The technology of the LFO

[JustinFun]

Firstly, apologies if I've missed an obvious thread or web-link on this. I did look, but couldn't find one. I've been building pedals for a couple of years now and mostly 'get' how amplification and clipping circuits work, thanks in no small part to the excellent 'technology of the fuzz face/ tubescreamer articles, DIYStompboxes and the gausmarkov site.

Where I feel completely in the dark though is in the area of modulation and LFOs. There seem to be a few different building blocks out there - some optical, some just transistor based. What they all have in common is that I don't know how they work!

So... can anyone recommend some basic reading matter on LFOs? Also if possible the envelope tracking segments of the auto-wah (another mystery to me at the moment).

Thanks.

[John Lyons]

What they all have in common is that I don't know how they work!

Yeah, me as well for the most part. I'll be watching this thread.

[.Mike]

...thanks in no small part to the excellent 'technology of the fuzz face/ tubescreamer articles...

Also if possible the envelope tracking segments of the auto-wah (another mystery to me at the moment).

This will get you started on envelope followers: The Technology of Auto-Wahs / Envelope-Controlled Filters.

For LFOs, I'm interested in learning the basics, too.

Mike

[merlinb]

An LFO is just an oscillator. The two common types used in stompboxes are the single opamp Schmitt trigger oscillator / relaxation oscillator, and the two-opamp triangle oscillator. You can read about them in any general electronics textbook.

After that it's a broader question of how you make part of a circuit voltage controllable. That's much more application specific.

[Jazznoise]

An LFO is just an oscillator. The two common types used in stompboxes are the single opamp Schmitt trigger oscillator / relaxation oscillator, and the two-opamp triangle oscillator. You can read about them in any general electronics textbook.

After that it's a broader question of how you make part of a circuit voltage controllable. That's much more application specific.

This. It's just a control signal. But let's have a few examples:

Optical Tremelos and Compressors use LDR's. JFET based Voltage Controlled Resistors are used by circuits like the Phase 90. The old Arp synths used an XOR gate for a square-wave ring modulator effect (Possibly the new Zvex trem too, or certainly something similar!) and we've a few resident experts on logic gate abuse here. Digital Potentiometers can be fed tables of data from to vary the given signal (I built a modest Ring Modulator around the concept, but I'd honestly say they're not worthwhile as in-line signal processors).

http://www.fairchildsemi.com/pf/H1/H11F3M.html << These are an underexplored set of trinkets (and for reasonably cheap) that I've picked up a set to look at. They seem promising! They do, however, look little creepy little maggots. The white packaging is..bizarre.

[JustinFun]

Thanks all, that's really helped. Going to have to do some reading up and then some playing to understand this better, I think.

[R.G.]

Just to amplify ( ) on the previous comments a bit, LFOs are nearly never the problem - it's what you're controlling with the LFO. As Merlin alludes, there are many different control elements, and each one needs its own kind of drive from an LFO to be effective at what it does.

More precisely, each kind of controlled element needs a unique combination of LFO signal characteristics
- "dc" offset
- peak to peak LFO signal size
- LFO control variations, in the sense of the LFO signal moving up from a minimum idle value, down from a maximum, or both up and down about the DC offset
- control by a voltage, a current, or some combination of voltage and current

These oddities come from the fact that almost all LFO-controlled elements do what they do as a side effect or imperfection of their otherwise normal operation. It would be nice if there was an isolated, voltage controlled resistor. These don't seem to exist as a single element, so they're made up by combining things to get more or less close to the ideal. The need for isolation comes from the fact that we usually don't want to listen to the wocka-wocka-wocka-wocka of the LFO, we only want to hear the signal as modified by the LFO. The ability to both modulate and not let the LFO bleed through to the audio is a huge factor in selecting a control element.

In fact, selecting a control element or circuit for low bleed through of the LFO and effective modulation range is usually where the circuit design starts. It's only once these questions get answered that the actual LFO itself gets designed, and that's done to keep the control element happy and minimize unpleasant side effects. I suppose, (...he said, taking another sip of coffee and wakening a bit more    ) that one way to look at it is that the selection and design of an LFO is a trivial wart on the side of the issue of selecting a controlled element to drive with the LFO.

[R.G.]

The issue of control elements is the real core of this.

In most cases, as a circuit designer, you'd be really happy with a voltage controlled resistor with no funny side effects. Sadly, these are hard to come by. The thing that comes closest to the ideal is the modulated light source coupled to an LDR, and this is why you see so many of these.  An LDR has quirks of its own (speed of response, distortion, light- and dark-adaptation, response sensitivity, linearity of resistance per unit of light... ) but it's really good at isolation and not having ugly effects on moderate levels of signal.

Optical isolation of the actual element controlling the signal is a huge step toward not getting control feedthrough.  It used to be that you had to use an incandescent lamp or a mechanical shutter to vary the light to an LDR, but with the advent of LEDs, this got easier. Morely used mechanical variation of the light on an LDR. The Univibe is the best-known example of an incandescent lamp varying light to an LDR. And the LED/LDR package we use so much is the modern version. Incandescent lamps vary their light by the power applied to them, and do so in a complicated way, as their light output is a function of the thermal emission of heat from their filaments, and the heat is a function of the product of the square of the RMS voltage across them and their own internally varying resistance as their temperature changes, and also that the light color out varies with temperature too. All of these variations are so nonlinear that an equations-up version of an incandescent lamp LDR control is nearly impossible to solve in closed form.

LEDs are much simpler. LEDs come very close to light output being linear with the LED current. So once you get the LDR doing what you want as a resistor in the controlled circuit, you can concentrate on a much simpler control current to make the LED make the light make the LDR do nearly what you were trying to do in the first place. 

There exist many controlled elements. FETs come close to a variable resistance in their channel if the voltage across the channel is small enough to never get near the channel pinchoff voltage. Accordingly, we see a lot of JFETs used in effects as variable resistors, as witness the Phase 90. JFETs have quirks ( ... he said parenthetically and comically to people who have fought these quirks) of their own, but kinda work. There is even an optically coupled photo-FET, and a commercial product combining an LED with a photo-FET to make an optically isolated FET based resistor, the H11F1/2/3.

Another is the forward resistance of a diode. A junction diode goes from high resistance to low resistance to small AC signals as its forward voltage goes from just barely not conducting current to lots of current levels. Since this is only about 0.1 to 0.2V, it doesn't sound promising for a modulator, but if you keep signals low enough, everything looks linear, and the diode is no exception. For signals up to about 25mV, using a diode for a modulator gives distortion under a few percent, and it sounds OK for effects work. Obviously, this requires feeding the diode with a control current to modulate its conduction, and taking the signal across the diode. A diode bridge circuit works well, and in fact the diode bridge modulator and diode ring modulators are used a lot in radio work where the signals are tiny. Thomas Vox amps use a diode bridge modulator for tremolo to good result.

Ordinary bipolar transistors have variable resistance between collector and emitter, and for similarly small signals, they can be used. This is the basis of the EH Pulsar tremolo and others.

If you decide to do a good, large signal/low distortion/highly linear modulator, you wind up doing something like an OTA or a ladder modulator. These are generally current controlled and have a differential form to cancel some of the feedthrough of control signal. The circuits quickly get complex for this.

There is a large body of information on modulators and how to control them that turning over the LFO rock uncovers. Some of the things under there are not pretty.