Wednesday, December 31, 2014

The New Xanu Oscillator - Breakdown and Explanation

The last time I posted here about a VCO, I forgot to mention that it's not very good in most applications as it was only linear FM and the waveforms were very... not right, I'll say. My latest design is somewhat complex, but can be simplified as it has eight different waveforms, a ton of different kinds of modulation (including an accurate 1 volt per octave FM input), and a very modular design made up of simple parts. Let's get started!

The Power:

As this is an electronic device, it needs power.
Most modular power supplies have four rails: main positive and negative voltage, a +5V rail, and ground. However, for this and other designs, a -5V rail is necessary. The easy way to make this rail is to power an inverting amplifier with the main rails and run the +5V rail through it. Now you have a -5V rail. Easy enough?

The FM Inputs:

For the inputs, you have manual frequency, exponential frequency, 1v/oct with built-in portamento (glide), and linear FM. This can be used on most VCO circuits, not just this one.
How it works: The three exponential inputs (manual frequency, exponential FM, and 1v/oct) are mixed together with the array of 100k resistors. The pot on the exponential FM input controls the level, and the manual frequency pot acts like a simple crossfader between the positive and negative 5V voltage rails. The exponential inputs are then attenuated slightly and run through a trimpot, which adjusts the 1v/oct tracking accuracy. That final signal is then sent into the final stage called an exponential converter which, as you might guess, converts the more or less linear voltages into exponential voltages, thus allowing you to control linear VCOs with exponential voltages. After that, the linear FM input is mixed into the exponential signal.

The part in the box is the built-in portamento (aka glide) section. It is made up of two buffers with an RC lowpass filter in between. Lowering the cutoff via the rheostat (or pot, whatever you use) smooths out the stepped voltage signal, thus making the voltage 'glide' to the next voltage value.

The Core:

The core of this new VCO is a simple Schmitt-trigger ramp (inverted saw) oscillator with linear FM capable of a wide range of frequencies.
How it works: A Schmitt trigger can be configured as an oscillator by using a capacitor connected from the trigger's input to ground and a resistor (or in this case, a diode acting as a resistor as it would in a diode filter) connected from the trigger's output to its input. This configuration makes a square wave with a very thin, nearly inaudible pulse width (because of its quietness, it's not counted as one of this oscillator's waveforms). Feeding that signal back to the beginning not only allows the trigger to oscillate by itself, but it also fills the capacitor with electricity when the square wave is high, and the capacitor depletes when the square wave is low. The unique duty cycle of this oscillator means that the capacitor's output forms a ramp wave.

Then you have the problem of getting the ramp output to other parts of the oscillator or even other parts of the synth. With too much output load (that is, too much circuitry after the ramp is extracted from the cap), the oscillator will lose stability. That is why both the square (trigger) output and ramp output are fed into Class-A transistor amplifiers - to act as buffers. They also make the signal very loud, so the resistors attenuate the amplified signal to a more manageable level.

Waveshaping the Basic Waveforms:

Like I said, this oscillator has 8 waveforms, one of which (ramp) has already been covered. Most good synthesizer oscillators have at least four waveforms: saw, triangle, sine, and square. This next section will show you how to get those waveforms.
How it works: the first wave, saw, is easy to get: invert the ramp wave and you have a saw wave. On most oscillators this doesn't matter as they sound identical. This VCO can go into LFO territory, however, so the different wave shapes are very useful. The triangle wave is formed using a simple comparator to get a 50% duty cycle square wave and an active lowpass filter. The output capacitor is for decoupling so you get an even waveform. For a sine wave, this oscillator uses an approximation based off the triangle and uses another active filter and large capacitor. For most of the oscillator's frequency range, the triangle and sine waves will actually form the proper waves despite the fixed filters.

The square wave section also uses a comparator, but instead of grounding the reference side of the comparator, there is a linear pulse width modulation section that allows manual and voltage control over the pulse width.

Waveshaping the Complex Waveforms:

By this point, we have the basic waveforms. But, every good oscillator has those, so why not add some flair with more unique waveforms?
The top section is a simple wavefolder that uses the sine wave as its input. The mangle pot controls how loud the sine wave is going into the folder - more volume, more folding. You could make this voltage controlled by feeding the sine wave into a small resistor then have the other resistor replaced by a transistor, vactrol, or optocoupler. The transistor section is what does the actual folding and acts a bit like a compressor except instead of limiting the volume, it folds the excess waveform over. After that, the signal splits off for an amplifier with a trimpot to adjust gain and a comparator which creates a ton of interesting square waves based off the folded sine wave. This square output is not counted as one of the main waveforms since the square wave shaper can sound very similar with PWM.

The bottom section is composed of two parts: the spike generator and the morpher. The spike generator is a simple RC highpass filter which could be voltage controlled if you wanted with either a transistor, vactrol, or optocoupler. The morph section splits the voltage into a normal diode and a zener diode. The voltage cutoffs of each of these make for an interesting waveform out of each diode. The pot acts as a crossfader between the two. This can also be voltage-controlled with a VC crossfader.

Putting It All Together:

Put all those parts together, and you have the Xanu Oscillator:
As one of the notes say, the trigger outputs do not go to the front panel, as explained earlier. This oscillator could do with a sync input, but otherwise I don't see what else you'd need or want in an oscillator. Of course, it could be narrowed down to using just the core with the triangle and square waveshapers for a simple, less feature rich oscillator. Or, you can add more things like yet more waveshaping (a saw animator would be interesting, as well as a suboscillator), a range switch to swap between core capacitors for ultra-low frequencies, more FM inputs... the sky is the limit, really.

I hope you enjoyed reading this, and if you have any questions, feel free to comment below. Oh and happy new year!

No comments:

Post a Comment