go to Japanese Version

MS20 Clone VCF 2

We Love MS-20!

This is prototype.It's so small.I use LED as
clipper diode
The first version of MS-20 were designed under low-cost strategy in 1978. And it don't use OTA at the VCF and they use transistor as dynamic resistance. It was in the custom chip called KORG35. I heard that it is trouble for repair men of good old musical instruments.

The VCF which was adopted with original performance synthesizer that Jugen Haible made himself is based on early MS-20's VCF. (His page is here and his performance synthesizer called JH-720 Solo Synthesizer is here )

And there is another interesting story. Martin Czech (His home page is here) who lives in German as same as MR.Jugen Haible, mailed to the original designer of MS-20 and ask the schematic diagram of KORG35. KORG gave it him with some limitation .
there is the whole story about it. What really amazing is the attitude of Korg towards us DIYers.


It's really small. The limter of this version is just
small signal switching diorde

I add the trim for adjust amount of resonunce .
I learn about sound making from SteinerVCF. and HPF is good to add something special to the sound , But there is no need for HPF with basic sound making. So I omit them for the PCB space. If you need it, You can add it easily. If it is 2nd filter for you, you should add HPF on it.
I layout Q2 and Q3 to make them easy to paste with, this will keep them same Temperature. I use Japanese transistors and their pin structure is different from overseas one's. Check it out please, But if you use Japanese one then, this layout allow you to pasete them easy.
You may use enough epoxy glue to cover them all to keep their temperature as same. But there is already plastic mold so I think that it's enough. I pasted them tightly to make them easy to exchange their heat. And I made a small cap with plastic sheet with air. Air will be most easy way to shut out the changes of the temperature.

I made a small cap with plastic sheet with air
I select out the transistor with the value of Veb. I use cheap DVM. When you change it for maintenance then you should change both, So molding is no matter for that.

I use 2SA1015 for PNP, and 2SC1815 for NPN. In my experience, when you can choose the rank of hFE, you should better choose lower hFE one. It's more easy to select for Veb than higher one. If you get 10 GR rank (its hFE is 350-700) transistor, you can get 2 or 3 pair but if it's Y rank (its hFE is 120-240), you can get 3 set of 3 pair or so.
I learn article of MR. Tanaka's page at (* TRANSISTOR *), the current are flowing from 2nd TR to 1st TR, and it get changes dynamic resistance of both one. so in order to offset each other we should change the value of C. MR. Jugen Haible also change its values.
MR. Tanaka said that it's different for 3 times. So to fit each cut off frequency, we should change their values.
In this schematic diagram, 10n and 33n of C are used, nevertheless should be 10n and 30n. There is already 10% error. and capacitors are always have errors of their value. In the other hands, errors of the transistor, 1mV difference of Veb makes 4% difference of Ic. and it's makes 4% error of dynamic resistance. (MR. Tanaka teach about this directly for me. Thank you so much!)
If you want it to be correct, You should get 4 equal valued capacitor and use it as series to get correct value and next, select transistors. I think that heat combination is good for usual operation for this filter. But to fit each cut off frequency, there is so many things to do. And it's not easy.
In fact, when I build it, I fried my Q2 by mistake. and I put another kind of NPN there. It worked. I think it will cause something trouble when you make sound carefully.
Now MR. Jugen Haible build PolyKorg Clone. And he use another schematic for VCF. It's more closely to original KORG35. And He said that the transistor of both VCF and VCA is selected by the value of VEB, not hFE.

If there is difference between cut off frequency of the poles in the same filter, they can't make self-oscillation or wave form that produce by self-oscillation may be distorted.
And the other hands, positive feedback filter will get big volume when it gets self-oscillation. So you should better set it as not to get self-oscillation.


To get 1V/Oct responce, change the attenuator of CV inputs. In this schematic diagram, you should change R11 to 1.8k. But I realize it will fit in 2.2k. I don't know why there. There is some 100k in parallel for summing stage and it makes something there. I will stady it for later. I will report here my result of cut and try.

I ask MR. Jugen Haible, what's Trim1 acting here? And he said that it's basically a limiter for low cufoff frequencies, and not to VCF going into the sub audio range. He is adjusting it by his ear.
In this part, The signal flows from emittor follower by PNP Transistor to NPN Transistor, and it forms like ARP's matched pair antilog amp styles. (Its actual action is so complex and it's so hard to understand)

Trim2 will adjust self-oscillation. In my case, there is no self-oscillation at high frequency, but when it's going down then it get self-oscillation. so in this case you should better adjust it in low frequency area.
There is another way to controle self-oscillation. You fix the rate of amplitude of opeamp and change tht value of R13. It forms attenuator with R14 and C2.
Frequency will be sweeped by CV and if this tuning is bad, it's going to be so hard to make sound. But these trims should be tuned as you like! It's the most fun (and also hard to do) part for D.I.Y.
The value of C2 and R14 is also play inportant part for self-oscillation. I change this values, But I can't find how it works here well.

I build this twice. In my first case, I use Transistor as Y rank (It's mean that its hFE is 100-200). and I use more higher one for the other one (it's 521 or so) and each Veb is 572mV. Higher hFE version can't suppress self-oscillation by 220k+100k(TRIM).
You should better change the value of 220k to 180k for the value of hFE that your transistors have.
You should better add 2.2k at the resonunce VR as MR. Jugen Haible did. I think that if you omit it, it's going to be so hard to tune and hard to oparate.


If I try to check them by blind method then I can't find which is which. It's different schematic diagrams but they have same sound charactors. Dual OTA version may have more clear and bright sound.
In the music tune, I always add the process to my sound "yogoshi" for fitting with another non synth sound. "yogoshi" means "blur" or "colly" in japanese. So I think that bright sound charactors are not necessary condition for my synth.

In this schematic, it may have many feedthrough. But if you add coupling capacitor at the end of the circuit then it's going to be under order of 1mV.
There is feedthrough. But if you put it into the music tune, no one cant find it

The linearity of cut off frequency is not better than Dual OTA version. You can play melody by Dual OTA version but in this version, it's hard to play it
If you add trim and tune it finely then you can paly several octerves, I think.

It's cheap, compact and good sounds. it's good! but there is some weak point.
the charactor of transistor comes so directly. the value of hFE or Veb will change the charactors of whole filter's charactors directly. If you want to use it as filter for polyphnic projects, then you should better get all the parts at onec to arrenge the charactors of transistors.
In the case of the ONE ONLY filter, you can get the good sound easily but it may be so hard to tune.

In this report, MR. Tanaka and MR. Jugen Haible gives so much advice for me as I write down in page. Thank you so muc!

Back to the HomeMadeSynth MENU(Japanese Only)
back to the Takeda's home page (Japanes Only)

"MS20 Clone VCF 2" Copyright 2002,2003 Motohiko Takeda, Crow Hill Laboratory
Motohiko Takeda Mail:takeda@aleph.co.jp