Will the real MS-20 filter please step forward

Started by Strategy, April 16, 2012, 07:15:39 PM

Previous topic - Next topic

Jordan A.

Quote from: Cliff Schecht on June 27, 2012, 01:03:27 PM
So people have tested the Rene schematic with a 13700 and it didn't work properly? I'd much prefer this part over a 3080 for the OTA's. If it doesn't work I can run some simulations to figure out what needs to be changed. I suspect the 13700 has *slightly* different requirements for the ibias pin but it should not be a big deal to switch to the superior (and cheaper!) chip. I'll look into this as well as a single supply, the latter will be trickier to implement methinks.

No, I've built it several times using 13700s.  Seems to work just fine, definitely sounds good, nice range of frequency control, etc.

Also I've lowered the power rails to +/- 9 volts without noticing any big change in normal use, but I have not messed around with a single rail supply, I would love to see what you come up with!

The weather here is too nice for me to be dorkin' around online, but it's fun to see people messing with this one,

jordan

Cliff Schecht

The single supply circuit I am running seems to work alright. I still have to figure out the expo current source and resonance circuit but the heart of the filter works off a single supply (on the bench at least :D). So far I'm running the circuit off of 15V (what my powered breadboard has available) but surely 9V won't be an issue either. Next I need to build a proper current source and the resonance section to complete the system, but I don't anticipate either of these giving me problems. I'll keep you guys updated and try to post some schematics once I get a bit further.

WhenBoredomPeaks

Isn't the Korg Monotron already contains an MS10/20 filter which is single supply/low voltage? Its schem was published by korg.

I think in the league of simple 12db low pass filters the king is the Polivoks. It's a shame that according to the forums there are not too many of us who could make it work (including me, it didn't work on the breadboard nor on a homemade etched PCB for me)
I think i gonna soon order a PCB for it and gonna continue messing with it, (it would be nice to turn it into 9v/single supply) it's sound would be pretty cool for filtering distorted guitar signals.

garcho

#123
I just want to second Jordan and say that there are no issues I noticed when using LM13700 or 9V bipolar. I haven't tried this circuit with the 3080 or 12-15v so I can't compare the two. Put that iron down it's summer! (for some of us  ;))

  • SUPPORTER
"...and weird on top!"

Cliff Schecht

#124
After spending the entire day messing with various expo current source circuits, I found an NPN variant that works well with a single supply. It's functionally equivalent to what the MS-20 uses (or good enough at least). Since the NPN's only sink current in this circuit, I mirror the expo current up to a PNP current mirror and source current to the Iabc pin from the second PNP. Works in simulation and should work fine on the bench.

Here is what I have so far (verified in simulation and parts on bench, may need values tweaked!): http://www.aronnelson.com/DIYFiles/up/ms20_9vsupply.pdf

Cliff Schecht

Also worth note is the expo circuit currently is backwards from standard convention, that being a 0V input to the cutoff pin is the highest cutoff frequency and a 10V input is fully off. For guys just running an LFO this is no issue (and not fixing the polarity saves us an op amp), if you were to run a CV voltage from a knob then the cutoff knob would work backwards (if the pot is wired correctly!).

garcho

  • SUPPORTER
"...and weird on top!"

Jordan A.

Great, thanks Cliff.  I just built it up, using BC550 and BC560 for the BJTs, 1N914 for the diodes, and 10k for R3 and R7.  Sounds good!  I'll have to put the bipolar one and single supply version side by side, but the single supply circuit from you definitely sounds good, in that nasty kind of way.  When fed with a wave with sharp transitions (pulse/saw) and high resonance you get the really interesting frequency locking stuff as you sweep the frequency this VCF is known for.

nice work,

jordan

Cliff Schecht

Sweet! I was wondering if anyone was actually going to build this circuit, that pesky Iabc bias circuit took me quite a bit of brain thinkings to get working correctly and I didn't want it to go to waste. The PNP version (meaning no NPN transistors in the bias source) would rail out the op amp before the BJT could actually bias itself on so I switched to the NPN version that mirrors up to PNPs. Seemed to work fine in sim (and the bipolar version has worked for me on the bench in the past). Either way glad it works for someone else!

Is the bass response good on it? I always use a large input cap on filters so I can get those low down dirty PWM'ed square waves to punch hard (they sound weak and buzzy if the input cap is too small). Any other complaints or tweaks? I think 10k was the right choice for those missing resistor values but you could try tweaking them if you don't like the range of the filter (smaller means more range). You can also tweak the 1.8k resistor but then you lose the 1V/octave effect which is why this circuit is used in the first place.

I think this effect would also benefit from some voltage controlled resonance. It could be eassily implemented with a second 13700. I'd like to have both cutoff and resonance controlled by pick attack to form a super funky envelope follower.

ml

Quote from: WhenBoredomPeaks on June 23, 2012, 01:49:51 AM
VREF is always half of the supply voltage. (it is 4.5V when you use a 9V battery to power the pedal)

Great, thanks! I did a bit of reading and am under the impression that the Vref is only needed in the case of a single supply. 
If using +/-15V, then instead of Vref, it should just go to ground.  Is that right?

Lots of interesting stuff going on here. I'm interested to see how Cliff's single supply version sounds!

WhenBoredomPeaks

Quote from: ml on July 06, 2012, 04:26:16 AM
Quote from: WhenBoredomPeaks on June 23, 2012, 01:49:51 AM
VREF is always half of the supply voltage. (it is 4.5V when you use a 9V battery to power the pedal)
If using +/-15V, then instead of Vref, it should just go to ground.  Is that right?

yeah

nordine

will jump in the wagon, since i love this filter and this particular chip  :P

noiw, what has me really interested is the feedback path, i tried an opamp solution last time i tinkered with it, but to no avail, will see if this time it works, cause feedback path and clipping ala Escobedo's 9V MS20 sounds nasty

garcho

#132
So, I've tried breadboarding Cliff's schematic a few times, and keep getting the same problems. Jordan said it sounded fine when he did it, so I'm sure I'm missing something and f-ing it up. I used 10k resisters for R3 and R7, tried Si and Ge diodes, as well as a variety of LEDs, and I wired the wiper from a 50k pot to the point marked 'cutoff', with terminals 1 and 3 going to +9V and 0V. I get constant oscillation, which can vary in tone while I move the cutoff pot from 0 to 9. The resonance doesn't change much in the sound, but will sometimes cause the sound to go silent temporarily. When I add a LFO via another 50k pot and 100k resistor in series, the oscillation and resonant wackiness increase. The signal from my guitar sounds fizzy and crappy. I've tried swapping out the ICs a few times with no luck. Here are some voltages I took:

voltage is 13.36 and VCC/2 is 8.05
when cutoff is 0V and the guitar signal is about 0.65V (going through a booster):
LM13700:
1 - 1.29
2 - 0.15
3 - 4.65
4 - 4.72
5 - 1.88
6 - 0.00
7 - -0.04
8 - 0.10
9 - 0.10
10 - -0.04
11 - 13.36
12 - 4.65 to 4.85
13 - 4.67
14 - 4.85
15 - 0.15
16 - 1.29

TL074:
1 - 5.95
2 - 5.95 - 1.90
4 - 13.36
5 - 4.71 to 4.81
6 - 4.45
7 - 4.40
8 - 2.60
9 - 4.72
10 - 4.76
11 - 0.00
12 - 5.00
13 - 5.43
14 - 4.62

Q1
c - 0.90
b - 12.75
e - 13.23
Q2
c - 13.23
b - 12.75
e - 12.80
Q3
c - 12.80
b - 8.23
e - 7.75
Q4
c - 7.75
b - 8.20
e - 8.20

when the cutoff voltage is at 50% (6.5V) the pins' voltages are significantly lower. I didn't test it with cutoff at 13V. I can post the other voltages if it helps.

any ideas?

EDITS: added transistor voltages, remeasured VCC/2
  • SUPPORTER
"...and weird on top!"

Cliff Schecht

Something is wrong with your reference circuit I think. The OTA inputs should both sit at 1/2 VCC which they clearly are not. I think you should start here. Even the half supply point you measure isn't actually 1/2 VCC or really even close to what it should be (~6.7V). I would start here and figure out why the 1/2 supply part isn't working right.

garcho

Thanks Cliff, will try and hammer this out tomorrow
  • SUPPORTER
"...and weird on top!"

Jordan A.

I forgot to mention that when I built the single supply version I buffered the bias voltage with an opamp, it was a while ago but I think I remember the bias voltage acting whacky until it was buffered. 

This is pretty much the first (solid state) thing I've built with a single supply, so opamp biasing using a resistive divider is new to me, but maybe buffering VCC/2 is something to try, Garcho?

jordan

garcho

Where is the source of VCC/2? From the base of T3? From pin 10 of the TL074?
To buffer the VCC/2, I take the bias from the voltage divider and send it to the non-inverting input of an opamp, then tie the inverting input to the output, and the output is the buffered VCC/2? So where in Cliff's schematic is the voltage divider? What point to I connect to the non-inverting input?
Am I way off? Sorry, I need someone to hold my hand.
  • SUPPORTER
"...and weird on top!"

Cliff Schecht

You are spot on. Use something like a 100k/100k voltage divider (both resistors equal value is all that is really important, use a larger value to prevent excessive current draw) and buffer this with an op amp unity gain follower. I didn't show this in my schematic because I wasn't sure if a buffered 1/2VCC point was needed, although just looking at how much it gets used I think it's a pretty obvious point for doing so.

What I might do is add a dual op amp to the design to buffer the low-impedance input as well. The BJT's in the OTA have an input impedance that varies with the bias current which could cause loading issues to some older effects pedals if not buffered (plus this lets us add an input gain/attenuation knob to set the levels for different guit-fiddles).

garcho

Well, I've got the VCC/2 at buffered 6.7V, but both inputs on both OTAs are at 6.4 or so. The sound is better than it was, but still super f-ed up, with squealing, noise, fuzz and crap-outs, and no filter effect at all. I'll keep triple checking everything, but any ideas how that could be?
  • SUPPORTER
"...and weird on top!"

garcho

Anyone else looking at this? Still can't get the single supply version worked out...
  • SUPPORTER
"...and weird on top!"