large the oscillation signal is after the turn diodes on

(I know you already know all this stuff Cliff, I'm figuring it out as i go)

I think I like the idea of a trimmer for resonance gain and smaller clipping diodes, for general guitar use.  Maybe a zener clipper at the end would serve as a super simple limiter, it might allow the resonance do it's wild thing while preventing speakers from being shredded?  I'll try some more stuff out...

jordan

[Cliff Schecht]

I don't like trimmers. To me they are a necessary evil and circuits that rely too heavily on them annoy me (I'm trying to refrain from being too snobbish about circuit design on a DIY forum ). With that said, if you are intending to increase the flexibility then why not just have a resonance pot that you can tweak externally? This is one of my favorite knobs on a synth usually..

That's exactly the reason I recommend the compressor. The voltage controlled resonance circuit (if this is implemented, I LOVE this on a synth fwiw) would only use one side of a 13700 and so the second side could be used to implement a simple one or two knob compressor. I like the idea of a compressor more than a limiter because as you said filters can vary all over the place amplitude-wise. This might help make the effect more usable in a much wider variety of places. This is the reason most synth filters don't make great guitar filters, they tend to get much higher Q than anything you typically find in effects pedals.

[Jordan A.]

I don't like trimmers. To me they are a necessary evil and circuits that rely too heavily on them annoy me (I'm trying to refrain from being too snobbish about circuit design on a DIY forum ). With that said, if you are intending to increase the flexibility then why not just have a resonance pot that you can tweak externally? This is one of my favorite knobs on a synth usually..

That's exactly the reason I recommend the compressor. The voltage controlled resonance circuit (if this is implemented, I LOVE this on a synth fwiw) would only use one side of a 13700 and so the second side could be used to implement a simple one or two knob compressor. I like the idea of a compressor more than a limiter because as you said filters can vary all over the place amplitude-wise. This might help make the effect more usable in a much wider variety of places. This is the reason most synth filters don't make great guitar filters, they tend to get much higher Q than anything you typically find in effects pedals.

Sure, I just stuck a trimmer in so that I could easily change the gain of the feedback amp while playing it.  I do have an external resonance control (actually two, one for each VCF).  I think this is a parameter that people may want to tailor to their own needs, ideally a few other people will breadboard it and we can decide on what to do with it.  I posted earlier about a value change for one resistor that will make the resonance control feel more "normal", alternately you can stick a trimmer in and set it to your liking, or you could switch between two resistors with a "tame" and "wild" switch.  I'm quite happy deciding what to do for myself, but my preference may not be to everyone's liking.  I'm just trying to feel out what you guys, and specifically Strategy, will like.  I agree that in the final product a trimmer here is probably unnecessary, in the meantime it seems useful, until the gain of the feedback amp is decided on.

I did a stereo series/parallel filter box a few months ago (very similar to what we are talking about here, but with two 4 pole LPFs), and I did include two NE570 compressors, with just a simple blend control to dial in how much compression was wanted.  I stuck it on a dual pot so one knob controlled both sides.  I also included a simple FET distortion circuit x2 before the VCFs, with one dual pot controlling both channels.  Both of those additions are super handy.  This is just to say that I'm a fan of big bloated circuits, but I've been trying to reign in my own tendency towards wanton excess in order to keep this project accessible.  I think an almost more handy feature for this specific filter would be a noise gate, to mute any oscillation when you are not playing...

On another note, changing the clipping diodes to 1N914 seems to be just about right for my guitar (with P90s).  The resonance when oscillating doesn't get speaker rippingly loud compared to the guitar signal and the general character seems intact.

And I know this isn't where you wanted to go with it, but Strategy, when you get one built up you must try them in parallel.  If you are a fan of complex formanty sounds parallel offers a lot of variation you don't get with series operation, as you can have the HP up higher than the LP without the sound going away, extra yoy.

cheers,

jordan

[Strategy]

I'm back after a few days gone...

resonance control for both HP and LP will be important. that makes two minimal compressors. One BIG advantage to including those is that when you raise the Q there is the appearance of overall volume drop in many filters. Which is why so many synths need a compressor after them so that when you modulate your sound from an low-resonance, fairly unfiltered sound to a very filtered one you get relatively even volume across the whole sound.

I did not realize that a compressor could be done with part of an LM13700. What a flexible chip! And I just dug up a ring modulator circuit that also uses this same chip! What CAN'T it be used for? 

Strategy

[teej212]

going back to the power supply- I think that it would a smart move to make it 9v powered. not doing this might turn a lot of people away, and compared to the rest of the circuit, this would not be that difficult to add. is the hesitance in the power supply being unstable?

[Mark Hammer]

Frostwave Resonator panel layout so you can get a sense of the elegant simplicity!

EDIT: I forgot that it had VC for resonance! I never found a completely familiar musical use for that but it is a popular bonus feature for filters in the modular synth universe

Aww man!  Now you made me go and start missing Paul Perry all over again!!

Why'd you have to go do that, and where the dickens is Paul??