Will the real MS-20 filter please step forward

[Strategy]

So I just built a "Korg MS-20" based OTA (13600) filter, something I found on a dutch forum whilst googling "MS-20 filters." I have lost the link so I'll have to refind it, although I'm hesitant to post it as there are some errors in the document.

Anyways, in spite of the errors (and the fact that I can't read Dutch) it sounds good. It is switchable between Lowpass and Highpass. Runs on 9V. Seems very similar to the Tim Escobedo Sallen Key circuit except with clipping diodes and highpass settings added.

Now, I have an original Korg MS-20 keyboard and the real magic of it is that you have both high pass and a lowpass filter - I'm not sure if they are in series or in parallel, but you can use the two together to create bizarre, vocal-like "formant" sounds, sort of a bandpass filter. Paul Perry's Frostwave "Resonator" circuit is the only pedal I know of to have an identical configuration. They are no longer made and I've never seen a schematic to know how devoted it is to the MS20's original design (I did have the luck to borrow one for many months a few years ago.) As a "pedal", having the two filters together is really, really much more interesting than having just lowpass or highpass because you can develop lots of interesting "extreme resonant stuck wah" sounds, or you can apply different CV's to each filter (say, two LFO's) for very complex modulations. Sounded great with basically everything I put through it from guitar to keys, drum machines etc.

If I wanted to build a "workalike" of the original filter could I put a highpass and a lowpass in series? ...Or would I want to have the signal split into each filter and then the two mixed back together?

Given that I know very little of circuit development, anyone know of a DIY MS-20 filter project that is devoted to the "dual highpass and lowpass" of the original synthesizer?

Could this be a forum project in the making? Anyone else interested in this? I have some MS-20 service doc's ( I think) but probably not the technical expertise required to adapt the real synth circuit to a 9V, DIY-friendly pedal format. Willing to help where I can if there's any interest

STRATEGY

[Jordan A.]

Oooh, fun.  You probably have seen Jurgen Haible's MS-20 schematic?  It's hiding out in the schematics section with the JH-720 stuff.  

Another good one is the Rene Schmitz version http://www.uni-bonn.de/~uzs159/

Those seem to be the two general approaches, Jurgen used transistors and Rene used OTAs.  I've tried both of those and a couple of others floating around out there, they all sound kind of different, but all sound good to me and stick out as "MS-20ish".  Note that Rene's has a crazy amount of resonance available, way beyond oscillation, for crazy frequency locking stuff kind of like Ian Fritz's Threeler.  You can turn down the gain of the feedback amp if needed easily enough.

I personally like the Rene one, I have thought about doing a dual pedal kind of like you are talking about with it.  I think with a VCF with such wild resonance and such interesting behavior past self oscillation it would be cool to include a gate circuit, so when the resonance was cranked it would be muted while you were not playing.

Back to the main point, the MS-20 filters were in series for sure, but if I were making one I would include provisions for running them in parallel as well, both ways will sound cool and do things the other way can't.  I think a useful starting point for this kind of effect is the general layout and interface of the Mu-Tron Biphase, most of what you would want it to do is covered already.

There is a nifty way to have series or parallel options using switching jacks (without an extra patchcord), so you don't need a dedicated switch for it like the Biphase has, I can sketch it up if we get that far.  Some people like to mess with that switch mid performance though, so you may want to leave it.

Another cool mode is "difference" where you input the same signal to both filters and then sum the outputs but with one of them inverted, I've only done it with two low pass filters but there may be some cool stuff with HP and HP/LP combos.  Two LPFs set up this way sounds kind of like a band pass response, and kind of like a phaser.

I think an envelope follower is almost a must, in addition to at least one LFO, preferably two.  In a recent VCF pedal I did I used an envelope follower with an "attack" control which, when set to the fastest attack setting let a ton of ripple through, which sounds kind of like FMing your filter with the same signal going through it, except an octave higher.  I might consider something like that or just a provision for FMing the filter with the input signal.

For a project that is already going to be big I say just suck it up and do the bipolar power supply, it's barely more work than all the biasing and AC coupling you need to do otherwise, but that may not be the majority opinion.

I have a bunch more ideas for this type of pedal, and a bunch of it has actually been tested already, so I'm excited to hear someone else bring it up!

jordan

[rousejeremy]

Is this at all based on the MS20 or is it just a marketing gimmick?
http://martoneaudio.com/BrainFreeze.htm

[Jordan A.]

Is this at all based on the MS20 or is it just a marketing gimmick?
http://martoneaudio.com/BrainFreeze.htm

I bet it's very similar (or identical) to one of the well known variants out there, it's only a couple bucks in parts and the schemos are well known.  It probably sounds cool but if I was doing it I would definitely include some modulation options on board, kind of a pain to have to patch up something you are going to want 99% of the time.

cheers,

jordan

[Strategy]

Ive been in touch with Martone, almost bought a Brain Freeze. Again, it's like the one I built: *either* LP or HP, but not both simultaneously, which is my desired pedal :-D

@ Jordan A. do you think the Rene Schmitz circuit is scalable down to 9V? I wonder if mods are required to make it work that way or you just get a lower overhead overall?

I'm pondering just building 2x this circuit I just made, hardwiring one for hp and one for lp and then boxing them up in series!

Strategy

Is this at all based on the MS20 or is it just a marketing gimmick?
http://martoneaudio.com/BrainFreeze.htm

I bet it's very similar (or identical) to one of the well known variants out there, it's only a couple bucks in parts and the schemos are well known.  It probably sounds cool but if I was doing it I would definitely include some modulation options on board, kind of a pain to have to patch up something you are going to want 99% of the time.

cheers,

jordan

[frequencycentral]

Subscribed! We'll talk about this later sweetheart. 

[Mike Burgundy]

I'm not the only Dutchie on here, but holler if you need some translation services ;P

[FUZZZZzzzz]

the layout and schematic you refer to are on www.synthforum.nl (a dutch synthesizer forum)

heres the thread http://www.synthforum.nl/forums/showthread.php?t=138512&highlight=ms+20+filter

hope you can see the attachments without an account

[Skrogh]

If you combine the two schematics, so you put a 10k or 50k potentiometer to fade between the low-pass and hi-pass INPUTS (low-pass is into the OTA, Hi is into the cap in the last stage), you will get the filter response of the MS-20 (more or less, the resonance is a bit different in the OTA version linked). This will create a notch or band-stop filter. Those can sound very vocal.

[Jordan A.]

Good news, the Rene Schmitz version seems to work fine at +/- 9 volts, I just lowered the supply voltage on mine and at a first glance not much changed.  I'll spend a little more time with it to see if anything major needs fussing with but it seems promising.

Some changes I would make/general notes if using it for instrument level signals:

-the resonance control is set up for bonkers amounts of feedback, really cool for whacky sounds but if you want a more "normal" amount of feedback change the 1k8 resistor going from the inverting input of A3 to ground to something like 6k8 - 8k2 (or whatever you like, bigger resistor means less feedback).

-for instrument levels i would change the LEDs in the feedback loop of A3 to something with a lower voltage drop, 1N914 or whatever small signal diodes you have around work, or whatever you like.  With diodes with a larger voltage drop the feedback gets dangerously loud compared to your guitar signal.

-I used an LM13700 with no modifications, I sometimes see people tying the input to the darlington buffers to v- when they are not used, maybe someone who knows what they are talking about could weigh in on that?

-I used the first PNPs i found in the box, since we are not too worried about 1v/oct performance or drift you probably don't need to be too picky about these.

-it might want a high impedance buffer first if using it with guitars, a FET would be cheap and easy, or another opamp/whatever you like.  It does work fine without it and sounds good to me as is, so whatever is your preference.

-for high pass you just do the old trick of grounding the input resistor, lifting the second caps grounded leg and inputting your signal there, just like the schemo from the dutch forum.  You can do this with a switch or switching jacks, or just hardwire it, or mess around with faders like Skrogh says.

-it doesn't take up much room on a breadboard and it's not picky about exact parts or values, might as well do it up and get it working how you want first.

ok!

jordan

:edit for more notes:

yup, it sounds good with a good range for the cutoff frequency.  I measure between 8 and 13 milliamps drawn for the positive rail and 9 and 14 milliamps for the negative rail at +/- 9 volts, for just the circuit as drawn by Rene, LFOs/buffers and what not would add a bit.  Current drawn changes with cutoff frequency, higher frequency = more current.

[Strategy]

I have a deadline so I won't be able to give this more juice for a couple of days but you all rock for the great tips/comments!

[Ronan]

Thanks to all too, I am interested in building two filters in parallel, with different CV's, and am now 10 times wiser, appreciate the input.

[Jordan A.]

I had forgotten about Tim Stinchcombe's study of the various MS20 VCFs, interesting reading for sure http://www.timstinchcombe.co.uk/index.php?pge=korg

It looks like Korg put the HP first, if it matters to anyone.

jordan

[space_ryerson]

It looks like Korg put the HP first, if it matters to anyone.

I think you're right. There's a neat trick you can do with an MS-20 where you turn up the resonance on the HP, and set the cutoff knob pretty low. This creates a big bass hump, which you then filter with the LP. Mega bass!

[Processaurus]

It looks like Korg put the HP first, if it matters to anyone.

That's a good strategy, as any self noise/hiss from the HP filter circuit gets filtered out by the LP filter after it.  Cool thread!

[Strategy]

OK, finally some time to reply.

So, the great ideas in the thread have presented a number of questions about how to proceed with a "true" HP/LP Korg MS-20 filter. In the meantime I have spent a few days with my "semi-true" Dutch-layout MS-20 filter - single sided power supply, etc. While it does sound awesome, I can say that the low headroom is a little bit noticeable.

I'm wondering how this could become a forum project with a proper layout and how I could sort of chip in and do my part given that I don't know how to create layouts. In any case I would be able to commit to organizing a group buy of PCB's, perhaps using the Dorkbot-PDX service (based in my town). Any of you layout savvy people up for participating?

So first, questions:
1 - What are peoples' objections to an onboard bipolar power supply, if any? 18v (or so) wallwart would be required. I know most of us favor 9V but the +/-15V psu could be incorporated into the circuit simply and give a LOT of the headroom. And I'm thinking back to my time with the Frostwave Resonator, which sounded killer, and used the full power supply but without making the project extra big or bulky. It just means you need that particular wall wart and can't use batteries or your 1-spot. In my opinion the sonic benefits outweigh the inconvenience or cost of having 1 extra wall wart.

2 - What are peoples' opinions about 13xxx chips vs. CA3080? there seem to be more producers/versions of the 13xxx chip and relatively good availability, although they are not cheap. CA3080's aren't the most expensive vintage chips in the world, BUT, I just sort of have a tendency to avoid obsolete chips on principle. Opinions? This would determine to some extent what circuit is used as a starting point, e.g., the Rene Schmitz version, one of the Korg versions, or a 13xxx version.

3 - How much modulation is enough? I am torn between the simplicity of the Frostwave Resonator (CV inputs, no onboard modulation - supply your own LFO's!) and the idea of having one LFO per filter and possible an envelope generator. I know there are plenty of compact versions of those that can plug in.

My idea for the "MS-20 Filter Pedal" project - bare bones version:
Classic configuration, serial with HP before LP. Plenty of resonance!
Mix/blend control
input gain/volume
CV input for HP
CV input for LP

Mega version:
HP into LP plenty of resonance
Mix/blend control
input gain/volume
CV input for HP
CV input for LP
CV input for resonance
LFO for HP
LFO for LP
Env. follower
The onboard CV's should be able to go to zero so that you can use just external CV's when you want or to use the filter in a static/manual way.

I propose to name this FROSTY THE FILTER or something else that acknowledges the legacy of erstwhile forum member PAUL PERRY who many of you will remember. My project seems to be clone-ish of his out of production Frostwave Resonator so it seems appropriate to pay respects where due

Opinions appreciated and layout-creators sought!

Please post your interest in a possible board run if this were to go forward!

STRATEGY

[Cliff Schecht]

1) You can easily use the MC34063, UCC380x or a similar SMPS controller to build a 9V to +/-15V power supply. If you wanted to get real clever you could even find an off-the-shelf coupled inductor with two secondary windings to generate your +/-15V in a very small footprint. The first option requires you to run two separate chips (one as a boost and another as an inverter) but would have cheaper/easier to source magnetics. The flyback is the optimal solution for small, simple power (especially because we aren't talking more than 50-100mA current draw).

2) The 13700 has a larger input range and will give you more headroom before distortion. The CA3080 will act more non-linear which can be a very cool thing in a synth type filter (especially when the distortion is introduced IN the filters and/or feedback loop). With that said I've used the 13700 in two filter designs and had no real qualms about it. The filters were DC coupled wherever possible so the bass response was great (sounded HUGE with low frequency PWM'ed square waves).

3) This is a tough one but IMO the modulation options should be made as flexible as possible with as little circuitry as possible. You can use switching jacks to disable internal LFO's when an external CV is used, this is commonplace in the synth world. If you do separate LFO's then it's nice to be able to sync them and also for guitar purposes a flexible envelope follower is always nice. I think in the end it depends on what the end user wants to use the filters for.

[Strategy]

@Cliff, I'm not familiar with those power options, will research!

I think I am also leaning toward 13700. The project I just built (link above) uses 13600, what's the relative advantages of that vs. 13700?

re: uses of the filter pedal - I think it should be a "use for everything" project, which is where the Resonator shines - it was very popular in spite of not having on board modulation, even as a 'static' filter you can dial such crazy overtones. I would probably not do switching-jacks to disable LFOs because it can be cool to mix LFO with your incoming CV signal, too.

Strategy

1) You can easily use the MC34063, UCC380x or a similar SMPS controller to build a 9V to +/-15V power supply. If you wanted to get real clever you could even find an off-the-shelf coupled inductor with two secondary windings to generate your +/-15V in a very small footprint. The first option requires you to run two separate chips (one as a boost and another as an inverter) but would have cheaper/easier to source magnetics. The flyback is the optimal solution for small, simple power (especially because we aren't talking more than 50-100mA current draw).

2) The 13700 has a larger input range and will give you more headroom before distortion. The CA3080 will act more non-linear which can be a very cool thing in a synth type filter (especially when the distortion is introduced IN the filters and/or feedback loop). With that said I've used the 13700 in two filter designs and had no real qualms about it. The filters were DC coupled wherever possible so the bass response was great (sounded HUGE with low frequency PWM'ed square waves).

3) This is a tough one but IMO the modulation options should be made as flexible as possible with as little circuitry as possible. You can use switching jacks to disable internal LFO's when an external CV is used, this is commonplace in the synth world. If you do separate LFO's then it's nice to be able to sync them and also for guitar purposes a flexible envelope follower is always nice. I think in the end it depends on what the end user wants to use the filters for.

[Cliff Schecht]

In the 13600 the bias of the output buffer is dependent on the current flowing into the Iabc (current control) pin. This is not what you want for audio because that bias jumping all willy nilly can cause the buffer circuit to make popping sounds. The 13700 has a buffer whose bias is independent of the Iabc current which makes it much better for audio. With that said, for filtering applications I ignore the output buffer completely and use a TL07x or similar op amp to buffer the OTA/cap integrator circuit. Prevents the stages from loading each other and makes for a more accurate filter (assuming you match the caps decently) which IMO gives circuits a better sounding resonance.

[frequencycentral]

I've layed out (single sided) PCBs for both 3080 and 13700 based VCFs, I've found that the 'mirror-image-ness' of the two OTAs/buffers of the 13700 can make for layout difficulties, requiring some complex routing and jumpers, and wasted space. Whereas 3080s you can just line 'em up - even adding FET buffers takes up little space in comparision with 13700. Of course, this would not be an issue if you go for double sided. Examples:

http://dl.dropbox.com/u/967492/121%20VCF/Roland%20121%20VCF%20Build%20Doc.pdf

http://dl.dropbox.com/u/967492/Neu%20Multimode%20PCB.gif
http://dl.dropbox.com/u/967492/Neu%20Multimode%20PnP.gif

Having said that, I'd always go for 13700 over 3080 if possible and practical, for price and availability.

Re the LFO(s), have you considered that the TAPLFO2 would make a very flexible addition to this design?