LFO circuit that doesnt tick in your audio path?

[Se7en_Costanza]

I need an LFO circuit, preferably a square, triangle or sinewave to control an LED/LDR type configuration in a tremolo effect.

I've tried 555 timers, schmitt triggers and clock ic's, which were all the worst for this so i steered clear from them, and moved to op amp comparator type ones (tremolous lune, valve wizard weird chorus ones i've tried) which work great in theory, but there is a constant ticking problem that makes its way in the audio path, now normally it's quiet enough to not notice it but the tremolo im making has a distortion in the output stage, so the ticking gets amplified intensely.

In terms of how i've reduced the ticking, ive tried everything forums have told me to 'cure' it, and it has reduced the noise significantly, but not enough. E.g. decoupling the lfo's and the audio paths power IC rails, having seperate bias voltages for the lfo to the audio op amp, and using low power op amps, altering the squarewaves rise and fall time with a capacitor, and still the ticking is there.

Is there an LFO circuit out there that has absolutely no ticking problems? or does anyone have a schematic for one that they've had success with in the past?

Any advice would be a huge help.

Tremulus Lune Schematic (Only the LFO)
http://www.tonepad.com/getFile.asp?id=84

Weird Chorus Schematic (Only the LFO)
http://www.valvewizard.co.uk/jennygreenteethschem.jpg

[GibsonGM]

The Trem Lune LFO shouldn't tick if you have built it right AND laid out your wiring well.   If you have the '100R/10000u cap trick' in place at its power source, and it's ticking...most likely parts/wiring layout is to blame.
 
I built an Easyvibe and moved the LFO offboard, put it up by the footswitch in a wah shell.  I routed the LED wires along the chassis and to the main board at right angles....not a single tick to be heard.    If you build it on breadboard, you probably can't get good results, either...

I think you can get the same result with the Trem Lune LFO if you are careful.    Each wire is an antenna...act accordingly and all that...

[Mark Hammer]

Boss implemented an interesting circuit, described here - http://moosapotamus.net/files/stompboxology-mo-tremlo.pdf - that uses the same basic 2 op-amp integrated square-wave LFO.  It adds just a tiny bit of lag to turn a pure square wave into a "lazier" square that takes a few milliseconds to rise and a few to fall.  Insomuch as the audible tick results from a sudden instantaneous draw on the circuit current (which decoupling and low-power op-amps are intended to address, by requiring less current, and ptoviding a "reserve" to counteract sudden current draw), making the current draw less sudden makes the consequences of that current-draw much less audible - essentially sub-audio.  Using low-power op-amps and decoupling is still a good idea, but this circuit makes it even better.

[anotherjim]

Several of our LFO designs aren't really optimised. In the TL, it looks like an existing single amp oscillator has had a buffer added, but look at the size of the resistors loading that buffer. A 1k "bypass" has been added across the LED's to reduce intensity. I'm not sure that was a good idea. If the LED is too bright, increasing series resistance is the way to do it. When you see resistors below 10k, you should ask if that is necessary -  sometimes it will be and unavoidable. However, a 10k path to ground on 9v circuits is close to needing 1mA, and that can add up if there are many. 1k paths (like in the TL) are x10 worse. You should bear in mind that high total current consumption will degrade the benefit of any supply bypass capacitors and increase interaction between audio and LFO supply paths.

Clicking mainly comes from the square wave switching current. To reduce the effect...

Use a low power op-amp. Audio grade amp low noise performance is not a concern.

Use high resistance timing networks >470k. Resistor noise is not a concern. Existing circuits can often be re-scaled. x5 resistor & 1/5 capacitors. It should be possible to avoid needing non-polarized electrolytic timing capacitors as film types (<1uF) can easily cover LFO speeds if the timing resistors are made bigger.

Don't flash the bypass LED from the square wave - use the triangle instead or low-pass filter the square with an RC filter.

Close couple and enclose LED/LDR so that LED current can be the minimum needed for maximum depth.

Don't directly switch a flashing LED using the stomp switch. There is a lot of capacitive coupling between the contact rows which the audio path will pick up.

Screen the audio in and out leads in the box if the I/O isn't pcb mounted.

Twist the wires to each audio control pot/switch tightly together.

Phase shift and bridge type oscillators naturally produce sine wave and can't click *. However, single control frequency range is limited without switching components. For Vibrato and Tremolo, these can do what most people want, but it's probably not going to go slow enough for Flangers or Phasors which is why those op-amp Schmitt/Integrator circuits are so common.

* They can click if you drive a LED too hard using insufficient series resistance, since as the threshold voltage of the LED is passed, there will be a significant step change in current draw.

[Se7en_Costanza]

How should an LFO and the Audio path be layed out on a pcb, im guessing as far away from each other as possible? or is it just certain parts of the circuit need to be away from each other, like the squarewave output or the power rails? unsure.

[Kipper4]

Tonepad pcb for Another Jims diagram.


http://fuzzcentral.ssguitar.com/tremulus/tremuluslayout.gif

[PRR]

> Boss implemented an interesting circuit, described here -

Consolidated to a 3x5 recipe card:

[Mark Hammer]

Thanks, buddy.  I owe you one.

[doug deeper]

What I do is, split V+ from your supply with two 10 ohm resistors each feeding it's own filter cap, one goes to LFO, the other signal.
Works much better then the single resistor method IMO.

[patrick398]

Hey doug, could you elaborate on this slightly. I've just build a lune as it happens and it's ticking...not an awful lot but it's there and i'd like to eliminate it completely. Do you mean isolating V+ (say giving it it's own track on vero) and then run two 10ohm resistors into (100u?) filter caps then one goes to pin 8 of IC2 and one goes to pin 8 of IC1?

http://tagboardeffects.blogspot.co.uk/2013/03/tremulus-lune.html

[GibsonGM]

Like a "Y", Patrick....at the branch, each side gets a 100R resistor (I prefer 100R but probably makes little difference, we're not talking big current here), with a cap to ground at that point from each resistor (R's don't run "Into" them, the caps are parallel to the R's, to ground).  I'd suggest oh 470u, could use 1000u (I'd try smaller first to save space. sometimes even a 100u will work). 

This creates a little voltage reservoir so that sudden changes in demand on the power supply (transients, caused here by the demand for current to suddenly take the LFO straight "on") can be met without disturbing the circuit in the form of a *click*. 

Changes like this are why some builders use perf board instead of vero - it's a lot easier!     

[duck_arse]

looking on those layouts, you might try moving the "ground" wire in from column 1 to column 8, and move the "9V" wire from column 18 to column 11, row 8. this will give a nearly star ground and supply.

[blast] of course, the layout I picked to look at was the centre one, the other two are different. oh, well. translate wot I said for the other layouts. [/blast]

[patrick398]

Thank you Mike and Stephen for the replies I was thinking of just tacking a little daughterboard on with the caps and resistors. Sorry for being a bit dense but is this what you mean? Obviously i'd still need to cut the jumper connecting pin 8 of both IC's. If just moving the ground and V+ wires would solve my issue that would be mighty fine but if not i don't mind cramming this little board in somewhere. This build turned out particularly tight for some reason.



Thanks again!

[duck_arse]

midboard, all 3 layouts, is a double link/single hole (tsk! tsk!). if you was to cut those links out, and replace them with the 100R's instead, then shit the 9V wire to the same strip as the double link hole, you'd only then need to string the electros between pin's 8's and 4's.

I dunno the state of your board, these mods might be easy, might make a real mess ......

[patrick398]

Great suggestion thanks for that. I never would have seen that, need to train my eye some more i think. Due to the fact that the build ended up being so tight however, i don't think i'd be able to squeeze these changes onto my board which is a real shame because it's a really neat solution. Would my daughterboard work do you think?
Thanks again

[duck_arse]

the daughterboard - might work, if supply filtering is the fixxe. might not help if large switching currents on the earth lines is the problem.

at the very least, moove your parts on the dortabord around so it will look, again, a little like a star ,,,, erm - thing.

shift R1 left to row 1, R2 right to row 5. C1 across to row 2, C2 across to row 4. then put the red, black, blue and green  wires at row 3. [you need to cut those wire links to the pin 8's on the board, tho.]

[Transmogrifox]

Another way to do this is to use something like a phase-shift sine oscillator that always makes gradual changes in currents.  You can amplify that and clip it like a distortion pedal to get a square wave shape.  This solves sine and square, anyway.

Tri is not as simple because it's typically a bang-bang oscillator like the circuit posted here.

One thing not mentioned is to put resistors into both ground an VCC with a big capacitor filtering the LFO.  The main point is to get the switching currents from the local capacitor and avoid getting them from upstream.

You might also consider ElectricDruid's TAPLFO chip.  I don't think it generates the bang-bang currents, so the filtering problem is less severe -- you only have to take off the high frequency PWM stuff.

[ElectricDruid]

You might also consider ElectricDruid's TAPLFO chip.  I don't think it generates the bang-bang currents, so the filtering problem is less severe -- you only have to take off the high frequency PWM stuff.

That's true. It *might* produce some noise, but it's more likely to be MHz ultrasonic digital hash from the processor. Nowhere near as audible as DC clicks, and easier to filter if necessary. Although I've never had trouble when mixing up PICs and audio from digital noise getting into the audio via the power rails.  I've also recently done the VCLFO10 (an updated version off my earlier now-long-in-the-tooth VCLFO 9D) which includes a waveform smoothing feature. This applies a digital RC filter (two pole IIR) to the waveform before it leaves the chip to help remove any clicks. I'm planning a new version of the TAPLFO to include similar features - TAPLFO 3.
I also have a STOMPLFO chip in the pipeline, which is an 8-pin version of the TAPLFO designed specifically for stompbox use. Lots of waveforms, no clicks, tap tempo, dead easy to use. That's the plan. I've got it all done bar the tap tempo and adding an example circuit to the datasheet. I was going to demonstrate the features with a LM13700 SVF filter, but I've had a nightmare with the filter! The LFO is fine, but the filter makes some horrible thumps when the cut-off goes into sub-audio...oh the irony...
I could do a tremolo of course, but I wanted something different from the TAPLFO's demo circuit.

Anyway, I've got lots of new stuff lined up for the new year. It's going to be busy.

Tom

<edit>Sorry, that post became a bit "news from electric druid". But the truth is that it so happens that lfo's that don't tick is something I've been thinking about and working on a good deal lately.

[Rob Strand]

Boss implemented an interesting circuit, described here

Very interesting motivation.
Initially I thought those parts helped the circuit switch reliably with slow opamps.

[amptramp]

A separate power supply for the oscillator and other circuitry may be the easiest solution.  If you operate off isolated supplies you could eliminate clicks.  This may require two batteries, two One-Spots or two sections of a spider supply.

A pure sinewave oscillator at low frequencies is also a possibility.  It has no abrupt current load transitions.