Skip to content

Phase Splitting

Bloodhoney

At its core, phase splitting is the process of taking a single input AC audio signal and dividing it into two separate outputs that are identical in amplitude but 180° out of phase with each other. When one signal is at its positive peak, the other is at its exact negative peak. Mathematically, if your input is \(V_{in}(t) = A \sin(\omega t)\), our two outputs are:

  1. \(V_{1}(t) = A \sin(\omega t)\)
  2. \(V_{2}(t) = A \sin(\omega t - \pi) = -A \sin(\omega t)\)

In audio engineering, this is done for three primary reasons:

  1. Driving push-pull amplifiers;
  2. creating balanced audio lines for noise cancellation;
  3. and—specifically in the pedal world—achieving full-wave rectification to double the frequency (the classic analog octave-up effect).

Here is a deep dive into how phase splitting is achieved across different domains.

1. Magnetic Phase Splitting (The Audio Transformer)

This is the old-school, passive method used in vintage fuzzes like the Tycobrahe Octavia, utilizing something like the 42TM011 transformer we discussed. You feed the input signal into the primary winding. This induces a magnetic flux in the core, which in turn induces a voltage across the secondary winding. To split the phase, the secondary winding must have a center tap tied to a reference voltage (usually ground or a specific bias point). Because the center tap is anchored at \(0V\), as the magnetic field pushes the top of the secondary winding positive (e.g., \(+1V\)), it pulls the bottom of the secondary winding negative (e.g., \(-1V\)) by the exact same amount. You now have two signals, perfectly inverted relative to the center tap.

2. Active Analog Phase Splitting (Transistors & Op-Amps)

Transformers are bulky, expensive, and introduce their own inductance and parasitic capacitance. Modern circuits often use active components instead.

The Split-Load (Cathodyne) Inverter

Using a single BJT (like a 2N3904), you apply the signal to the base. You then place identical value resistors at both the collector and the emitter.

  • The signal taken from the emitter follows the input (in-phase).
  • The signal taken from the collector is inverted (180° out of phase). While elegant, the output impedances of the collector and emitter are vastly different, which can unevenly load subsequent stages.

The Dual Op-Amp Approach

For high-fidelity audio design, relying on a premium dual op-amp (such as the OPA2134) yields pristine phase splitting with perfectly matched impedances.

  • Op-Amp A: Configured as a standard non-inverting unity-gain buffer. It outputs \(V_1(t)\).
  • Op-Amp B: Configured as an inverting amplifier with a gain of precisely \(-1\) (using matched precision resistors in the feedback loop). It outputs \(V_2(t)\).

3. Phase Splitting in the Digital Domain (DSP)

In the digital realm—whether routing arrays through JAX for modeling or writing C++ for an ARM Cortex-M architecture like the Daisy Seed—phase splitting is computationally trivial. Because the audio is just a stream of floating-point values, 180° phase inversion is achieved simply by multiplying the signal vector by \(-1\).

The Payoff: Full-Wave Rectification (The Octave-Up)

Why go through the trouble of phase splitting in a fuzz pedal? To fold the waveform in half. Once you have your two out-of-phase signals \(V_1\) and \(V_2\), you route each of them through a germanium diode (like a 1N34A). Diodes act as one-way valves.

  1. During the positive half of the cycle, \(V_1\) is positive (passes through its diode) and \(V_2\) is negative (blocked by its diode).
  2. During the negative half of the cycle, \(V_1\) swings negative (blocked), but \(V_2\) swings positive (passes through its diode).

When you tie the outputs of both diodes together, you have taken the negative bottom half of the original sine wave and flipped it up to the positive side. The resulting mathematical function is an absolute value:

\[ V_{out}(t) = |V_{in}(t)| = |A \sin(\omega t)| = \frac{2A}{\pi} - \frac{4A}{\pi} \left( \frac{\cos(2\omega t)}{3} + \frac{\cos(4\omega t)}{15} + \frac{\cos(6\omega t)}{35} + \dots \right) \]

Because the wave now peaks twice as often in the same amount of time, the fundamental frequency has literally doubled (\(f_{out} = 2f_{in}\)). Your ear perceives this as an octave jump, and the harsh, abrupt nature of the diode folding creates the aggressive harmonic splatter characteristic of a great analog octave fuzz.