Sunday, 15 December 2019

Synthesizer Build part-12: THE KORG MS20 FILTER.

A good working version of the famous Korg MS-20 filter by Rene Schmitz, with updated stripboard layout and wiring diagram with LP & HP .

This is a filter I absolutely had to include in my DIY synthesizer project, for one because Sam Battle from LookMumNoComputer raves about it and it sounds amazing in his videos and the other reason is that it has the option to go between Low-Pass and High-Pass and I didn't have a High-Pass option in the synthesizer yet. I also added the Band-Pass option in the layout drawing however I  tested it and it adds no extra benefit to this filter. Of course, combining this filter in High-Pass mode with one of the other lowpass filters gives you the bandpass option too and with a much better sound! (Check out the video at the bottom of this article to hear this filter in series with the Moog Ladder Filter. A really cool combination.) And if you build the Dual Korg Filter you'll have an even better bandpass option but that's for an other article.
There was a discussion on the Synth DIY Facebook group a while back about the possibility of switching this filter between 6dB per Octave and 12dB per Octave. I've tried it and it works but only in LowPass mode. More about this further down the article.
Btw, I tested this filter on dual 12 Volt and it works just as well as on 15 Volt so no need to change anything if you are feeding it from a +/- 12V powersupply. Of course you need to open Resonance a little more than on 15V but it's all within the normal throw of the potmeters so no problem there.

For this filter I used the 'Late MS20 Filter' schematic from Rene Schmitz which you can find by clicking here.
As he mentions in the text with his schematic, the gain of the opamps is hightened so you can get some weird sounds out of this. That's certainly true. :)
This filter is definitely different from other filters. It doesn't sound like the Moog Filter but it does have that 'ripping the fabric of the universe' synth sound and it's a real Speaker Ripper!. It's a 'Sallen-Key' type filter and it produces really divers sounds. (I always think of this filter as the 'Heavy Metal Guitar' of the synthesizer world.) You can get deep bass tones out of it and if you connect an Envelope Generator like the little AD/AR to the CV1 input you'll get a squarewave changing into a really bassy sinewave as the note progresses. It's awesome to experiment with this filter.
In HighPass mode the Cut-Off Frequency potmeter doesn't work over its complete throw. I found it usually only works over the first 50%. This is also true of the other MS-20 filters I built for the Dual version so this is normal behaviour for this design.
This filter is self oscillating in both the Low and High-Pass configuration. The more Resonance you give it, the more the two yellow LED's light up. I found that by using an LM13600 instead of a LM13700 you can  tame the filter a little. The 13600 seems to be a bit less aggressive although the difference is really small. Btw, instead of the TL074 you can also use the TL084 or the LM324. They are pin for pin compatible and work just fine. I personally tested them all successfully.

6dB vs 12dB.
Although this is a 12dB per Octave filter it is possible to get an output with 6dB per Octave roll-off if you tap the signal from pin 7 of the TL074 (A1 in the schematic drawing) with a 470nF capacitor, just like the normal output. The 6dB won't work in HPF mode because the signal is output from the stage before the one where we input the Highpass audio signal.
I have experimented with 6dB and at first I dismissed the option but having recently rebuilt my filter using (vintage) Polystyrene capacitors (and Polyester box caps for the 470nF output caps) it now sounds much better and the 6dB function sounds better too. So I made new layouts which includes the 6dB/Octave function with a switch to give you the choise between 6dB or 12dB. In the rebuilt filter I also used two BC558 transistors that were matched to within 4 points of their Hfe using a simple multimeter transistor tester. I'm not sure if this is necessary but it can't hurt. See the video below for a demo of the 6dB sound.

LM13700 pinout:
I took the part of the Rene Schmitz schematic that shows the CA3080's and I added the pinout numbers for the LM13700 chip which is just 2 CA3080's in one chip with added buffers which are not used in this filter. I'm not showing the whole schematic on my website because I don't want to draw visitors away from his website. It's Rene's schematic, not mine. Here are the pinout numbers:

Below here's the stripboard layout. I made my own version from the ones that are circulating on the internet to which I added the potmeter connections and the audio in and out, CV in, plus the switch connections for Low-, Band- and Highpass with the altered position of the 1nF cap at pin 12 of the LM13700. So with this and the schematics you should have all the info you need to build it right the first time. BTW, the 2 transistors used on the layout below are BC558 PNP's but you can also use the 2N3906 but those have to be put in the other way around.
Please note: I did add the Band-Pass option to the layout but if I were you I would just leave it out. But do some tests and decide for yourself.

(Last revised 1-March-2020 Corrected mistake with 470K resistor. 07-07-2020 added 6dB/Oct option.)

About the DPDT switch wiring for HP/LP mode: Connect the top two pins and the right middle pin together and connect that to the High-Pass input. You could even just forget about the top right pin and bypass it but I thought it was neater to include it. The lower right pin goes to ground. The audio signal goes into the middle left pin, and the lower left pin goes to the Low-Pass input. The Band-Pass switch is simply connected to the Low-pass input and the audio input. If the filter is in Low-Pass mode the BP switch won't have any effect but in High-Pass mode the switch will connect the audio to the Low-Pass input aswell so both inputs get the audio signal thus creating the bandpass characteristic.

Close up of just the strip-board lay-out (Print this one and use it for your build. The lay-out is guaranteed, tested and verified faultless). The green wire-bridges are coloured green because they indicate a connection to ground. Don't forget to cut all the copper strips under the IC's.

(Last revised 1-March-2020 Corrected mistake with 470K resistor. 07-07-2020 Added 6dB/Oct option.)

Bill of Materials:

About the components:
I used fairly cheap LM13700 chips I got out of China and they worked just fine. 
I used a DPDT toggle switch (Double Pole Double Throw) to switch between High-Pass and Low-Pass but you could also use a jack socket for High-Pass input with a build-in switch that connects C4 to ground when nothing is plugged in there.
The Cut-Off Frequency potmeter in this build is a 100K one, but you can use any type you wish because pins 1 and 3 are connected to the + and - of the power supply so it is nothing more then a voltage devider. I saw that Sam Battle uses a 4K7 for this in his layout so use what ever value you want. (Beware that the voltage difference over that potmeter is 30 Volts so don't go too low with the value or you'll fry your potmeter. Remember Ohm's law!)
I don't think the Resonance potmeter is that critical either but you better stick to the schematics for that one. Keep it a 100K potmeter. I used a logarithmic one but linear will work fine too. As mentioned earlier, you can use an LM13600 instead of the LM13700 and instead of the TL074 you can use the TL084 or the LM324. They all work just fine.
As I mentioned earlier, I rebuilt the filter recently and used polystyrene caps instead of ceramics. It did make a difference in my filter but the first version I built of this filter was a mess to look at. Building something neat and tidy always makes it work better I found out. So don't expect miracles by changing the caps from ceramic to polystyrene.
One thing that might be worth experimenting with is the 10K feedback resistor over opamp A3 in the schematic. The one by the LEDs. You could put in a 15K trimpot and see what it does if you change the feedback resistance. I haven't tried this yet myself but I'm going to and I'll report here if I get any results.

Here's a picture of the panel I made for it. Because I recently re-built the filter I had one unused mounting hole left in the panel so I put a white 3mm LED in there. It's a bit bright =)

Here's a picture of the insides. Those big brown ceramic caps at the top and bottom are just for de-coupling. They're spread out a bit because they're so big. I advise to use polystyrene caps for all the other capacitors although that's not necessary for the filter to work. I just found polystyrene caps to sound better but you mustn't attach too much importance to the use of polystyrene caps. Any type will do.:

The video below shows a demo of the filter after I had re-built it in July 2020 and added the 6dB per Octave option. The audio in this video is a bit crap and I'm just ploinking along the keyboard to test the sounds from low to high notes. (If I could play properly I would have done that, but there we are... =) 

Here's an other and older video demo-ing some  extra effects. In Low-Pass mode the Resonance Control has very little influence on the sound. In High-Pass mode it has much more effect but in this filter the Cut-Off frequency is what it's all about. That's what you use to get the cool sounds:

This is a test with the Korg filter (in Highpass mode) and the Moog Ladder filter (in Lowpass) in series and using my 8 step sequencer and reverb from the CaraOK effects module:

Sounds amazing doesn't it? Especially with added echo or phase-shift effects. It sounds very 'Giorgio Moroder' to me ^____^

WARNING: Beware your speakers!! This filter can oscillate at below audible frequencies and the cones of your bass speakers will take a hell of a beating if you've got some serious amplification going. If ever a filter could be called a 'Speaker Ripper' this one is it. Quite literally. (I added this warning because tonight I almost blew my speakers up with this filter.)

I want to direct your attention to a very useful page from Scott Stites' website. He talks about all the different aspects of this filter, using two of these filters in tandem and his approach to adding a Band-Pass mode to it. If you want to build this filter, you have to read this text I think.: Click here for Scott Stites website.

Lastly, here's an other very interesting document I found by Sound Semiconductor entitled "Designing Voltage Controlled Filters for Synthesizers with the SSI-2164."
It goes into great detail into how filters work and how to design them and places specific emphasis on the Korg MS-20 filter.

Okay, that's it for this one. I hope you enjoyed this and other articles and if you did why not leave a comment. I'd love to hear your thoughts or questions in the comments below. Any questions will be answered by me asap!
Right, see you on the next one!


  1. Hey
    I just built this thing. And I got really excited when it worked, but I soon realized that the cutoff effect decreases over time and after 2 minutes you can't hear anything changing anymore, although the resonance keeps working. Any idea why that could be?

    1. That is weird. I've never heard this sort of problem before. Could there be a faulty capacitor somewhere? I've built 3 of these and they all work fine.

  2. Just found this article after receiveing some PCB's I designed using the Rene Schmitz late Korg MS20 design. Thanks for the info, now I have to try and find some polystyrene caps.

    1. Thank you Leonardo. You don't "absolutely must use" Polystyrene caps. You can use those plastic Polyester caps or silver mica. As long as it isn't Ceramic.

    2. Thanks Eddy. On the hunt for some non-ceramic caps! Love the sound of your filter by the way!

    3. Thank you! Good luck with the build!

  3. This comment has been removed by the author.

    1. I'm sorry you're having so much problems with it. I wouldn't know what caused it. Maybe you should abandone this project and build the steiner parker filter instead.