LFOs - part I

LFO stands for Low Frequency Oscillator. We use LFOs in modulation and delay pedals to vary some parameter, such as phase, volume, delay time, etc., to create movement and give our guitar signal more life.

Oscillators are a really interesting subject and they're all around us. You're probably wearing one on your wrist. For this forum I'll focus on the two types of oscillators we use as LFOs in pedals:
Integrator-schmitt trigger oscillators make triangle waves and square waves.
Phase shift oscillators make sinewaves.

In this article I'll focus on the integrator-schmitt trigger oscillator and save the phase shift oscillator for part II.

The integrator-schmitt trigger oscillator primarily makes square waves and triangle waves. With a few extra parts, it can make psuedo-sinewaves, sawtooth and pulse waveforms.
Here is a great visualization of an integrator-schmitt trigger oscillator in action.
https://www.falstad.com/circuit/e-triangle.html

The advantages of this type of oscillator are:
1. It's guaranteed to start and run.
2. The frequency adjustment range can be huge and very low frequencies are possible.
3. The frequency and amplitude are very predictable & repeatable.
4. It doesn't take very many parts to implement.

There's really only one disadvantage:
1. It's prone to ticking.

The integrator-schmitt trigger oscillator shows up in pedals like the EQD Sea Machine, Chaos Machine, Keeley Magnetron, MXR Phase 90, MXR Electric Mistress, Catalinbread Pareidolia, DOD FX75, Keeley Magnetic Echo, Boss CE-2, etc.

It's a good oscillator as long as we can mitigate the ticking. Guitar signals are small, on the order of a few hundred mV and the human ear is very sensitive. A disturbance as small as a few millivolts will be audible. It's important that things inside the pedal don't make any unwanted sounds.

Referring to the falstad webpage, the opamp on the right is an integrator and the opamp on the left is a schmitt trigger. The integrator ramps it's output voltage up or down at a rate determined by the upper 10K resistor and the capacitor. Replace the resistor with a pot and we have a Speed control. A schmitt trigger is a comparator with two thresholds: on for turning on and one for turning off. Whenever the triangle wave hits one of those thresholds, the schmitt trigger switches rapidly from on to off or off to on, which commands the integrator to ramp the triangle wave in the other direction. In this type of LFO, the voltage on the timing capacitor reverses every half cycle, so it's important that this capacitor is non-polar. If the LFO is intended to run very slowly, say longer that 10 seconds to go around once, then the capacitor needs to be low leakage because the charging and discharging currents are very small. A film cap is the best choice. Not all opamps are good for slow LFOs. The FET input opamps, like TL072, are good because their input bias current is small enough to not affect the operation. LM324 is another good choice for all except the slowest LFOs because the input currents are sufficiently low and the outputs swing close to the rails.

The Tick
Rapid switching in the schmitt trigger is the source of the tick. With opamp-based schmitt triggers, the output slews as fast as it can from one rail to the other. There is a spike in power supply current during the slew and that can spread all over the board thru the power and ground traces. Also, any signal traces or wires running near the schmitt trigger or Speed control can act as antennas and pick up the high frequency content present in the rapidly switching schmitt trigger output. The LFO may be switching at 1Hz or less, but the fast edges on the output of the schmitt trigger contain frequencies across the audio spectrum.
The solution is to slow down the schmitt trigger's switching speed so it doesn't draw large currents during switching or radiate high-freq noise. The "tick fix" I proposed several months ago does just that. The downside, if you want to call it that, is the peaks of the triangle wave get rounded off a little and the triangle amplitude changes a little bit when the frequency is changed.

There is another "tick fix" out there that is just plain dumb. It shows up in MidFi Pitch Pirate Deluxe. The output of the schmitt trigger is loaded with a large capacitor. That capacitor might slow down the schmitt trigger's output, but it causes large power supply current spikes. It's never a good idea to load the output of a conventional opamp with a capacitor because that capacitor could make the opamp oscillate, introducing more extraneous noise into the guitar signal.
 
Last edited:

K Pedals

Well-known member
LFO stands for Low Frequency Oscillator. We use LFos in modulation and delay pedals to vary some parameter, such as phase, volume, delay time, etc., to create movement and give our guitar signal more life.

Oscillators are a really interesting subject and they're all around us. You're probably wearing one on your wrist. For this forum I'll focus on the two types of oscillators we use as LFOs in pedals:
Integrator-schmitt trigger oscillators make triangle waves and square waves.
Phase shift oscillators make sinewaves.

In this article I'll focus on the integrator-schmitt trigger oscillator and save the phase shift oscillator for part II.

The integrator-schmitt trigger oscillator primarily makes square waves and triangle waves. With a few extra parts, it can make psuedo-sinewaves, sawtooth and pulse waveforms.
Here is a great visualization of an integrator-schmitt trigger oscillator in action.
https://www.falstad.com/circuit/e-triangle.html

The advantages of this type of oscillator are:
1. It's guaranteed to start and run.
2. The frequency adjustment range can be huge and very low frequencies are possible.
3. The frequency and amplitude are very predictable & repeatable.
4. It doesn't take very many parts to implement.

There's really only one disadvantage:
1. It's prone to ticking.

The integrator-schmitt trigger oscillator shows up in pedals like the EQD Sea Machine, Chaos Machine, Keeley Magnetron, MXR Phase 90, MXR Electric Mistress, Catalinbread Pareidolia, DOD FX75, Keeley Magnetic Echo, Boss CE-2, etc.

It's a good oscillator as long as we can mitigate the ticking. Guitar signals are small, on the order of a few hundred mV and the human ear is very sensitive. A disturbance as small as a few millivolts will be audible. It's important that things inside the pedal don't make any unwanted sounds.

Referring to the falstad webpage, the opamp on the right is an integrator and the opamp on the left is a schmitt trigger. The integrator ramps it's output voltage up or down at a rate determined by the upper 10K resistor and the capacitor. Replace the resistor with a pot and we have a Speed control. A schmitt trigger is a comparator with two thresholds: on for turning on and one for turning off. Whenever the triangle wave hits one of those thresholds, the schmitt trigger switches rapidly from on to off or off to on, which commands the integrator to ramp the triangle wave in the other direction. In this type of LFO, the voltage on the timing capacitor reverses every half cycle, so it's important that this capacitor is non-polar. If the LFO is intended to run very slowly, say longer that 10 seconds to go around once, then the capacitor needs to be low leakage because the charging and discharging currents are very small. A film cap is the best choice. Not all opamps are good for slow LFOs. The FET input opamps, like TL072, are good because their input bias current is small enough to not affect the operation. LM324 is another good choice for all except the slowest LFOs because the input currents are sufficiently low and the outputs swing close to the rails.

The Tick
Rapid switching in the schmitt trigger is the source of the tick. With opamp-based schmitt triggers, the output slews as fast as it can from one rail to the other. There is a spike in power supply current during the slew and that can spread all over the board thru the power and ground traces. Also, any signal traces or wires running near the schmitt trigger or Speed control can act as antennas and pick up the high frequency content present in the rapidly switching schmitt trigger output. The LFO may be switching at 1Hz or less, but the fast edges on the output of the schmitt trigger contain frequencies across the audio spectrum.
The solution is to slow down the schmitt trigger's switching speed so it doesn't draw large currents during switching or radiate high-freq noise. The "tick fix" I proposed several months ago does just that. The downside, if you want to call it that, is the peaks of the triangle wave get rounded off a little and the triangle amplitude changes a little bit when the frequency is changed.

There is another "tick fix" out there that is just plain dumb. It shows up in MidFi Pitch Pirate Deluxe. The output of the schmitt trigger is loaded with a large capacitor. That capacitor might slow down the schmitt trigger's output, but it causes large power supply current spikes. It's never a good idea to load the output of a conventional opamp with a capacitor because that capacitor could make the opamp oscillate, introducing more extraneous noise into the guitar signal.
Awesome info!!!
 

cooder

Well-known member
Fabulous, Chuck! Just back from holidays without internetness I'm delighted to discover the boneyard....! What a source of head explosions and enlightenment. ;)
Could you refer to the tick fix or put a link up, I have missed that. Cheers!
 

Chuck D. Bones

Circuit Wizard
Could you refer to the tick fix or put a link up, I have missed that. Cheers!

Hey Cooder, try here: https://forum.pedalpcb.com/threads/how-to-fix-the-lfo-tick-in-the-sea-horse-and-dark-rift.2360/

This fix works on all of the Schmitt-Integrator LFOs we've tried. The ones that already have a resistor in series with the schmitt's inverting (-) input pin are the easiest to do. If you have to add that resistor, then you'll need to cut or bend the IC pin and install the resistor between the IC pin & the pad.
 

cooder

Well-known member
Cheers Chuck! I actually saw that one earlier but wasn't sure if that's the one you were referring to, thanks for clarification.
Also: could you elaborate on what you would recommend as the best practise with ground and power planes on pcb designs? So far I have used a solid pour ground plane on bottom of pcb, no plane on upper side of pcb. Power connections just by traces. Do different pours have advantages in lower noise or less chance of inducing ticking? Just wondering because the newer pedal pcbs seem to have a pour on top layer, not solid but cross patterned or so. What's your take and does it depend on what circuit?
 

Chuck D. Bones

Circuit Wizard
I look at it this way...
From an electrical perspective, more ground is better. It acts as a shield between traces and the more copper there is, the lower the resistance. From a thermal perspective, ground planes act like heatsinks and it makes soldering to the ground more difficult. That's why wagon-wheel pads were invented. We still have the ground plane, but the heat flowing out of the ground pads into the plane is reduced by the spokes. The cross-hatched ground plane is just an extension of the wagon-wheel concept. Still acts like a low-resistance ground path & shield, but there is less thermal conductivity, which facilitates soldering. And it looks pretty. My advice is pour as much ground as you can, but don't feel like you have to fill in every little nook & cranny. Definitely make sure there aren't any floating islands. If you can't easily tie a poured area to the overall ground, then either move something in your layout, add a via or just don't pour there.

As for ticking or any other kind of interference, you have three methods of dealing with the problem:
1. Reduce the noise at the source.
2. Reduce the coupling paths, including shielding.
3. Make the circuit less sensitive to noise.
Sometime you have to do all three. Always start with #1, because if the noise isn't there in the first place, you don't have to do #2 or #3. To minimize coupling, keep noisy traces away from sensitive traces. Same goes for off-board wiring. If there is anything you can do to make your circuit less sensitive to interference, do that too. Reducing impedances can help. High impedance circuits pick up noise more easily than low-impedance ones. I see pedal designs that mimic tube amps right down to the 1Meg pots. I think you know the ones I'm talking about. In many cases the impedances could be lowered 10x. Don't build in more gain than you need. Limit the high-freq gain, anything above 20KHz, because radio stations, computers and cell phones are radiating RF that will sneak into your pedals and cause havoc.
 
Top