MS-20 filter fun!

[moid]

Picture time! First one is the circuit I currently have - the power and ground of the LFO circuit are a differnt power and ground to the MS20 circuit. I cannot get the circuit to affect the MS20 if it shares ground with the MS20, and this is a big problem I suspect.


I had a go at trying to make Duck's sawtooth mod - so this is almost certainly wrong! I'm now thinking the switch and the pot swap places?


If either of you have any idea how to fix the power / ground issue that would be amazing. Thanks.

[ElectricDruid]

There are various things I can spot:

The MS-20 filter Frequency control is wired up wrong. The ends go to 9V and 0V, and the signal to the circuit goes from the wiper. You've got the wiper tied to ground!

The Triangle LFO out needs to go to the same place as the Freq pot wiper. You've got it wired to one end of the pot.

The variable-symmetry mod won't work because the diodes and new pot need to replace the LFO's 20K rate pot, not be placed on the LFO output.

The power/ground problem is probably a result of some of this mis-wiring. There's no reason why they shouldn't work on the same power supply, both power and ground. There might be the common "ticking" problems you get with LFOs, but that's a different issue.

HTH

[duck_arse]

see the parts between a and b. RV1 and the stop resistor [remember? stop resistors on pots?] determine how fast the timing caps can charge, and then how fast they can discharge. so you switch in a diode, and whoosh, that direction sees "no" timing resistance. reverse the diode via the switch, and it's the other direction that goes whoosh, instead. so, sawtooth forward, sawtooth reverse.

but if you add another pot, let us say 10k, between the 'a' and the switch common, you can program the extent of the whoosh-ness, from 0 to some, because you are slowing down the diode, as it were.

obviously, with the pissy output level for the triangle, you wouldn't use my resistor values for R1 and R3. you can connect a rate led and CLR from "sq out" to ground, but it will flash square-to-pulse, not ramp.

[moid]

Thanks Tom - I was going to say, well that's what I did last night and nothing happened, but I did it again today... and it worked Probably lots of random cables held together with prayers/curses and crocodile clips and one wasn't attached that well.  So both circuits are running off the same battery now and seem happy. A couple of odd issues - at the moment I am swapping capacitors in and out of the breadboard to test the circuit, and when I do that the speed pot on the LFO seems to get stuck - the circuit just keeps the speed at whaever it was before changing the capacitor (regardless of what I set the LFO speed to). Presumably this won't be an issue once that is all connected to a switch? I currently fix it by twiddling the Frequency and Resonance pots and at some point it starts working again! The audio does seem to sometimes cut out when some of the pots are at extremes, but altering them a little brings it back - that's the circuit I guess. The high pass option is a lot quieter than the low pass option - that's a niggle, but I doubt much can be done with that unless I chuck a boost circuit on there as well... probably going to be hard enough to squeeze this into the enclosure as it is.... hmmm I do have a small booster circuit left over though....a DOD one... hmmmm too many options! I will wait to see Duck's Sawtooth idea first


I just realised I didn't test the circuit by sticking an SPST betwen the LFO output and the wipe of the frequency pot so I can deactivate the LFO if I want a c o cked wah, but that should be fine.  I'll start making a vero layout and new schematic. Thanks again

[ElectricDruid]

So both circuits are running off the same battery now and seem happy.

Nice! That sounds like a big improvement.

A couple of odd issues - at the moment I am swapping capacitors in and out of the breadboard to test the circuit, and when I do that the speed pot on the LFO seems to get stuck - the circuit just keeps the speed at whaever it was before changing the capacitor (regardless of what I set the LFO speed to). Presumably this won't be an issue once that is all connected to a switch? I currently fix it by twiddling the Frequency and Resonance pots and at some point it starts working again! The audio does seem to sometimes cut out when some of the pots are at extremes, but altering them a little brings it back - that's the circuit I guess.

All of that sounds like "breadboard gremlins" to me. Dodgy connections, basically. Dealing with stuff like that when breadboarding was why I largely gave up and now build stripboard prototypes instead.

The high pass option is a lot quieter than the low pass option - that's a niggle, but I doubt much can be done with that unless I chuck a boost circuit on there as well...

In theory it should be the same volume, so there shouldn't be a need for a booster. In practice, if you set the cutoff frequency somewhere in the middle of the audio range and then switch back and forth between lowpass and highpass, you'll typically find the highpass is quieter. That's simply because most musical instruments have the najority of their energy in the fundemental and the lower harmonics, and not so much energy higher up. With highpass, that's the bit that gets removed, so you finish up reducing the volume a lot. It's not that the highpass output is actually quieter, but just an effect of the signal you're feeding it and the filtering action you're applying.
To hear the effect in reverse, set the filter frequency fairly low, and then feed in some hihats. On the lowpass output you'll hear basically nothing - very quiet indeed! On the highpass mode, the cymbals will be pretty much unaffected by the filter, since all their energy is above the cutoff.

HTH

[moid]

Good Lord that Mr Duck he draws fast! Faster than his own shadow I've heard it be said. Right well I need to put my thinking cap on when I goes a-lookin' at one of Duck's special drawings... they do funny things to my head you know. Thanks Duck, I will have a read, and a think, and then ask a selection of sensible and daft questions. Might be a day or two; I've got to take my son to London tomorrow so that's a going to be a very long day. Hmmm looks like I guessed everything wrong bar the diodes running in opposite directions. I should really stop using a ouija board to predict circuit connections...

As a prelude to that, here are some of my own fair pictures. A revised schematic for the LFO as corrected by ElectricDruid


and a vero layout of it


Hopefully those are working? Please yell if anything looks wrong with them. I know the breadboard version works, so if I can't get Duck's slippery slanted slopes sorted out, I know that the above is functional.

[ElectricDruid]

You've broken the connection between the pin 7 output and the 8K2 resistor. The rest of it looks good so far.

[moid]

All of that sounds like "breadboard gremlins" to me. Dodgy connections, basically. Dealing with stuff like that when breadboarding was why I largely gave up and now build stripboard prototypes instead.

I don't think my ability to work with circuits would allow me to do live prototyping on vero board sadly... when I did fix the breadboard version of the circuit I also had an LED attached to the LFO out (hoping to see it flicker like a rate LED) but it blew up with the power coming from the same source as the MS20... ouch.

The high pass option is a lot quieter than the low pass option - that's a niggle, but I doubt much can be done with that unless I chuck a boost circuit on there as well...

In theory it should be the same volume, so there shouldn't be a need for a booster. In practice, if you set the cutoff frequency somewhere in the middle of the audio range and then switch back and forth between lowpass and highpass, you'll typically find the highpass is quieter. That's simply because most musical instruments have the najority of their energy in the fundemental and the lower harmonics, and not so much energy higher up. With highpass, that's the bit that gets removed, so you finish up reducing the volume a lot. It's not that the highpass output is actually quieter, but just an effect of the signal you're feeding it and the filtering action you're applying.
To hear the effect in reverse, set the filter frequency fairly low, and then feed in some hihats. On the lowpass output you'll hear basically nothing - very quiet indeed! On the highpass mode, the cymbals will be pretty much unaffected by the filter, since all their energy is above the cutoff.

HTH
[/quote]

Thanks that makes total sense - when I look at recordings I've made in Adobe Audition I can view them as spectrograms and the brightness of colour there reflects the volume of the frequencies; the bass and mids are disproportionately brighter (and therefore louder) than anything over 2K and upwards - some instruments make some loud high frequency harmonics, but as you note the way that MS20 filter works cuts stuff quite hard, so guitars really loose their volume at that point. I'll leave it alone; both my son and I prefered the low pass effect most - I will leave the highpass as an option because I'm bound to want it for something in future

[moid]

You've broken the connection between the pin 7 output and the 8K2 resistor. The rest of it looks good so far.

D'oh! Thanks for spotting that one... OK I think the below is fixed... I was congratulating myself on making such a small vero layout as well What's another row between friends? It's not that big (that's what she said anyway)

[moid]

Oooh I really should go to sleep but you've got me all fired up with questions and thoughts and urges... so here goes

see the parts between a and b. RV1 and the stop resistor [remember? stop resistors on pots?] determine how fast the timing caps can charge, and then how fast they can discharge. so you switch in a diode, and whoosh, that direction sees "no" timing resistance. reverse the diode via the switch, and it's the other direction that goes whoosh, instead. so, sawtooth forward, sawtooth reverse.

I think so... the stop resistor set a minimum amount of time / resistance to that part of the circuit? The diodes altering the direction of sawtooth makes sense ( I think), although how the two 10uF back to back capacitors work is not something that makes sense; I thought capacitors always had to be positive leg to negative leg? You've got them as positive to positive? And if I wanted my two different speed ranges, would I need a pair of 470uF caps in the same configuration and a SPDT switch to decide which pair are in the circuit?

but if you add another pot, let us say 10k, between the 'a' and the switch common, you can program the extent of the whoosh-ness, from 0 to some, because you are slowing down the diode, as it were.

So that pot would alter the slope of the sawtooth wave? so from triangle to hard sawtooth, while the diodes decide whether the sawtooth is reverse or forward facing?

obviously, with the pissy output level for the triangle, you wouldn't use my resistor values for R1 and R3. you can connect a rate led and CLR from "sq out" to ground, but it will flash square-to-pulse, not ramp.

OK so I need much smaller values for R1 and R3? I'll start with 56K and 10K respectively and see what that does. I did try measuring the LFO out of my circuit and on DC Voltage it says 6.24V... now I'm suspecting I should've tried AC Voltage? What sort of Volts measurement would sound like a good one to you?

Thanks for the rate LED - even a squarewave LED pulse is useful!

I've got more questions (sorry)

The two diodes and switch section. How do they attach to the rest of the circuit? does the X near "a" connect to the X at "sq out"? Where does the X at B go?

VCC/2 that is the output of a voltage divider from 9V yes? So 4.5V?

I presume I need to add in the +9V and ground conection to the LM358 chip as I have before?

SW1 - is this an SPDT of the ON / OFF / ON variety? I've got a few of those.

If I don't have a C100K pot, can I use a normal 100K lin pot instead? Not a problem if it works best with C100K pots; I need to make an order of parts for other stuff I'm running out of so I can get some more of those at the same time.

Thanks again for the temporary loan of your grey matter

[ElectricDruid]

obviously, with the pissy output level for the triangle, you wouldn't use my resistor values for R1 and R3. you can connect a rate led and CLR from "sq out" to ground, but it will flash square-to-pulse, not ramp.

OK so I need much smaller values for R1 and R3? I'll start with 56K and 10K respectively and see what that does.

It's more a question of the *ratio* of the two resistors, rather than the values. Scale both down by ten and you won't change anything much. Scale *one* of them down by ten, you'll see a big change.
They set the point at which the schmitt trigger flips the triangle wave from going up to going down or vice versa. If you set that point low, the triangle wiggles about rapidly at low amplitude. If you set it much higher, it slows the triangle down because now it has to charge up (or down) for a lot longer to reach the trip point again. But you get a bigger amplitude. So adjust those resistors to get the *amplitude* you want, and then go back and change the timing cap and pot to get the *frequency range* you want again, because if you mess with the trip point, you'll change the overall frequency range of the whole LFO.

[duck_arse]

moidy moidy moidy - do as we all do - throw all your parts on the bench in front of you, then just pick something up and shove it in, see if it's the right thing or not. don't worry your pretty little [wooden] head about C taper or B taper, until youse gets the hot flashes going.

the 2 x 10uF caps are minus to minus, both (+) are out-pointing. this is a method of making a bi-polar cap. that spot in that osc is one place there is no real + to point the cap at. you need double your final value caps, so 2 x 10uF equates to 5uF - capacitors in series and all.

as for the a and b connections, just straight down, a goes to pin 1/sq out, and b goes to pin 6/ R2/ C1.

Vcc/2 is yes a voltage divider, so 4V5. don't fit a bypass cap to ground from the Vcc/2 point as is usual, because that point wants to jump about each cycle. also, as Druid says about those resistor values - I'm too lazy to learn the setting points on that circuit type.

and yes, you always need to supply supply and ground. we don't show it on snippets is all.


and, to confuse you, if you are going to run a led and ldr from this as you so often do, wire it with the supply+ > CLR > Anode -led- Kathode > tri out/ pin 7/ opamp output. as this IC will pull down to ground but stops short of the supply by 2 volts, you can hide the led turn-on voltage, about 1V7 for red, in that 2V deadband. more swing and drive, like your dancing.

Faster than his own shadow I've heard it be said.

perhaps check the date on that drawing ......

[moid]

moidy moidy moidy - do as we all do - throw all your parts on the bench in front of you, then just pick something up and shove it in, see if it's the right thing or not. don't worry your pretty little [wooden] head about C taper or B taper, until youse gets the hot flashes going.

Why I've never been called a pretty little head before! I think I might be getting one of those hot flashes myself! Oh my!

the 2 x 10uF caps are minus to minus, both (+) are out-pointing. this is a method of making a bi-polar cap. that spot in that osc is one place there is no real + to point the cap at. you need double your final value caps, so 2 x 10uF equates to 5uF - capacitors in series and all.

Well that was new to me! Thanks.

as for the a and b connections, just straight down, a goes to pin 1/sq out, and b goes to pin 6/ R2/ C1.

Vcc/2 is yes a voltage divider, so 4V5. don't fit a bypass cap to ground from the Vcc/2 point as is usual, because that point wants to jump about each cycle. also, as Druid says about those resistor values - I'm too lazy to learn the setting points on that circuit type.

and yes, you always need to supply supply and ground. we don't show it on snippets is all.


and, to confuse you, if you are going to run a led and ldr from this as you so often do, wire it with the supply+ > CLR > Anode -led- Kathode > tri out/ pin 7/ opamp output. as this IC will pull down to ground but stops short of the supply by 2 volts, you can hide the led turn-on voltage, about 1V7 for red, in that 2V deadband. more swing and drive, like your dancing.

OK I think I've got all that on the circuit (see below)

Faster than his own shadow I've heard it be said.

perhaps check the date on that drawing ......
[/quote]

Good grief! You're predicting my screw ups 7 years in advance? Quick use those powers to tell me next week's lottery numbers, I want to retire now!

Progress update - I built duck's circuit - used a 510K resisitor in place on a 560K ( I do have one... somewhere in a mixed bag of 5 series resistors). Good news - the Red LED lights up and doesn't explode. Bad news - it does not flash at all. I stuck the DMM on pin 7 and it says the resistance is 32.55K - oh it just slowly rose to 32.84K, and fluctuates by about 0.01 - 0.15K... and has slowly fallen to 32.38, OK there does seem to be a rise fall pattern; just a slight and very slow one. One of the ON SPDT positions gives a lower reading of around 27K, the other ON gives the same reading as the OFF position.

Connecting the flasher to lug 2 of the Frequency pot kills all audio going through the MS20 and turns off the two 3mm red LEDs on the PCB. So that's probably not a good thing?

For diodes I used 1n4148, hope those are ok. I found a C100K pot too! I've added the 10K pot duck mentioned.

Any guesses what to do next? Larger pair of capacitors? I guess get the LED on the flasher flashing and then work out why the circuit doesn't connect to the MS20?

[moid]

It's more a question of the *ratio* of the two resistors, rather than the values. Scale both down by ten and you won't change anything much. Scale *one* of them down by ten, you'll see a big change.
They set the point at which the schmitt trigger flips the triangle wave from going up to going down or vice versa. If you set that point low, the triangle wiggles about rapidly at low amplitude. If you set it much higher, it slows the triangle down because now it has to charge up (or down) for a lot longer to reach the trip point again. But you get a bigger amplitude. So adjust those resistors to get the *amplitude* you want, and then go back and change the timing cap and pot to get the *frequency range* you want again, because if you mess with the trip point, you'll change the overall frequency range of the whole LFO.

Thanks Tom for explaining that. I guess I'll begin by reducing the value of the 560K resistor and see what happens.

[moid]

With resonance maxed on the MS20 and the 560K reduced to

510K nothing happens
430K nothing happens


390K I got some oscillation, and the guitar also came through and is modulated and the flasher LED flashes! Quite rapidly

At 330K the flasher LED goes off - I get oscillation but guitar is weak and I'm not sure if there is modulation - maybe, but subtle.

back to 390K I can't tell if the SPDT switch does anything - no audible difference, but I need a much slower range of pulse - aha seem to have a bit of that - the C100K pot is very fast for a small part of the turn, and then about half that speed for 90% of the rest of the pot range with a very narrow transition between the two. I think the modulation is more shallow than it was on the previous Schmitt trigger circuit. Going to try a smaller pot next?

[moid]

Changed the pot to 25K - this has a more gentle sweep from one speed to the other, but the flasher doesn't go fast or slow enough. I would say the pulse varies from 4 Hz to 1Hz (aproximately).

Changing the capacitors to 22uF gives a rate of about 3Hz to 1Hz - actually worse range I think? Not sure?

I've added a switch to go between the pair of 22uF and a pair of 470uF. The 470uF are actually too slow, so that shows that this new flasher circuit gets different ranges from the capacitors as the previous circuit. Maybe the stop resistor needs to be smaller? Indeed it does 100R is soooo much better! Now I have pretty slow up to audio rate (a bit) on the 470uFs, and fast to super audio rate on the 22uFs

I changed the pot to 10K which seems to give good control over the sweep.

I probably need to make a drawing of this lot!

At the moment the only problems are:
I think the effect doesn't go as deep as the previous flasher though, and I cannot get the diode switch to do anything beyond slight pitch changes when it is activated. How do I get greater depth from the pulse and fix the diodes - are they the wrong type? Using 1n4148 at present.

[moid]

and a new mistake - the 330K resistor wasn't a 330K... it was a 330ohm...so perhaps the circuit would work better with less than 390K in it - would that help increase the depth?

[ElectricDruid]

and a new mistake - the 330K resistor wasn't a 330K... it was a 330ohm...so perhaps the circuit would work better with less than 390K in it - would that help increase the depth?

Draw us a schematic of what you've *really* got and we'll try and tell you!

[moid]

Picture time! Hopefully this makes sense... at the moment I wonder if decreasing the 390K resistor will give me more depth (higher values turned the circuit off), so maybe 330K or slightly less would be the thing to try? If not that then I'm not sure. No idea if that will fix the slope diode issues. Not sure if the diodes should be a different type?

[moid]

OK well changing the 390K resistor does not give better depth - it just slows the pulse length down as the resistance gets smaller - I tried 330K and 220K and neither changed depth; just slowed the overall speed of the LFO down. The diodes didn't magically start working either. Ermmmm help?