Re: MS20 filter questions

[snk]

Hello,
It's been a little while since I built a circuit, and I'd like to give the Korg MS20 filter a try.
I will be using the 9V version from Tim Escobedo.
Everything seems good, except that I don't have any 100k resistor left... and i'm impatient 
I have 90k and 110k resistors, so I think I'm safe to use of of these values (or use 2 resistors in series), but before going for over-spec or under-spec values, I'd like to know what these resistors do in the circuit, and if it's best to use the regular spec, reach for an upper value (110k) or a smaller value (91k)... ?
Would anyone have an advice (try 110K? try 91K? wait to get some 100K resistors?)?
Thank you in advance

[antonis]

Use 110k for all 100k except for top right one..

[snk]

Thank you, Antonis!

[antonis]

You're welcome..

For further reading (as you can afford..)
https://www.timstinchcombe.co.uk/synth/MS20_study.pdf

[snk]

Thanks !
(A bit too mathematics for me, but very nice  )

[snk]

Ok, so I've built it (using 110k resistor except where advised, where I used two 51k in serial).
I used two 3mm red leds and a LM13700.
It works, but despite being in the ballpark, it doesn't really sound like an accurate model of the MS20 filter (I know there is two version of the MS20 filter) :

1- First, the filter doesn't go high enough (both in LPF and HPF). On a regular MS20, with the filter fully open, you can hear much more high frequencies pass through. Here, it sounds quite damped.
Is there any way to "shift" the filter frequency, or make it have a bigger amplitude and allows for more treble to pass when turned fully clockwise?

2- The resonance doesn't go into self-oscillation (which the Korg MS20 is famous for). It seems that this circuit tamed the resonance a bit. On a MS20, since you pass 12 o clock on the resonance, you get into gritty self-oscillation quickly.
Is there any way to increase the resonance so it could get into self oscillation more easily? Should I try with other leds, or is there a resistor taming things a bit?

3- It seems that bass is slightly cut at the input. With a guitar, it should be ok, but with a bass, the lowest octaves doesn't have enough presence.
Is it the 100nF cap at the input shaving bass range a bit? Should I try 220nF or 330nF?

[JustinFun]

That schematic doesn't look much like the MS20 filter to me.

This is supposed to be good - been on my 'to do' list for a while:

https://www.schmitzbits.de/ms20.html

[ElectricDruid]

1- First, the filter doesn't go high enough (both in LPF and HPF). On a regular MS20, with the filter fully open, you can hear much more high frequencies pass through. Here, it sounds quite damped.
Is there any way to "shift" the filter frequency, or make it have a bigger amplitude and allows for more treble to pass when turned fully clockwise?

I'm not surprised. The Rene Schmitz schematic linked above shows 100K used for the frequency control into Iabc on a +/-15V supply. The schematic you posted uses the same 100K value on a 9V supply, so the Iabc current will be proportionally less and have much less range - notably down, as well as up. Reduce the value of the 100Ks coming from the Freq control. Try 47K maybe?

2- The resonance doesn't go into self-oscillation (which the Korg MS20 is famous for). It seems that this circuit tamed the resonance a bit. On a MS20, since you pass 12 o clock on the resonance, you get into gritty self-oscillation quickly.
Is there any way to increase the resonance so it could get into self oscillation more easily? Should I try with other leds, or is there a resistor taming things a bit?

The Schmitz schematic shows *gain* in the resonance feedback loop: 10K/1.8K gives about x7. That's hard to fix on your schematic, unless you add some boost into that loop. There isn't any resistance you can easily tweak to improve it.
HOWEVER the notes under the 9V schematic claim that it does oscillate, and we have no reason not to believe them, so perhaps there's an error somewhere.

3- It seems that bass is slightly cut at the input. With a guitar, it should be ok, but with a bass, the lowest octaves doesn't have enough presence.
Is it the 100nF cap at the input shaving bass range a bit? Should I try 220nF or 330nF?

100n into 100K gives a HPF at 16Hz, so that's probably not the cause. I'm not sure what is. That said, increasing the input cap a bit certainly won't hurt and is easy to try, so why not?

HTH

[Mark Hammer]

I built  Motohiko Takeda's version of the MS20.  Nice compact PCB layout.  Good sound.  I can send you the PDF if you like.

[ElectricDruid]

I built  Motohiko Takeda's version of the MS20.  Nice compact PCB layout.  Good sound.  I can send you the PDF if you like.

The guts of that one are much like the Schmitz one: +/-15V supply, gain and clipping diodes in the feedback loop, standard PNP expo convertor driving the OTA Iabc inputs. The 9V version is heavily simplified, and it looks like somewhat to its detriment.

[snk]

Thank you all for your answers and your help!

Yes, this is a very simplified version of the original VCF (as the original, featured in a synth, works on a symmetrical power supply -among other things-), but I thought the differences were mainly related to the power supply, and not so much sound-wise.
I have a vintage Korg MS20 here, and the filter is much more full spectrum (frequency wise), with the ability to have an overdriven self-resonance past 12-o-clock, and all in all it makes thing "fatter", while this 9V version is much "harsher" (it goes into harsh saturation much quicker, not into warm overdrive). I wish I could get into this territory (while not expecting a 1:1 emulation, because of the lack of clipper and the simplified power supply).


I have found that the 100k/680pF combo was setting a cutoff point around 2.3kHz (which might be suitable for a guitar, but very far from the full range spectrum of the original filter)... I may either change the resistor (as Electric Druid suggested) or the capacitor : is one better than the other?
It seems that with a 220pF cap, I'd go up to 7.2kHz, which should be great...


About the leds, it's 5mm green on R Schmidt design, and 3mm on the T Escobedo version... I understand they have a different forward voltage, but I don't think that it makes such a big difference, right?


About the resonance, would decreasing the value of the 10k resistor to ground (going from pin 9 of the OTA to the filter pot) change anything?
(On the R Schmidt design, I have read that there is a 1.8k resistor which sets the max resonance)

[snk]

I built  Motohiko Takeda's version of the MS20.  Nice compact PCB layout.  Good sound.  I can send you the PDF if you like.

Thank you very much for the kind proposition, Mark!
Yes, why not?

(I didn't want to go into a full project with a symmetrical power supply -else I would have chosen an eurorack version-, but maybe that's the way to go if I want a better sound?)

[snk]

I'm not surprised. The Rene Schmitz schematic linked above shows 100K used for the frequency control into Iabc on a +/-15V supply. The schematic you posted uses the same 100K value on a 9V supply, so the Iabc current will be proportionally less and have much less range - notably down, as well as up. Reduce the value of the 100Ks coming from the Freq control. Try 47K maybe?

Ok, thank you for the explanation : I was expecting less headroom, but not a narrower frequency spectrum.
Do you advise to reduce the resistor value instead of the 680pF capacitor?

The Schmitz schematic shows *gain* in the resonance feedback loop: 10K/1.8K gives about x7. That's hard to fix on your schematic, unless you add some boost into that loop. There isn't any resistance you can easily tweak to improve it.

Too bad. Well, it's good already (it resonates somewhat like a dirty version of a Roland synth, which is fine)... but soundwise it is far from the infamous MS20 scream with overdriven self oscillation.
I guess from your answewr that lowering the value of the 10k to ground next to the resonance pot wouldn't change much the behaviour?

HOWEVER the notes under the 9V schematic claim that it does oscillate, and we have no reason not to believe them, so perhaps there's an error somewhere.

Yes, I'll check back the circuit.
I have read that the input volume/drive/gain can also affect a lot the resonance, so maybe that's it?

100n into 100K gives a HPF at 16Hz, so that's probably not the cause. I'm not sure what is. That said, increasing the input cap a bit certainly won't hurt and is easy to try, so why not?

Yes, that's strange... could it be the consequence of using a 9V circuit (instead of a +/-15V)?

Thank you a lot for all these answers !

[garcho]

Q&D is Quick & Dirty!

The 9V version is heavily simplified, and it looks like somewhat to its detriment.

Certainly. Most of Escobedo's "snippets" are starting points for experimentation, not fully troubleshooted optimized guitar pedal circuitry, so it makes sense that it might need some finessing and experimentation. It can be deflating if you assume they're totally finished products and when you build them, they don't sound "right". Luckily, there's probably a thread here on the forum for anything he ever posted (and most of those circuit snippets are from 20 years ago!).

I didn't want to go into a full project with a symmetrical power supply

You can use a LT1054 for a Q&D bipolar 9V power supply, to go along with the Q&D filter. It's significantly simpler (read: no mains wiring) than getting a 12/15V bipolar PS going, as you mentioned (eurorack). And way cheaper.

Do you advise to reduce the resistor value instead of the 680pF capacitor?

The most important thing to keep in mind when changing the voltage is the current going to I_ABC on the OTA. You can fry them pretty easily compared to most ICs!

Also, clean, dynamic signal input for LM13700s can be tricky, as well as getting the thing to stop &%$ing hissing.

[Mark Hammer]

Here's the PnP layout, although it may not be to scale.  Don't know how to post pdf files here, and I think Motohiko has taken the site down, so I can't link to it.  You can PM me with your e-mail address and I can mail it to you.
I used red LEDs.

[snk]

Do you advise to reduce the resistor value instead of the 680pF capacitor?

The most important thing to keep in mind when changing the voltage is the current going to I_ABC on the OTA. You can fry them pretty easily compared to most ICs!

Sorry, english is not my native language and i don't have many electronic skills so i am not sure to understand (i can't "keep in mind" something i don't know yet  ) : are you implying that changing the capacitor value (instead of changing the resistor value) would impact the voltage going to the OTA, risking to fry it?

[ElectricDruid]

The Schmitz schematic shows *gain* in the resonance feedback loop: 10K/1.8K gives about x7. That's hard to fix on your schematic, unless you add some boost into that loop. There isn't any resistance you can easily tweak to improve it.

Too bad. Well, it's good already (it resonates somewhat like a dirty version of a Roland synth, which is fine)... but soundwise it is far from the infamous MS20 scream with overdriven self oscillation.
I guess from your answer that lowering the value of the 10k to ground next to the resonance pot wouldn't change much the behaviour?

Reading the Tim Stinchcombe paper, it's pointed out that the gain-with-clipping-diodes in the feedback path gives a very specific behaviour: the gain is higher for lower voltages, and then decreases as the signal level rises. This significantly affects the sound of the resonance, and no gain and no clipping is not going to give you the same sound. If I was tweaking, I'd put that part back in. It's easy to run an op-amp at 9V and overdrive circuits are what we know best, so sticking an overdrive in the feedback path of the filter should come naturally!

The 10K to ground next to the resonance pot is just a cheap I-to-V convertor. The output current of the OTA is taken to ground by the resistor. If the resistor has a current of X flowing *through* it, then it must have a voltage of Y *across* it, so then we can tap the voltage off the top of the resistor for the rest of the circuit. It doesn't really have anything to do with the resonance circuit. The same thing appears at the output of the first OTA stage, notice.

HTH

[snk]

Thank you

In the meantime, I made further tests and measurements.

I will try with an overdrive prior to the filter to see how it goes, and then I will go back to the soldering table to change either the resistor or the cap value to shift up the max frequency.
I might also try with 5mm green, blue or yellow leds (instead of the 3mm red I am currently using).

With an output volume and the resistor (or cap) new lower values, I thinkit will be good 
My only concern will be to find a way to have less resonance at minimum setting, and more at maximum settings (but as advised, I'll play with input gain/overdrive and leds to see how it goes).

[ElectricDruid]

Changing the cap and the resistor aren't completely equivalent. Changing the cap shifts the whole range up or down. Changing the resistor changes the *size* of the range. Since the voltage out of the pot goes down to zero, the minimum current into Iabc will always be zero (which may not be ideal - there's no other frequency offset trim on this circuit either). The maximum current will be determined by the resistor value and the maximum pot voltage of 9V. It can't exceed 2mA or you'll fry the LM13600/13700, but otherwise you can experiment. R=V/I, so 9/0.002 = 4500 = roughly 4K7. Actually the situation isn't quite that bad, since the Iabc input is a couple of diode-drops above ground, say 1.2V. That means the 9V input is actually only 7.8V above Iabc. So 7.8/0.002 = 3900 = 3K9. So you could *probably* go that low safely, but it's cutting it very fine and I'd use a 4K7 to save any unfortunate accidents!

[snk]

Awesome, Electric Druid, thank you so much for the insight!

I noticed that the beginning of the pot rotation doesn't produce much effect, so maybe I can try to shift the whole range a little bit, and also narrow the size of the range slightly. I will experiment, I have everything I need to know
(according to this very nice website of yours, it seems that 470.0 pF / 68.0 KΩ would give 5.0 kHz)

In the meantime :

- I noticed that the layout I was using didn't feature the connection of the resonance pot to Vref : I made the wiring according to the schematic, and now the resonance is a better.

- I also found the first build of this filter I did a while ago (and lost for 2 years !), using another layout, and a 13600 (instead of 13700) and blue leds : so, I could compare them
The "early" build from 2 years ago with the blue leds sounds quite different (it is a bit more "ballsy" while distorting, and it self-resonates a lot). I will try blue leds on my new build, and may swap the 13600 and the 13700 to see if it changes anything.