Understanding the MXR Envelope Filter

[mdh]

I've been struggling with a MXR Envelope Filter built from the old Tonepad layout (A-gua2, Rev 1: http://tonepad.com/getFileInfo.asp?id=5), and I'd like to try to make some headway by breaking the circuit down into functional blocks that I can breadboard and test individually.  The main reason I want to do this is that I'm stubborn, and I'd like to see if I can make the circuit work with the TI 4069UBEs that I have, even though build reports seem to indicate that they don't work in this circuit.  If I play around with it on the breadboard, maybe I can figure out what needs to be tweaked to make it work.  I also have some 4049s that I'd like to try out on the breadboard.  These considerations, along with Occam's Razor, make me want to break it down to the essential bits.

So... looking at the schematic, I'm thinking that after the input cap, R7 leads to the filter portion of the circuit, and R8 leads to the envelope follower.  Within the filter portion, IC1c is a gain stage to drive the filters; IC1a and IC1f, with associated capacitors and resistors, comprise the filters, which are swept by the switches in IC3.  I'm a little unclear on how that works, though.  Could I just stick a dual-gang pot (wired as variable resistors) in place of the in/out terminals of IC3 for the purpose of manually sweeping the filter?  Again, this is just for playing around with the circuit on the breadboard, I'm not trying to come up with a manual sweep mod or anything like that.

The envelope follower leaves me very confused.  I'd rather not speculate too much on what's going on in there.  Unfortunately, that's where I suspect the problem will be, so it's the part I'll need to make an effort to understand.  Any hints would be most welcome!

If anyone is willing to help walk me through this thing, I would be most grateful.  I'm very willing to study specifications and graphs in data sheets, but I need some guidance to get started.  Thanks!

[markusw]

I had the same question some time ago and was really lucky 

http://www.diystompboxes.com/smfforum/index.php?topic=20811.0

Markus

[StephenGiles]

wrong post

[StephenGiles]

I've been struggling with a MXR Envelope Filter built from the old Tonepad layout (A-gua2, Rev 1: http://tonepad.com/getFileInfo.asp?id=5), and I'd like to try to make some headway by breaking the circuit down into functional blocks that I can breadboard and test individually. 

[Paul Perry (Frostwave)]

The switches are turning on and off the resistors in series with them (think of it this way, if you connect a 10K resistor fo only 50% of the time, it looks like a 20K resistor!) so, replace the switches by a dual  pot, maybe 500K.

[mdh]

I had the same question some time ago and was really lucky 

http://www.diystompboxes.com/smfforum/index.php?topic=20811.0

Markus

Excellent, I must have just missed that thread.  Thanks!

[mdh]

I've been struggling with a MXR Envelope Filter built from the old Tonepad layout (A-gua2, Rev 1: http://tonepad.com/getFileInfo.asp?id=5), and I'd like to try to make some headway by breaking the circuit down into functional blocks that I can breadboard and test individually. 

I could interpret this as, "Way to go, have at it!" or, "Jeebus, man, you're a fool!  Go back to playing with your tinker toys!"  I'm sort of curious whether you meant one or both of those sentiments 

[mdh]

The switches are turning on and off the resistors in series with them (think of it this way, if you connect a 10K resistor fo only 50% of the time, it looks like a 20K resistor!) so, replace the switches by a dual  pot, maybe 500K.

Perfect, that's what I thought (except I didn't get the rapid switching angle... makes a lot of sense now).  Thanks!

[slacker]

I've been struggling with a MXR Envelope Filter built from the old Tonepad layout (A-gua2, Rev 1: http://tonepad.com/getFileInfo.asp?id=5), and I'd like to try to make some headway by breaking the circuit down into functional blocks that I can breadboard and test individually. 

I could interpret this as, "Way to go, have at it!" or, "Jeebus, man, you're a fool!  Go back to playing with your tinker toys!"  I'm sort of curious whether you meant one or both of those sentiments 

Mr Giles often suggests breaking large circuits down into blocks for building and testing, so I assume he meant it in a good way.
I can't believe the timing of this thread, I was just about to ask exactly the same questions
I'm thinking of building something based on this but couldn't figure out the whole schematic, mainly the bit that I now know is the "clock".

[StephenGiles]

I've been struggling with a MXR Envelope Filter built from the old Tonepad layout (A-gua2, Rev 1: http://tonepad.com/getFileInfo.asp?id=5), and I'd like to try to make some headway by breaking the circuit down into functional blocks that I can breadboard and test individually. 

I could interpret this as, "Way to go, have at it!" or, "Jeebus, man, you're a fool!  Go back to playing with your tinker toys!"  I'm sort of curious whether you meant one or both of those sentiments 

Way to go, have at it!

[Mark Hammer]

I received an off-line query from someone about the MXR EF clone at Tonepad recently.  As usual, it was an "I built it but it doesn't work, did I get the wrong chips?" sort of query.  That's something that comes up often enough that I started to cogitate about it, and it dawned on me that maybe the culprit is not the brand of invertor chips, but rather the specific CMOS switch chip in tandem with the components that result in the amplitude of the clock pulse.  Is it possible that the "on" pulse going from the clock to the switch is simply not enough to turn the switch on for some brands?  If so, then that might mean that one could use virtually ANY brand of CMOS chips if we knew how to change certain component values and adjust the properties of that clock pulse to do its job.

Does that make sense, and does anyone know what to zero in on in the circuit to accomplish that?

[mdh]

I received an off-line query from someone about the MXR EF clone at Tonepad recently.  As usual, it was an "I built it but it doesn't work, did I get the wrong chips?" sort of query.  That's something that comes up often enough that I started to cogitate about it, and it dawned on me that maybe the culprit is not the brand of invertor chips, but rather the specific CMOS switch chip in tandem with the components that result in the amplitude of the clock pulse.  Is it possible that the "on" pulse going from the clock to the switch is simply not enough to turn the switch on for some brands?  If so, then that might mean that one could use virtually ANY brand of CMOS chips if we knew how to change certain component values and adjust the properties of that clock pulse to do its job.

Does that make sense, and does anyone know what to zero in on in the circuit to accomplish that?

That was me... but then I decided to try to "learn to fish," so to speak, rather than to post a debugging thread with voltages.  I freed up a largish breadboard by finally committing my GEO test oscillator and GEO FET matcher to perf, so I should be good to go.

FWIW, with respect to the interaction of the 4066 with the 4069s, I originally had TI 4069s and a TI 4066 in there, but I've also tried a Motorola 4066, which resulted in the same behavior that I described in the PM I sent you.  I'm guessing that the problem is in that clock pulse generator.  This would be consistent with the Tonepad build reports, in that it seems like swapping IC2 out for a non-TI brand improves the behavior of the circuit.  So I think that what I've done here is volunteered to try to become an expert on this portion of the circuit, and work out whether we can make it work consistently with chips from different manufacturers.  I have a long way to go, but I'm willing to try    Maybe I'm naively optimistic, but I think if I can break it down into simple enough blocks, I have a chance of figuring it out.

[mdh]

I can't believe the timing of this thread, I was just about to ask exactly the same questions
I'm thinking of building something based on this but couldn't figure out the whole schematic, mainly the bit that I now know is the "clock".

Cool.  I've been meaning to play around with your Slack Trem, and I seem to remember seeing something else you did with inverters.

[slacker]

If the problem is that the pulse isn't high enough to switch the 4066 then it might be possible to increase it by putting a large value resistor across pins 8 and 9 of IC2 turning it in to an amplifier. There's an example oscillator circuit using this method in the Philips 4069 datasheet. I guess that could affect the sensitivity of the envelope follower though so that might need some tweaking as well.
Could the problem be that the oscillator isn't even oscillating?

and I seem to remember seeing something else you did with inverters.

I think the only other thing I've done using invertors was the LFO from the slackfilter. I'm hoping I can mod the Envelope Filter into a simpler, better version of this. Pretty much like Mark and Markusw talked about in the other thread.

[mdh]

If the problem is that the pulse isn't high enough to switch the 4066 then it might be possible to increase it by putting a large value resistor across pins 8 and 9 of IC2 turning it in to an amplifier. There's an example oscillator circuit using this method in the Philips 4069 datasheet. I guess that could affect the sensitivity of the envelope follower though so that might need some tweaking as well.
Could the problem be that the oscillator isn't even oscillating?

Could be anything... I haven't actually had a chance to implement my master plan of breadboarding this stuff, and the last time I probed around my PCB build, I didn't really know what I was looking for.  Thanks for the data sheet reference... I think I'll be looking at all the CMOS inverter data sheets I can get my hands on.

[StephenGiles]

Could be a further example of cutting a circuit to the bone, so that it just worked using the parts on which MXR got a good discount on - sound marketing sense of course, but doesn't help you.

[mdh]

I was wondering whether anyone might be able to clarify for me what kind of behavior I should expect from the oscillator comprised of IC2a,b,c,d.  I gather that it should be oscillating all of the time, so that if I build it in isolation (which I have, on the breadboard) it should oscillate.  The trouble is that I don't know what the frequency should be, and I'm not sure that I have any instrument capable of detecting ultrasonic oscillations.  My DMM has a frequency setting, but I have misplaced the manual, so I don't know the range over which it can measure.  I don't have a scope either, just a high-impedance buffer for my soundcard, so I assume I won't be able to detect anything above audio range.

If I'm not equipped to study the oscillator directly, maybe I can study its effect on a 4066 switch.  If I put a voltage across the switch (with a current-limiting resistor in series, I assume... I'll look at the data sheet), then I should be able to measure the effect of changes in the oscillator circuit on the current passing through the switch, right?

Finally, am I correct in thinking that introducing a signal (from, e.g., an LFO) somewhere around IC2d pin 8 will allow me to look at the transient behavior of the current through the 4066 switch, or should I introduce the signal just after the attack pot (R28)?

[slacker]

I've got this on the breadboard now so hopefully I can help. The oscillator should be working all the time, mine seems to be running somewhere between 10-20KHz, so you should be able to measure it with your DMM. Probe around either side of c10 and you should get some readings. The other thing you could try is change c8 (the 100pf) for something like a 1n or larger cap. That drops the speed and if the oscillator signal is switching the 4066 you'll be able to hear it as a whine on the output of the filter.

[Transmogrifox]

If I'm not equipped to study the oscillator directly, maybe I can study its effect on a 4066 switch.  If I put a voltage across the switch (with a current-limiting resistor in series, I assume... I'll look at the data sheet), then I should be able to measure the effect of changes in the oscillator circuit on the current passing through the switch, right?

Yes, exactly.  This is one of the first things that I would try.  I didn't go to read the other thread about the theory of how this filter works, so I don't know what has already been explained about how this works, so I'll repeat it here for the sake of everyone who did not follow the link to the other thread (simple convenience):

The oscillator should be approximately triangle wave .  IC2d theoretically will switch when the voltage is near 1/2 Vdd (these typical numbers are quoted in the datasheet).  The envelope follower circuit provides a DC voltage to shift the "0" point on the triangle oscillator center.   When the triangle oscillator center is low, IC2d only turns on for a short pulse, since the triangle wave crosses over the switching threshold and back for a very brief period of time.  When you dig into your guitar, the dynamic increases, thus the DC voltage shift at the gate of IC2d increases, thus pushing the triangle wave '0" reference more positive, which cases the switch to be on for a longer period of time. R22,23,26,14,16 all will effect where this "0" point resides in the "off" state.  You can also twiddle R14,16,20 to change the amplitude of the oscillator.  All the changes in this part of the circuit are pretty interactive, so you may find it most useful to mess with R23 and R26 the most if it appears to be not working.  I really think that the ratios of the resistors there are theoretically optimal for the type of circuit this is.  The only reason you would need to change these is if the turn-on threshold of IC2 is much higher than that of a different brand.

Finally, C10 is an interesting circuit part that had me scratching my head for a few minutes.  It's a small-valued capacitor that feeds signal from the square wave output of the oscillator.  It is small enough that all it allows through is a short pulse to pop IC2d open once for a short time every oscillation period.  What this does is it makes sure the switch is on for at least a very short duration each clock cycle.  If both switches were turned off altogether for a long period of time, it would screw up the bias on the filter section so it would sound pretty weird when you started to play up until the filter was able to reach bias conditions.  It would be distorted sounding on the attack.

So...I would recommend that you do as you said, by hooking up a resistor and a switch together.  I would also suggest putting a capacitor in parallel with the resistor to even out the DC.  Breadboard the oscillator section with IC2, and apply a voltage from a pot buffered by an op amp to R23 on the IC2a side.  Watch what happens to the voltage at the switch/resistor setup you made as you twiddle the pot.

FYI, This "test setup" would make an interesting distortion pedal.  At the junction of your pot you could add an amplified guitar signal through a capacitor and then take the output from the switch.  It could also work as a somewhat clean amplifier for lower amplitude signals--in this case the triangle wave oscillator amplitude controls the gain.  I got off topic....

[mdh]

Wow, thanks for that post, Transmogrifox.  I breadboarded it twice yesterday, and I finally got the oscillator running at about 30-80kHz, depending on the chip (according to my DMM).  I found that one of the TI 4069s oscillated closer to 1-3kHz, so I wonder if that one has suffered some static damage or something.  I still haven't been able to measure the amplitude of the oscillation, and haven't yet had success measuring the oscillator's effect on the 4066.  Alas, I've run out of weekend, and probably won't have much time to work on it this week

One thing I have noticed consistently, however, is that about half of the supply voltage is getting dropped across that 3k resistor between +9V and Vs (R4 on the Tonepad schem).  I had asked Mark about this when I PM'd him a couple of weeks ago, and he thought that resistor should just be limiting current, and shouldn't drop a significant voltage.  I figured maybe there was a solder bridge or bad resistor value somewhere on my PCB, but since I've breadboarded this circuit (and variations) twice with the same result, I don't think that's it.  The thing is, if the 3k resistor is dropping 4.5V, then the whole circuit is drawing 1.5mA, which seems on the low side for all that appears to be going on in this effect.  But if that resistor was dropping 0.5V, giving a Vs that's still close to 9V, then the circuit would only be consuming 167uA, which would be lovely, but I don't believe it.  So all of this makes me wonder what would happen if I replaced R4 with a smaller value, or jumpered it altogether, but I'd like to know beforehand if I'm likely to let out the magic smoke.  Or am I missing something here?  If the designer wanted a supply voltage around 4.5-5V, why not just use a zener diode or voltage regulator?

Slacker, what voltage are you seeing at Vs?

And thanks for your patience, everyone.  I'm determined to get this, but I am groping around sort of half blind here.