A fairly simple quad op amp oscillator for out-of-phase LEDs

[midwayfair]

It's not without its problems, but I haven't seen this exact set-up anywhere before:



It's explained a bit on the schematic, but there was limited room. Also, I know some things might look a little funny (namely that there doesn't appear to be anything actually inverting the LFO ... but they're definitely blinking out of phase!), but I really does work pretty well despite the simplicity.

Here's what it does well:

-It can go fully dark and quite bright, so you can get a good switching range out of your LDRs.

-The depth control turns both LEDs on fully as it's turned down. This is important in most tremolos. I assume there's some way making them both go dark, but I'm not sure what it is.

-You can use all your favorite Tremulus Lune controls.

-At most depth settings, the LEDs are pretty well balanced.

Here's what it doesn't do so well:
-The LEDs are a little unbalanced at lower depth settings. The "out of phase" side (the LEDs on the op amp without the depth control) seems to be a little shallower than the "in-phase" side until the depth control is turned up a little farther. About halfway, they're equal. I don't know why this is happening, but I wasn't able to find a solution to the problem.

-The onset of the depth control is a little sudden. About halfway up is really deep.

-I tried everything I could to soften the waveform and the LED behavior, but I can't get anything that works without affecting the other set of LEDs. The tremulus lune has resistors in parallel with the LEDs; that didn't work so well here. I might experiment further, but perhaps someone else will have suggestions.

-I find it very peculiar that it works with two LEDs in series on both sides, but not one LED on both sides, and it seems like all four LEDs need to be hooked up to get the out-of-phase side to oscillate. Also, it helped to have at least one LED on each side be blue, so the Fv seems to matter in some way.

-Finally, I can't tell if this is just that I'm working on a breadboard in a room without three-prong wiring, but it's a little noisy. I'm not getting any ticking, but there is some "woosh" and hum when the LFO cycles. It could be in my audio path and not the LFO. I'm recommending a low-current consumption IC, but all I had were TL074 and LM354 (I killed a couple others doing something dumb ...).

[midwayfair]

Did some more work with a clear head, and it's now much improved! The depth control is basically the same as before, but it's been moved. The LEDs are no longer interactive. The wave form is waaaaay smoother, now, too, much closer to a sine wave, and can be softened further with the same parallel resistor trick that's used in the lune. (I think the second set of buffers really helps smooth it out, because it's much harder at the output of IC2.)

One nice thing about this is that the LFO is never inverted. Aside from the fact that I simply couldn't get any inverting op amps to work right, almost no extra parts are required compared to the original dual op amp version from the Lune, except the trivial extra space for the quad op am. It just buffers and then sends the LEDs to opposite power rails.

The balance trimmer pulls double duty. It provides a second 4.5V reference for the depth pot and lets you balance the on/off times of the LEDs. I'd seen this in a bunch of places

Overall I'm really pleased with it now and I can think of some good uses for it.

The depth pot's value and taper could be tinkered with. An audio taper might actually be a little better, depending on how much depth you need early in the sweep.

Time to do more work on audio paths to use with it!

[duck_arse]

sourcing and sinking.

jon, are you still having any trickery with numbers/colours of leds, and did you try putting the leds into the feedback loops? I'd been wondering about the "in the f/b loop, sinking" bit last I was playing with oscs.

[midwayfair]

sourcing and sinking.

jon, are you still having any trickery with numbers/colours of leds, and did you try putting the leds into the feedback loops? I'd been wondering about the "in the f/b loop, sinking" bit last I was playing with oscs.

I still need two LEDs in series to get the LEDs to go fully dark. This seems to be true regardless of what color the LEDs are, so I have no idea what that's about (meaning, a single blue LED doesn't turn on/off as much as two red LEDs), though maybe I'm just not being thorough enough. It's got to be a voltage drop; adjusting the current limiting resistors doesn't do the same thing. It has no trouble driving a blue + diffused (the blue will blink a little harder -- sounds more triangle in-circuit) or two blues. Two diffused, or a diffused and the IR diode in a VTL5C1, are very smooth.

Putting them in the feedback loop and putting a resistor to the power rail or ground gives very little depth, even with multiple LED drops. I don't know enough about this method to know if I'm doing something wrong; I looked at RG's out-of-phase LED circuit for that. The blinking is also a bit more triangle-ish looking. Not nearly as smooth even with the shallow depth.

A couple things I found playing around with it more:
You don't need to use both buffers for the LEDs. If you connect LED3's anode to IC3's output pin, everything seems to work equally well. They don't seem to interfere with each other that way.

This left the extra buffer open, so I tried using it for the 4.5V reference for the depth pot -- it's tough to tell for certain, but it seems to be a little easier to dial in, and there aren't any trimpot settings that completely kill either LED.

However, one use for using both buffers for the LEDs is that you can use a SPDT switch to put them in phase or out of phase. You don't get to use both LEDs for the bypass indicator that way, but you can still use one.

Also, a 100nF cap from IC1's + pin to ground gets rid of ticking if you want the square wave.

I'm drawing a bunch of schems in Eagle, so I'll repost something more readable in a bit.

[midwayfair]

Okay, posted the new schematic. Should be easier to read now, too.

There are notes on the schematic, but the long and short of it is that one method (A) makes the reference voltage for the depth pot easier to dial in and allows the use of one pole from the footswitch to turn off both sets of LEDs (so they can pull double duty as bypass indicators), while the other (B) lets you put the LEDs in or out of phase without putting them in parallel (which would require matching) but there's no good way to wire the LEDs through the footswitch as bypass indicators. Also, although I'm sure there are some cool uses for flipping the polarity of the LEDs, you probably still need to switch some things in your audio path to take advantage of it, so it might not be the most useful thing ever. But you can do some fun stuff with stereo tremolo effects with that switch, too.

Oh. Another cool thing: At minimum depth in a tremolo, the LEDs are both "mostly on," but very very, very slightly dimmer than the peaks during oscillation. So the volume is actually self-correcting (at all depth settings) in that regard, and you could actually wire up a bypass that disconnects lug 3 of the depth pot as your bypass, similar to how the EA tremolo was originally designed to be a preamp as well as a tremolo (the switch disconnected the LFO).

[tubegeek]

.
(following)

[deadastronaut]

+1..

yep, ive noticed 2 leds in series are better than 1 too with my lfo experiments, and depending on the type/colour too...always handy for indicator bling though.

[samhay]

Cool - something new and wobbly i the works Jon?

I suspect you can do 'Plan A' with a dual op-amp like this:

[midwayfair]

Cool - something new and wobbly i the works Jon?

I suspect you can do 'Plan A' with a dual op-amp like this:

Thanks, Samhay. Tried it just now. This does make them blink out of phase, but the waveform is VERY hard (even without the wave pot involved) with two LEDs, so it works best without the extra rate indicators (not sure what to do about that ... maybe try one in parallel like the lune). A C taper might also work best for the depth; it was almost imperceptible until about 400K, but that might depend on the vactrols used. You still need an offset trimmer, as well, for the U2 Vb connection, which could be done with a 100K trim the way you have it. I had to adjust it quite a bit off center to get them to balance properly.

I think the extra buffer is smoothing out the waveform in the quad, but depending on the effect it's used in, I could see yours being equally effective.

Not doing something new, exactly. I'm revising something old.

[samhay]

Should have mentioned that I would use an A1M depth pot with this design.
I used an inverting op-amp in my Anharmonic tremolo, but this doesn't have a waveshape control, and I guess the 1M impedance is loading it down a bit.

[midwayfair]

I guess the 1M impedance is loading it down a bit.

Can you explain further? I wouldn't have expected the 1M to matter, since the 500K is still panning between the two oscillator outputs, or is that the 500K wave pot is comparatively a much smaller resistance between the two outputs at any given time than the "no" resistance without the 1M there? (More of my op amp ign'ace here, I suspect.) I thought that the square wave output ended up looking about the same.

----

I was thinking another interesting use for this setup would be a dual filter sweep, if you replaced ICA with an envelope. The depth control would be an effective intensity control that's different from the typical threshold control at the input of the envelope, and a balance pot would be an interesting external control to make the filter sweeps different. I can even imagine some sort of dual compressor design, where you can pan between two different methods of attenuation with external indicators -- make one LED set a gain control for feedback compression and the other LED set a volume control at the ouput for feedforward compression. Or compressing different frequency bands.

[midwayfair]

I have some questions about the waveform generation.

The rate pot is 2.2K (or slightly less) in series with a 100K pot, forming a low-pass with the 10uF in the feedback loop with R15. The triangle setting is tapped from there, but it's a softened triangle, or at least it seems to be. And obviously when there is more resistance total between pin 3 and pin 10, the wave is less of a PULSE.

Now, I know I can play around with the ratio of the rate pot, the 2.2K, and the capacitor to use a different pot value, like 22K + 1MC + 1uF. this doesn't seem to affect the waveform that I can see, though.

--Is there a relationship between the values of the components as a whole and how soft the triangle wave output at pin 1 is?
--Can the wave be softclipped inside the oscillator, or does This Kill the Oscillation? The Tri-Vibe clips it to Vb after the second stage, but that just makes the wave lopsided that I was able to see, at least in this setup.