Waveshape, timbre, etc.

I'm building a dead-simple theremin of my own design, using discrete JFET oscillators with BJTs for mixing and gain. Two oscillators at 600kHz, a BJT mixer, gain stage, and germanium diode + lowpass, clean and simple. So far, I have the thing on a breadboard, with a random wire that changes the pitch when you move your hand near it. It actually sounds pretty decent, producing a sine wave that is clipped on one side. I'll probably start figuring out the volume section next, I guess I need a higher frequency osc and some way to convert frequency to a control voltage for a variable gain amplifier.

The part I don't get is what the various "waveshape", "timbre", "brightness", and so on controls do on the pro models. I suppose being able to go from sine to square would be nice, but what do all those funky controls do to the signal, exactly?

I'll definitely have a pot operated limiter stage to square up the signal on demand, but doing voltage-controlled filters with discrete transistors seems kinda hard.

The most desirable place to control timbre is in the RF osc side, prior to the mixer. Beyond that point you might as well just go out and buy a guitar effects processor.

The waveshape of the RF oscillators controls timbre for the most part. Very sinusoidal = whistle-like or maybe high-tenor opera singer if you prefer.

More squared-off of a waveshape is going to be full of odd harmonics, so it will sound raspy. A similar thing will happen for a sinewave clipped off at the top.

A more triangular or sawtooth-like wave will give a different timbre.

As for how to do it ... bias of your JFETs will control distortion to some extent. You can also apply filter networks prior to the mixer stage.

The other trick often used is to provide some "loose coupling" between the fixed and pitch oscillator circuits by means of a wire connected to one oscillator placed somewhere in the vicinity of the other oscillator. These are sometimes called "gimmick loops". This will create some additional harmonics and will affect the waveshape of the heterodyned signal that is fed to the mixer stage.

Hope that helps.

Don

"The other trick often used is to provide some "loose coupling" between the fixed and pitch oscillator circuits by means of a wire connected to one oscillator placed somewhere in the vicinity of the other oscillator." - Don.

One can make the coupling a DC control function (so it can be adjusted by user potentiometer) by coupling the reference and variable oscillators through a vari-cap diode. Some of the Silicon Chip / Jaycar / EPE Theremins use this method.. There is an article somewhere here at TW which describes modifications to the SC Theremin and presents schematics showing the operation of various tone controls.

The function of waveshaping / tone adjustment has been extensively discussed here at TW, at every level of technical depth.. Worth a search .. Also, The Moog article on modding the EW (The bible) gives some info on the waveshaping functions.. "How do I distort thee" is one of the threads that jumps to mind.

Fred Mundell
Fundamental Designs Ltd.
Electronics Consultant.
<- See Profile Image for Email.
Designer of Theremins and other alternative electronic music controllers and instruments.

Hmmm... So if I have a varactor between oscialltors, how can I bias it? Seems like they usually have the anode grounded, and I suppose I would need the varactor between two points on the oscillators to add coupling.

One thing I just thought of is coupling the antenna oscillator to the pitch adjustment varactor in my reference oscillator. I wonder what phase noise sounds like...

Simplest biasing is shown here..

[img]http://www.therasynth.com/assets/images/varicap2.JPG[/img]

Or here is link to my page Scrapbook page (http://www.therasynth.com/html/temp.html)

C1 and C2 sized for the maximum coupling you require (make them 10pF and absolute maximum coupling will be less than 5pF) - Capacitance of diode will be the key factor determining coupling.. Any diode can be used - 1N400x Power diodes are particularly good for higher capacitance. NOTE - The configuration shown WILL distort the waveforms -particularly - If the amplitude of signals is high.. As these signals will dynamically alter the coupling... Use double diodes (K-K,or A-A at their common point) to minimise distortion.. but note that doing this halves the effective capacitance and one needs double biasing resistors.

Make the voltage range across the diodes as large as possible.. Fixed -12V with a variable voltage from -11V to +12V gives good control with 1N4007 Diodes.. One can place as many diodes in parrallel as one requires to increase capacitance.

Fred Mundell
Fundamental Designs Ltd.
Electronics Consultant.
<- See Profile Image for Email.
Designer of Theremins and other alternative electronic music controllers and instruments.

Some more..

[img]http://www.therasynth.com/assets/images/varicouple3.JPG[/img]

[b]Edit 17 Sept.. Corrected error in controlled distortion circuit (removed C5). [/b]

(same page as in last posting)

This shows (a) using 2 diodes for reducing distortion, and (b) adding another diode + biasing to give voltage controlled distortion..

Note, the last circuit (b) will always give more distortion than (a), but will give the same degree of distortion as the original single diode circuit when its control voltage is low, and almost no distortion when its CV is high.

Fred Mundell
Fundamental Designs Ltd.
Electronics Consultant.
<- See Profile Image for Email.
Designer of Theremins and other alternative electronic music controllers and instruments.

"One thing I just thought of is coupling the antenna oscillator to the pitch adjustment varactor in my reference oscillator. I wonder what phase noise sounds like..."

Interesting! .. It would not, I think, be "noise".. But would generate FM components.. However, as these oscillators are so close - you could end up with a novel way of heterodyning the 2 signals.. Hmmm.. Something someone with better maths understanding could probably answer quickly..

From what I have seen, any signal coupling between oscillators running at frequencies close to each other (and Theremin oscillators run extremely close.. 10kHz maximum difference at frequencies of 250kKz+ ) has the effect of causing them to synchronise.. This is exactly what one is aiming for - in a controlled manner.. as the oscillators pull towards the 'lock' state, the waveforms of BOTH change.. The trick is to get this distortion in a controlled way which produces musical harmonics, and to avoid the oscillators going into full lock too soon.. At full lock, both oscillators are synced and at the same frequency and there is no audio output.. One does not want this to happen until the difference frequency drops below 16Hz.. Also - As the oscillators pull each other towards lock, the law controlling the distance / pitch relationship changes.. This is used deliberately on the EW to linearize the low frequency end of the playing field - and messing with oscillator synchronisation affects overall linearity quite strongly..

"I'll definitely have a pot operated limiter stage to square up the signal on demand, but doing voltage-controlled filters with discrete transistors seems kinda hard."

Its not too hard! - Dont forget, you are dealing with HF not audio - And anything you do to reduce harmonics on either HF oscillator pre-heterodyning will have an effect on the harmonics of the audio output.. So reducing the harmonic content of one oscillator with a VCF will make an audible difference on the output signal.

Using the diode methods shown above, a simple VCF can be implemented.. Here is circuit which befaves like a grounded capacitor whos capacitance is DC controlled.. You can use this in a simple RC configuration to make an extremely simple low pass filter for high frequency waveforms - or you can expand this by adding as many poles with transistor buffers as you require.

Maximum capacitance occurs when the voltage is lowest (say -11V when the -ve reference is -12V).

[img]http://www.therasynth.com/assets/images/Varicap_to_GND.JPG[/img]

Fred Mundell
Fundamental Designs Ltd.
Electronics Consultant.
<- See Profile Image for Email.
Designer of Theremins and other alternative electronic music controllers and instruments.

Wow, thanks!

my pleasure!

The rule of thumb is this.. Every diode must have a path for DC current to flow* (well - it doesnt really 'flow' as the diodes are reverse biased.. but..).. All diodes must always be reverse biased* - even if only by a few mV.

There should be no DC path for any signal.

*Forward biasing can occur as a result of the AC signal being coupled to the diode exceeding the DC bias level - This is what causes major distortion in the single diode circuit, and can be implemented with the controlled distortion circuit.

Distortion occurs even if diodes are not forward biased, due to the fact that the capacitance will change as a function of the TOTAL potential across them.. It can be seen that the greater the AC amplitude, the greater the distortion of all kinds.

In the 2 diode circuit, a decrease in capacitance as a result of AC signal across one diode, will be compensated by an increase in the capacitance of the other.. NOT perfectly, but certainly better than the single diode configuration.

If you have a good 'scope, and start messing about with variations on these diode circuits - particularly if you have harmonically rich waveforms - you may find that you dont notice the months slip away, as you sit absorbed and intregued at what you can do with a few diodes, resistors, capacitors and a power supply! ;-))

If you want to generate pleasant 'distortion' you need to have large amplitude signals - If you want severe distortion, increase signal levels to greater than the bias voltages (or reduce the bias voltages by making the -V ref more positive and reducing the +Ve bias voltage..

Also.. It is VOLTAGE NOT CURRENT one is using to control the capacitance.. The value of the bias resistors do not actually affect the capacitance - but they do affect the impedence seem by the AC signals.. Too low in value, and they will load the AC signals fed to the network - High values (>100k) are desirable for this reason.

One can also use these circuits to implement waveshaping locally at the oscillator (ie - not as a circuit for coupling oscillators) - you can, for example, square the oscillator waveform and feed this square wave back in a controlled manner, thus changing the oscillator waveform and behaviour completely...

Fred Mundell
Fundamental Designs Ltd.
Electronics Consultant.
<- See Profile Image for Email.
Designer of Theremins and other alternative electronic music controllers and instruments.