Dr. Quack follower reverse sweep...

[jakefuzz]

I've seen this topic come up a few times but never dealing with just the led as the envelope driver. I am using the second flashing LED to control an opto-isolator. Here is the circuit I have on the board right now:


Works great for standard envelope sweeps but I need a reverse sweep. Is there a quick and dirty way of doing it to the quack follower? I tried a few things with no luck. I also tried the Mutron follower first but couldn't get it to work. I like how simple this design is and ripple really isn't a concern for what I am testing.

[deadastronaut]

hi jake:  is this a sound to light circuit?..in laymans terms.. .?
ive been after something like this myself....thanks.

no idea on the reverse thing though...sorry!..

[jakefuzz]

Hey Rob, it is. I just took the envelope follower from the Dr. Quack project and cut everything out up to the transistor (and the filter of course). Works great for forward sweeps.

I just read your envelope filter thread and was wondering if your follower works well? Yours is super simple!

[deadastronaut]

cheers, great i 'll try it, i have a spare side of an opamp, makes sense to use/try that instead...

as for my one, i have it on bread at the mo, everything is fine, but,  as soon as i connect the 386 (sound to light part) i get distortion on the wah output...hmmmm....

[vendettav]

cheers, great i 'll try it, i have a spare side of an opamp, makes sense to use/try that instead...

as for my one, i have it on bread at the mo, everything is fine, but,  as soon as i connect the 386 (sound to light part) i get distortion on the wah output...hmmmm....

i was thinking the same actually 
wonder what's Dr Q's wah part

[deadastronaut]

great just tried it out....works a treat..just got to adjust for superbrights as thats all i have at the mo....

so back to your question: so we want this to work in reverse...so the led is on, and ducks out when playing...hmmm....tinkering time i think!....

thanks jake

i stuck a 4.7M  where the 2.2M  is....bit better for superbrights i think..


edit: hmmmm still getting distortion out of the wah....

[jakefuzz]

I'm glad you got it working. You should try it in your circuit. I think you could lower the 22K resistor too to make it a little brighter, mine gets super bright just like that.

The reverse sweep must have something to do with reversing the second led and maybe putting a 9 volt potential on one end? Every time I try something like that the LED doesn't light up at all.

[deadastronaut]

yeah ive tried it with the wah, but its distorting with it too......hmmm....but the light reacts well with a pluck..

[Gurner]

I've had a few tins of falling down water, & have knocked the schem below up in very short order so might not be 100%, but should give you an idea of the approach towards a reverse sweep....



The LED is biased 'on' with no signal present (9V->LED->220R resistor->VCC/2 - decrease the resistor value for a brighter LED & vice versa).

Only positive signal cycle swings are allowed through that diode attached to the opamp output ...these positive swings start turning the LED off - adjust the gain of the opamp to get the best 'fade' response (&/or the 220R series limiting resistor depending on the LED colour/type used)

[jakefuzz]

Thanks Gurner! Ill get this boarded tomorrow hopefully and see how it works.

[jakefuzz]

Hmm I can't seem to get this one to work. It looks like it should work. I had to put in a dc blocking cap after the diode to get the LED to light  doesn't make sense. I am using a signal gain of 100, must experiment more...

[Gurner]

What happens with the LED when you disconnect the rectifying diode (ie isolate the opamp output diode)

To get the correct fade response, the correct biasing of the LED is going to be key (as is the signal gain) - what colour LED are you using? (any idea of the forward voltage spec)

Some DC voltages would be useful (but without the capacitor you've used)...

Voltage supply           (don't use a battery....my 220r resistor assumed a red led, approximately 10ma of LED current & a 9V regulated supply)
Opamp +ve input pin
Opamp -ve input pin
Opamp output pin
D1 Cathode

Is your VCC/2 source low impedance? (ie regulator, or buffered opamp output etc)

[jakefuzz]

I am away from my workbench right now. But it was a super bright white LED so probably 1.5 volts. But in the actual application I am using a 4N35 optoisolator with a quoted range of 0.8-1.4 I think. I just used a 9 volt battery. I could use the other side as a Vref buffer but I just used a 10k/10k voltage divider from the supply. Ill get some voltages when I get back and maybe try that impedance transformation. 

[Gurner]

You've a couple of issues....

But it was a super bright white LED so probably 1.5 volts.

A super bright white LED has a forward voltage of typically 3.2V.

I just used a 9 volt battery.

Not ideal for circuits that use LEDs...as the battery levekl fades so does the led brightness....best using a regulated 9V supply.

I could use the other side as a Vref buffer but I just used a 10k/10k voltage divider from the supply.

it'll need to be low impedance  ...a 10k/10k divider will mess up all the required DC levels.

Edit: Now I've revisted my schematic (& without beer in my belly), there are some tweaks needed (the points labelled VCC/2 actually need to be a specific voltage *not* VCC/2 ...& the level depends on the type of LED used) ...but it's my kids bath time routine now...so one for later!

[Mark Hammer]

It doesn't get talked about much, but downward sweep is "different" than upward sweep.

If it is a bandpass filter, very little signal lives up where the filter centre-frequency starts out for downward sweeps, but most of it lives where the upward sweep starts out.  That has implications for how fast you want the filter to move.  A slow-moving upward sweep can sound "dead sexy!", but a slow-moving downward sweep can sound plain dead. 

Moreover, since the frequency content of guitar signal varies with time, there is the matter of strategically planning out where the filter is going to be when it gets to this point in the note.  Downward sweep is far more effective for lowpass filter mode than for bandpass because, although the fundamental is always there, the note always starts out with more harmonic content than it ends up with.  So a downward-swept lowpass (especially with the resonance turned up) accentuates all those harmonics then sucks them away, leaving you with mostly fundamental.  It effectvely exaggerates the natural "harmonic arc" of the note.  A downward-swept bandpass, however, needs to vacate the top end before the top end itself leaves, and doesn't necessarily provide enough "meat" to the signal in the form of lower-order harmonics.

Conversely, an upward swept filter pretty much always has to move fast, regardless of mode, if its going to get to where the quickly-dissipating harmonic content lives.  This has implications. Thirty years ago, I used to use my compressor, 6-band EQ, and Envelope Filter to mimic reverse tape.  I would stick the compressor ahead of the filter, and the EQ after it with the bass turned down and the mids and highs cranked.  Because the MXR Envelope Filter lets you get a fairly slow attack, and the wonky time recovery on my compressor slowed that down even more, the filter would sweep upwards too slow to "catch" the treble before it left.  As a bandpass filter, that meant that the earliest part of the note was attenuated by the EQ. As it swept upwards, it entered the lower mids which were boosted by the EQ.  There was no treble left in the signal by the time it reached the top, but with the EQ pedal controls cranked, there was some distortion generated by the remaining mids.  That created the illusion of a note that seemed to swell in volume and end with more harmonc content than it started with.  Neat, huh?

Since starting point impacts on audibility of the output, start-point for downward sweep can not simply be "the opposite of upward sweep".  It needs to be planned out and actually start below the point of maximum sweep for the upward direction.  That's why when you see the direction switch for filters like the Mutron and such, you will see a resistive divider to provide the "right" initial voltage.

[Gurner]

Good points well made - completely agree with the above, my input was simply to the problem of how get an LED to light under no signal condition but then fade out with signal magnitude.

This ought to be better (I didn't save the eagle file, so have simply hacked the jpg - complete with spellling errors that I can't be arsed to correct!)...



It's not shown but an electrolytic cap straddling the LED should help holding the LED dim a little longer.

For a white LED, I reckon the series limiting resistor for 10mA @ 9V supply ought to be something like 580R.

the idea being that at the LED junction, there'll be about 5.8V (9V less the fwd voltage of the LED)

So we want to bias the opamp so that its output is just above that (to cater for the fwd voltage drop of the diode)...therefore assumning a fwd voltage 0f 0.6V for the diode, i reckon biasing the opamp for about 6.3v would be good. (so it's just biased off)

Signal comes in, the positive swings make it through the diode & start turning the LED off.

the values & voltage levels will need to be modified depending on the colour/spec of the LED

[Mark Hammer]

If a filter is preset for only one direction of sweep (as this one appears to be) its no problem whatsoever.  You simply tune it to the way you like it (presumably R6 is a fixed resistor and trimpot), and away you go.  The challenges arise when you want to simply flick a switch and reverse direction.

I think the modded MXR Envelope Filter presents a nice illustration of this.  Stephen Giles astutely noted some time back that there was this unused 4069 inverter section in the circuit, and that if one simply made use of it, you effectively "reversed" the duty cycle, such that what was formerly 70% duty ("on" 70% of the time) would now be 30%.  While that certainly changed the direction of the effect, in a dead-simple way, without demanding anything more than a toggle to bring in or lift the extra inverter section, it did not result in an optimal "starting point" for the downward sweep.  An ideal direction flip would want some sort of small adjustment to the initial duty cycle (an offset) to produce a more musical sweep, which involves a little more than a toggle.

[jakefuzz]

Got it Gurner Thanks a bunch man. Works perfectly and I can tune the response by changing R6 and my voltage divider values. The electro cap definitely helps to sustain the response a little.

[Gurner]

Got it Gurner Thanks a bunch man. Works perfectly and I can tune the response by changing R6 and my voltage divider values. The electro cap definitely helps to sustain the response a little.

Cool, there may be more elegant ways to approach the same result ...but not bad, bearing in mind I was winging it with that first design whilst enjoying a mild mannered beer buzz!

[Earthscum]

Couldn't you use a PNP instead of the NPN with some small modifications?