Skip to content

Signals & Wireless (RF)

How AM demodulation gets the voice back

AM demodulation gets the voice back from an amplitude-modulated carrier. The voice is the outline of the carrier's peaks, its envelope, so a receiver can trace it with a diode, a capacitor and a resistor: an envelope detector. A second, series capacitor then strips off the steady level the detector leaves, and what reaches the earpiece swings either side of 0 V, just like the voice did.

Fill and drain

On each positive crest the diode conducts and tops the capacitor up. Between crests it's off, and the resistor drains the capacitor a little. When the voice makes the crests taller the capacitor fills higher; when it makes them shorter, the resistor lets it down. The output is the voice riding on a steady level.

How short is short enough?

R × C must be long against the carrier, so the output holds between crests, and short against the voice, so it can follow it down. For a 50 kHz carrier and a 1 kHz voice, 10 kΩ × 10 nF = 100 µs sits between them. The exact limit comes from how fast the envelope falls, which is fastest halfway down:

R × C ≤ √(1 − m²) ÷ (2π × f × m)

For 50% depth (m = 0.5) at f = 1 kHz, that's about 276 µs: 100 µs follows the voice, and 1 ms can't. Deeper modulation and higher notes need a shorter time constant, so designers check the worst case, the deepest modulation at the highest audio frequency.

Live experiment

Pick the detector's capacitor

A 50 kHz carrier, 50% deep with a 1 kHz voice, through a diode into C1 across 10 kΩ. Pick C1: 1 nF sags between crests, 4.7 or 10 nF follow the voice and hold the crests, and 47 or 100 nF can't fall fast enough and clip the troughs.

Try it yourself: free lesson

“Hear the voice” is lesson 3 of 6 in Airwave Atoll, Signals & Wireless (RF) World 3. Free with an account. Lessons unlock in order, so you’ll do the 16 before it first.

Too long: diagonal clipping

Make R × C as long as the voice's own cycle, 1 ms, and the capacitor can only drain down a slow slope. When the voice falls faster than that, the output is left behind and every trough is cut off by a ramp. The voice comes out distorted, worst on loud, high notes.

Keep the swing, lose the level

The detector's output is the voice sitting on a steady level, about 2.3 V here. A speaker makes sound only from changes, and a steady voltage would just push its cone off centre and waste power. A capacitor in series blocks the steady part and passes the swing. With 10 nF into a 47 kΩ earpiece, the corner is 1 ÷ (2π × 47 kΩ × 10 nF) ≈ 340 Hz, so the 1 kHz voice goes straight through.

Weak signals and the diode drop

A silicon diode needs about 0.6 V before it conducts, so a weak carrier never gets through. Crystal sets used germanium diodes, which conduct from about 0.2 V. Real receivers amplify first and then detect, usually at a fixed intermediate frequency such as 455 kHz, where one tuned detector suits every station.

Free lesson

Now build it yourself

Each lesson comes in three short parts: see the idea, try it on a live board, then build it yourself, with quick checks as you go.

Try it yourself: free lesson

“Hear the voice” is lesson 3 of 6 in Airwave Atoll, Signals & Wireless (RF) World 3. Free with an account. Lessons unlock in order, so you’ll do the 16 before it first.

Or start from the very first lesson, no account needed

Practice questions

Three quick questions from the lessons. Pick an answer to check it.

  1. Question 1 of 3R × C far too long. At the voice's troughs, OUT…
    Show the answer

    Answer: Can't fall fast enough: the troughs are clipped into ramps. The capacitor drains along R × C. If that's slower than the voice falls, OUT is left behind and the troughs are cut off.

    • Why not “Goes negative”? The capacitor only drains toward 0 V through R. Nothing pulls it negative.
    • Why not “Ripples at the carrier”? Ripple at the carrier comes from an R × C that's too short. A long one holds the level very steadily.
    • Why not “Follows perfectly”? It follows the rises, through the diode, but the falls are limited by how fast C can drain through R.
  2. Question 2 of 3R × C far too short. OUT…
    Show the answer

    Answer: Drops between carrier crests: a ragged ripple. The capacitor empties between crests, so OUT saws up and down at the carrier frequency instead of holding the tops.

    • Why not “Clips the troughs”? Clipped troughs come from an R × C that's too long. Too short, the output falls all too easily.
    • Why not “Stays flat”? A short time constant makes the output move more, not less: it drains between every crest.
    • Why not “Doubles”? A detector can't make more than the crests it sees. It only holds them, well or badly.
  3. Question 3 of 3The detector's output is the voice sitting on 2.3 V. A capacitor in series to the earpiece…
    Show the answer

    Answer: Blocks the 2.3 V and passes the voice. A capacitor passes changes and blocks a steady voltage. What reaches the earpiece swings around 0 V.

    • Why not “Adds 2.3 V”? In series it blocks a steady voltage. It removes the 2.3 V rather than adding to it.
    • Why not “Doubles the voice”? A capacitor is passive: it can pass the swing but never make it bigger.
    • Why not “Blocks the voice”? A capacitor passes changes, and the voice is the changing part.
Keep practising: free lesson

“Hear the voice” is lesson 3 of 6 in Airwave Atoll, Signals & Wireless (RF) World 3. Free with an account. Lessons unlock in order, so you’ll do the 16 before it first.

Questions people ask

What is AM demodulation?

Getting the original audio back from an amplitude-modulated carrier. Because the audio is the outline of the carrier's peaks, a diode, a capacitor and a resistor can trace it, and a series capacitor then removes the steady level so a speaker can play it.

What is diagonal clipping?

Distortion from an envelope detector whose time constant is too long. The capacitor can't discharge as fast as the envelope falls, so the output slides down a straight slope instead of following the troughs, cutting them off. Shortening R × C cures it.

Why can't an envelope detector demodulate SSB?

Single-sideband sends no carrier and only one sideband, so the signal's outline no longer traces the audio. The receiver has to supply its own carrier and multiply, a product detector. The same goes for AM that is modulated deeper than 100%.

Every topic