Anti-Sidetone Network

Part of Telephony

A circuit that cancels the echo of your own voice in the earpiece while you speak on a telephone.

Why This Matters

When you speak into a telephone without an anti-sidetone circuit, your voice is also heard loudly in your own ear — the effect called sidetone. A small amount of sidetone is actually helpful; it tells you the phone is working and lets you modulate your speaking volume naturally. But too much sidetone is jarring and fatiguing: the speaker unconsciously lowers their voice, the distant party hears them poorly, and the conversation becomes a frustrating negotiation of “can you speak up?”

The anti-sidetone network solves this by creating a path that delivers a phase-inverted replica of your transmitted signal back to your earpiece, canceling out most of the sidetone. This is one of the first examples of deliberate signal cancellation in electrical engineering — a principle that later extended to noise-canceling headphones, echo cancellation in digital telephony, and active noise reduction in aircraft.

For anyone building a working telephone system from scratch, understanding the anti-sidetone bridge is essential for creating instruments that are comfortable and practical to use. A telephone without it will work, but people will avoid using it.

The Sidetone Problem

In the simplest telephone circuit, the microphone, earphone, and line are all connected in series or parallel. When the speaker talks, the microphone generates current that flows through the circuit — including directly through the earphone before it ever reaches the distant end. The speaker hears their own voice amplified, sometimes louder than the incoming voice from the other party.

The physical reason this matters: humans naturally adjust speaking volume based on what they hear in their ears. When sidetone is too loud, the brain interprets the loud feedback as “I’m speaking too loudly” and the speaker reduces their voice. This feedback loop drives the conversation toward whispered inaudibility. The distant party asks for repetition, the near party gets frustrated and confused, and fatigue accumulates over a long call.

There is a secondary effect: loud sidetone masks the incoming voice signal during speaking, making it harder to hear the other party when they try to interject or respond.

Bridge Circuit Principles

The anti-sidetone network uses a Wheatstone bridge topology. The bridge has four arms, with the microphone in one arm, a balancing impedance (the “artificial line” or balance network) in the opposite arm, and the earphone connected between the midpoints of the two bridge halves.

When the bridge is balanced, equal voltages appear at both midpoints relative to the circuit ground — so no voltage difference exists across the earphone. Any signal that flows through both arms equally (like sidetone from your microphone) cancels at the earphone terminals. But the incoming signal from the distant party arrives differently: it enters at the line terminals and creates a voltage difference across the earphone, which the ear hears normally.

The balance is never perfect in practice because the artificial line network must approximate the impedance of an actual telephone line, which varies with line length, temperature, and frequency. In older telephone instruments, the balance network was a fixed inductor and capacitor combination tuned for average line conditions. In more sophisticated designs, adjustable networks allowed field technicians to tune the balance for each installation.

Constructing the Network

A practical anti-sidetone network for a local telephone system requires an induction coil (also called a hybrid transformer in modern terminology), the balance network, and correct wiring.

The induction coil is a transformer with a center-tapped winding on one side. The microphone connects to one half of this winding, the line connects to the other half, and the center tap feeds the earphone through a capacitor. The balance network connects across the line terminals in parallel with the line itself. The component values determine how well the bridge cancels sidetone at different audio frequencies.

For a simple workshop telephone operating over short indoor lines (under 1 km), a fixed balance network of a 600-ohm resistor in series with a 2 µF capacitor typically provides adequate sidetone reduction across the voice frequency range (300-3,400 Hz). For longer lines or carbon microphones with different impedance characteristics, you may need to adjust these values empirically.

To test your network, speak normally into the microphone and assess whether your voice in the earpiece is comfortable — noticeably present but not overwhelming. If sidetone is still too loud, increase the capacitor value slightly. If you cannot hear yourself at all, check that the induction coil phasing is correct; reversed winding polarity eliminates sidetone entirely but creates other problems with received signal levels.

Adjusting for Line Conditions

The ideal balance point shifts depending on line impedance. A short line (100 meters, low resistance) has different impedance than a long line (10 km, high resistance and significant capacitance). A network optimized for one condition may perform poorly on the other.

Field adjustment requires an audio signal source and a way to measure the signal appearing at the earphone terminals. Inject a known audio tone into the microphone circuit and adjust the balance network until the tone in the earphone is minimized. This null-balance procedure achieves the best cancellation possible for that specific line.

In practice, most telephone instruments accept a compromise balance that works reasonably well for typical line lengths. The engineering goal is not perfect cancellation (which would require perfect knowledge of line impedance) but adequate cancellation that makes the phone comfortable to use across a range of realistic conditions.

Modern digital telephone systems handle sidetone entirely in software, inserting a precise negative replica of the transmitted signal into the received audio stream. This digital cancellation is essentially perfect but requires signal processing hardware. For analog systems built from components, the passive bridge network is the correct and elegant solution.