Will the real MS-20 filter please step forward

[Cliff Schecht]

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?

Stability would not be an issue. There are two options here WRT powering off of 9V. You can either bring in 9V and run it through some sort of switching supply(ies) to convert it to +/-12V or +/-15V. Or you can redesign the circuit to work off of a lower voltage and single supply. IME the second one can neuter a filter, you need a certain amount of headroom for the resonance circuit to work right especially. Also reducing the supply can in general change how the effect sounds because all of the currents being sourced to the OTA's are dependent on the supply voltage as well (meaning changing the supply voltage will change the cutoff and resonance). I'd still like some switchers onboard but I looked into what is available off the shelf and a flyback converter would probably require custom magnetics (not cheap, unless someone buys a bunch in bulk).

[teej212]

what about charge pumps?  Im not too educated in the area but if you got a voltage boosted from 9v to 30v you could make it a +15/-15 v from 9v, no?

[hoyager]

Talking psu and charge pumps, here's the converter from the lt1054 datasheet on stripboard, with zener regulators, which works well, but *maybe* doesn't have enough current for a 2 filters?




Andy

[Cliff Schecht]

Not quite. What you need is a charge pump that goes from 9V to 15V and another one that either goes -9V to -15V or you can run a simple inverting charge pump to change +15V to -15V. But you can't (easily) make +/- voltage from a 30V supply, or better stated you wouldn't want to.

As I stated earlier the problem with charge pumps is they are unregulated voltage. The output voltage is a function of the output current, meaning the output voltage varies with the amount of current the circuit is drawing. For OTA based filter circuits I think what we will find is that the filter starts to drift worse and worse as you push the cutoff frequency higher and higher. As you go higher in cutoff frequency more current has to be driven into the Iabc pin of the OTA which means that the biasing circuit will also have to draw this additional current. As the supply voltage drifts, the overall bandwidth of the filter will change as well, not something I want to unintentionally add into any filter circuit.

I still hold that the best solution is two switching power supplies or even a single flyback supply. I can design them with easy to obtain parts and off-the-shelf magnetics that would make these no more difficult to build than your run of the mill charge pump circuit. To be honest it annoys me that people are so adamant about NOT using switching power supplies when they usually are the most viable solution to the problem (especially in this case where we want clean, regulated power!). I want DIY'ers to get over their SMPS-phobias and start applying these circuits instead of bulky, unregulated charge pumps. I can understand people not liking SMD parts but most any SMPS that we would need can be made from one of a handful of general purpose controllers (MC34063 or UCC3804 for example) and a small range of inductors (anywhere from like 10uH to maybe 1mH) depending on application and current draw).

[Cliff Schecht]

Sorry but that charge pump circuit looks pretty massive. The power solution should not take up as much real estate as the circuit itself IMHO...

With that said, another usable option for us would be a 9-15V boost power supply and a charge pump to derive the negative supply rail. The negative supply in these circuits usually draws much less than the positive one and if you run an inverting charge pump off of the +15V, you will at least have a bipolar power solution where the plus and minus voltage track each other decently well.

[hoyager]

Cliff, what you've said is exactly what happens with the above charge pump circuit. I use 18v zeners and the voltage seems to stabilize at about 16v, but this is running just 1 filter. I'm keen to see a design using the 34063 which would be more stable. I've tried the double chip boost and invert but unless I made an error somewhere seems to have the same problem of voltage drop with surrent draw, but worse.

[nelson]

In the LT1054 datasheet see application 30 and 31 for both a regulated positive and negative voltage doubler.

With the V L of the LT1054 and diodes you should be able to get regulated +/-15v from a 9v supply.

There are also Push Pull PWM IC's like the SG3524. There's an app in the datasheet for a relatively simply flyback +/-15v supply.

[Jordan A.]

I'm going to reverse myself a little, with the clipping diodes I'm liking two 1N914s in series (in both directions) better than just one 1N914.  An LED with a similar voltage drop would probably give similar results.  This is really an area where people can pick whatever is their preference, just reporting my findings.

Cliff, or anyone who has done work with this, I would be exited to try out any power supply option you cook up if you can knock out a quick schemo, as long as I can find whatever I need at mouser/digikey.

jordan

[Cliff Schecht]

Yeah more diodes in the negative feedback amplifier circuit means more negated amplitude is sent back to the filter which will give you less resonance overall (if I understand what you are doing correctly).

I have figured out a couple of viable power solutions for this using off the shelf components. I'm thinking either a flyback controller where we have users wind their own coupled inductors from purchased/scavenged toroids (this is a DIY forum, right? ) or use a coupled inductor on an inverting regulator to generate both +/- voltage. I'm thinking the second option both because it offers a very compact solution and it uses only off the shelf components. Everything for that can be bought from Mouser for ~$5.

[hoyager]

Regarding clipping diodes for the resonance loop, I've tried alot of combos, and keep coming back to 2 zeners, but in standard clipping configuration, so about .5v forward voltage. 2 leds with a 1n34a on each end also sounds really good, but makes the self oscillation too loud. With the 2 zeners, or possibly any silicon diode, the res only takes over when the input level is backed off, and is about the same volume as the input signal was, which makes the filter much more usable imho.

[garcho]

here's a link to Thomas Henry's ±15v power supply, if anyone is still interested.

[Cliff Schecht]

That schematic is the most common method for deriving split supply rails from a single wall supply. We've discussed that circuit earlier and decided that we'd like to power this pedal from the same supply as any other 9V negative center pedal would take. I'd still like to at least do a version with a proper switching power supply (or set of them). If you sync the oscillators between the two chips then you can eliminate the possibility of them crosstalking and sending either supply off the deep end. Also with a well designed filter(s) the switching will be nearly invisible to the audio circuit, at least in the sense that it will never be heard by the human ear. The real trick here is to try guarantee stability and at least decent regulation under all operating conditions while making the supply easy to implement with inexpensive off-the-shelf through hole components.

TBH I've set this project aside for now just because this thread has been pretty dead. I have been toying with the power options and have a few I still need to try out, but nothing that will take that long I don't think. If Strategy says that he's ready to move forward I will start working up the schematics and boards. I think we should make a couple of versions to start with with different power supplies and LFO's to see what set of features we really want for guitar.

[garcho]

i'd love to start breadboarding, sorry i can't help with much

[Strategy]

i'd love to start breadboarding, sorry i can't help with much

Sorry folks work has kept me away from the forum. The Dorkbot-pdx circuit board orders are twice monthly so as soon as we have layouts to test I'm happy to do a test run of 5-10 boards and distribute them to the group! As for the power options I like the idea of being able to get people powered up from typical 9V like Cliff is suggesting.

I realize I need to weigh in on the many modulation options. I'm going to go back through the thread and respond definitively. The number of options are a little overwhelming to me.

Strategy

[Strategy]

As for LFOs, I'm thinking that if we give one LFO per filter that is a LOT of modulation by comparison to other pedals, which usually give one LFO with numerous waveshapes. Separate modulation of the two filters will allow some of the formant/"yoy" effects mentioned. As for waveshapes, do we want to have switchable between sine/square/tri? Gristleizer and some other circuits offer ramps but with two filters being modulated I think the nuances of ramp will be lost amongst all the "mangling".

What I like about this is that it does something that an unpatched MS-20 can't do: more than one LFO, one for each filter. On the original MS-20 you have one LFO for everything, oscillator, filter, etc., which is one of the real limitations of the synth. So in that way, a 2-LFO pedal, even if limited in terms of number of waveshapes, is offering up some possibilities beyond the original synthesizer.

This seems like a good balance between the minimal and maximal proposed modulation options. I still don't understand how sync/unsync would be implemented or whether its desirable. I don't have any devices with synced LFOs for comparison.

Paul

[Strategy]

One thing I'd like to suggest is that the CV inputs for filter cutoff freq and resonance should be in parallel with the onboard LFOs and should not deactivate the LFO's via switching. That way you can have the LFO's running AND pipe in additional CV's, for very complex modulations.

Strategy

As for LFOs, I'm thinking that if we give one LFO per filter that is a LOT of modulation by comparison to other pedals, which usually give one LFO with numerous waveshapes. Separate modulation of the two filters will allow some of the formant/"yoy" effects mentioned. As for waveshapes, do we want to have switchable between sine/square/tri? Gristleizer and some other circuits offer ramps but with two filters being modulated I think the nuances of ramp will be lost amongst all the "mangling".

What I like about this is that it does something that an unpatched MS-20 can't do: more than one LFO, one for each filter. On the original MS-20 you have one LFO for everything, oscillator, filter, etc., which is one of the real limitations of the synth. So in that way, a 2-LFO pedal, even if limited in terms of number of waveshapes, is offering up some possibilities beyond the original synthesizer.

This seems like a good balance between the minimal and maximal proposed modulation options. I still don't understand how sync/unsync would be implemented or whether its desirable. I don't have any devices with synced LFOs for comparison.

Paul

[garcho]

That way you can have the LFO's running AND pipe in additional CV's, for very complex modulations.

i like that.
i also much prefer 9V, just didn't read the thread close enough. i'm super excited to start playing guitar through something like this...

[Jordan A.]

I feel slightly responsible for cluttering up the discussion with too many options too fast, I like that we're getting back to practical territory. 

Any of you guy's can start breadboarding these things, if you don't have access to a bipolar power supply you can always just use two 9 volt batteries, it works at +/- 9 volts, definitely well enough to give you an idea of it's capabilities.

If you need ideas for LFO circuits I can try to do a quick pen and paper sketch (if I can manage to figure out how to post pictures here), pretty much any LFO that swings a few volts and has a comfortable frequency range will work well.  I think a good old two opamp integrator/comparator LFO is the ticket.  This is an (approximately) 1 volt per octave filter, so an LFO that is 10 volts PP (for example) will modulate the VCF through 10 octaves when the modulation depth control is all the way up, which is more than enough range, IMO.

I'll try to draw up a proposal for an LFO mode switch, that I think is very nice.  The option to sync (and reverse) the sweep of the VCFs is very important, I think.  When you get it breadboarded up I think you will see what I mean, with two free running LFOs the modulation is a bit "wandering", for sure a cool effect but it's nice to be able to get a more "focused" sound when you want it.

ok!

jordan

[garcho]

Any schematic anyone recommends more than any other? To start I'll use a simple LFO and 2 9v batteries.

I'm excited to eventually use the TAPLFO with this project.

[Jordan A.]

If you have the time you could try both of these -

http://www.jhaible.de/jh_720_vcf.gif

http://www.uni-bonn.de/~uzs159/

I've been mostly messing with the Rene one, but i do remember breadboarding the Jurgen Haible one a while ago.  I'm not sure how Jurgen's will do at +/- 9 volts, but I know the Rene one will work.

Jurgen's has the benefit of being really tiny and cheap to make, and it will probably work with just about any NPNs, PNPs and opamps you have.

The tapLFO would absolutely work well.

jordan