Frequency Modulation

Switch on an FM radio, pick up a walkie-talkie, or use a wireless microphone, and you are using FM — frequency modulation. The carrier frequency swings up when the audio is positive, swings down when it is negative, and returns to center during silence. FM's defining characteristic — information carried in the carrier's frequency variations rather than its amplitude — gives it properties that make it ideal for high-quality broadcasting and for clear local voice communication: excellent noise immunity, the capture effect, and the ability to use efficient non-linear amplifiers.

How FM Encodes Information

In frequency modulation, the instantaneous frequency of the carrier is made to deviate from its center frequency in proportion to the instantaneous amplitude (voltage) of the modulating audio signal. The amplitude of the carrier remains constant throughout — it is only the frequency that varies. This is the opposite of AM, where the frequency stays constant and the amplitude varies.

The mathematical description of an FM signal is:

To make this concrete: imagine a transmitter set to 146.520 MHz. Without modulation, it transmits exactly 146.520 MHz continuously. When you speak into the microphone and generate a 1,000 Hz audio tone at 2 V peak, the carrier frequency swings from 146.520 MHz upward to 146.5225 MHz (a deviation of +2.5 kHz) and then downward to 146.5175 MHz (a deviation of −2.5 kHz), oscillating between these extremes 1,000 times per second. The center frequency always returns to 146.520 MHz between deviations.

Notice what determines the frequency deviation: it is the amplitude of the audio signal. A louder audio tone causes a larger frequency swing. A quieter tone causes a smaller swing. The rate at which the carrier swings back and forth — the speed of the modulation — is determined by the audio frequency. A 1,000 Hz tone causes the carrier to sweep back and forth 1,000 times per second. A 2,000 Hz tone sweeps the carrier back and forth 2,000 times per second.

FM modulation: the audio amplitude (top) controls the instantaneous carrier frequency (middle). The resulting FM waveform (bottom) shows compressed cycles when audio is positive (carrier above center frequency) and expanded cycles when audio is negative (carrier below center frequency). The carrier amplitude remains constant throughout.

View Larger

Frequency Deviation

Frequency deviation (symbol Δf, spoken "delta f") is the peak shift of the carrier frequency away from its center frequency, caused by the maximum amplitude of the modulating signal. Deviation is always specified as a peak value (plus or minus), not a peak-to-peak value.

The deviation is set by the transmitter design and the audio levels. Different FM services use different standard deviation values:

• Narrow-band FM (NBFM) — two-way and business radio, walkie-talkies: maximum deviation ±2.5 kHz or ±5 kHz depending on channel spacing (±2.5 kHz for 12.5 kHz channels, ±5 kHz for 25 kHz channels), audio bandwidth about 300–3,000 Hz
• Wireless microphones and in-ear monitors: typically ±40–75 kHz deviation for full-bandwidth audio
• Commercial narrowband FM (NBFM): ±2.5 kHz or ±5 kHz depending on service and country
• Wideband FM (WBFM) — broadcast: Maximum deviation ±75 kHz (USA), 75 kHz time constant pre-emphasis

Over-deviation — modulating the transmitter more than the specified maximum deviation — causes the FM signal to exceed its channel allocation. On a 12.5 kHz-spaced radio system, an over-deviated signal spreads into adjacent channels and interferes with other users. The transmitter's deviation is measured by service technicians using a deviation meter or a communications test set, or by looking at the signal on a spectrum analyzer. Broadcast transmitters use audio processors with precise limiters so that deviation never exceeds ±75 kHz.

The FM Modulation Index (Deviation Ratio)

The FM modulation index β (also called the deviation ratio when Δf is the maximum rated deviation and f_m is the maximum rated audio frequency) is the ratio of the peak frequency deviation to the modulating audio frequency:

β is not a fixed property of the FM system — it changes depending on both the audio frequency and the audio amplitude (which sets the deviation). For a transmitter with maximum deviation of ±5 kHz:

• At a 1,000 Hz audio tone at maximum amplitude: β = 5,000 / 1,000 = 5
• At a 3,000 Hz audio tone at maximum amplitude: β = 5,000 / 3,000 = 1.67
• At a 1,000 Hz audio tone at half amplitude (2.5 kHz deviation): β = 2,500 / 1,000 = 2.5

The significance of β becomes clear when you examine how FM bandwidth depends on it. Small β (called narrowband FM, or NBFM) produces a signal that is not very different from AM in bandwidth. Large β (called wideband FM, or WBFM) produces a much wider signal but with greatly improved noise performance — a fundamental tradeoff in FM system design called the FM improvement threshold or FM quieting.

Calculator: FM Deviation Ratio

FM Bandwidth and Carson's Rule

Calculating the exact bandwidth of an FM signal requires Bessel functions (discussed briefly in the next section), which give the precise power levels of each sideband pair. For practical engineering purposes, however, an approximation rule developed by John R. Carson in 1922 gives a result accurate enough for most calculations:

Carson's rule captures both contributions to FM bandwidth: the frequency deviation (how far the carrier swings) and the audio frequency (how fast it swings). If either increases, the bandwidth increases. A signal with very high deviation but very low audio frequency is still wide (swept far, though slowly). A signal with very high audio frequency but very little deviation is also wide (swept fast, though not far). The bandwidth is dominated by whichever term is larger.

Note that Carson's rule gives the bandwidth containing approximately 98% of the signal's total power. The actual spectrum of an FM signal extends to infinity in theory (the Bessel function series never completely terminates), but the outer sidebands become negligible for practical purposes. For receiver filter design and channel planning, the 98% bandwidth from Carson's rule is the standard specification.

Calculator: FM Bandwidth (Carson's Rule)

FM spectrum comparison using Carson's rule. Left: NBFM two-way radio with ±5 kHz deviation and 3 kHz audio gives 16 kHz total bandwidth. Right: WBFM broadcast with ±75 kHz deviation and 15 kHz audio gives 180 kHz total bandwidth. The multiple spectral lines are Bessel function sideband pairs; Carson's rule gives the bandwidth envelope containing ~98% of total signal power.

View Larger

NBFM vs WBFM

FM systems are broadly categorized as narrowband (NBFM) or wideband (WBFM) based on their modulation index and resulting bandwidth:

The trade-off between NBFM and WBFM illustrates a fundamental principle of FM: wider deviation provides better noise performance (the FM improvement factor), but at the cost of more bandwidth. FM broadcast stations accept 200 kHz of channel spacing (and about 180 kHz of signal bandwidth) because the resulting audio quality and noise performance are excellent for music. Two-way radios accept the narrower audio quality of NBFM because the goal is intelligible speech and fitting as many channels as possible into a limited band.

The Bessel Functions: FM Sidebands in Detail

Unlike AM, which produces exactly three spectral components (carrier, USB, LSB) for a single audio tone, FM produces theoretically infinite sideband pairs for any non-zero modulation index. The amplitude of each sideband pair is determined by Bessel functions of the first kind, denoted J_n(β), where n is the sideband order and β is the modulation index.

The FM sideband structure consists of:

• A component at the carrier frequency f_c with amplitude J₀(β)
• First-order sidebands at f_c ± f_m with amplitude J₁(β)
• Second-order sidebands at f_c ± 2f_m with amplitude J₂(β)
• Third-order sidebands at f_c ± 3f_m with amplitude J₃(β)
• And so on to infinity...

Several important things to note about the Bessel function behavior:

The carrier power changes with modulation index. Unlike AM, where the carrier power is always P_c regardless of modulation depth, the FM carrier component's amplitude is J₀(β), which actually goes to zero at certain values of β (approximately β = 2.40, 5.52, 8.65...). At these special values of β, all of the signal's power is in the sidebands — the carrier disappears. This is useful in laboratory tests and in certain communications systems, but is worth knowing because it means FM's spectral structure is fundamentally different from AM's.

Total power is constant. Unlike AM, where the total transmitted power increases with modulation depth, FM modulation does not change the total transmitted power. The carrier amplitude is constant. What FM does is redistribute the constant total power among the carrier and an increasing number of sidebands as β increases. This is why FM transmitters can use more efficient (Class C or Class D) output amplifiers — the output amplitude never changes, so there is no need for the linear amplifiers required by AM or SSB.

For practical purposes you do not need to memorize the Bessel function values. Carson's rule gives a good approximation to the signal bandwidth, and the key understanding is simply that FM produces multiple sideband pairs, unlike AM's single pair. The Bessel functions do explain one useful trick: at a modulation index of 2.405 the carrier component disappears completely, which lets technicians set deviation precisely by watching the carrier vanish on a spectrum analyzer.

The Capture Effect

One of FM's most distinctive and useful properties is the capture effect (also called the FM capture effect): when two FM signals on the same frequency arrive at a receiver simultaneously, the receiver tends to lock onto only the stronger signal and almost completely ignore the weaker one. This is fundamentally different from AM behavior, where two AM signals add linearly at the receiver and both are heard simultaneously (with a mixture of the two audio contents).

The capture effect arises from the way FM limiters work. The FM receiver contains a limiter stage that strips away all amplitude variations from the received signal, passing only the frequency variations to the discriminator. When two signals of different amplitudes are present, they combine to produce a complex signal that appears to the limiter approximately as the stronger signal with some frequency variations imposed by the weaker signal. The limiter, by stripping amplitude variations, also strips most of the weaker signal's contribution — effectively "capturing" the receiver for the stronger station.

The capture effect is approximately 6 dB: if one FM signal is 6 dB stronger than a co-channel interferer, the receiver will capture the stronger signal and provide a nearly interference-free output. Below 6 dB difference, capture is not complete and both signals degrade. Above 10–12 dB difference, capture is essentially perfect and the weaker signal is virtually inaudible.

The FM capture effect: when two signals share a frequency, FM receivers lock onto the stronger one. With approximately 6 dB difference, capture is essentially complete — the weaker signal is almost inaudible. Within 6 dB of equal strength, both signals degrade each other. This behavior differs fundamentally from AM, where both signals are always heard simultaneously.

View Larger

In practice, the capture effect is both beneficial and occasionally problematic:

• Beneficial: If two transmitters are on the same frequency — two walkie-talkie users talking at once, or a distant FM station on the same channel as a local one — the receiver captures the stronger signal. Listeners hear clean audio from one station rather than a garbled mix of two.
• Problematic: A stronger unwanted signal captures the receiver completely, so a weaker wanted signal can disappear without warning. This is why aviation radio still uses AM, where two simultaneous transmissions are both heard.
• Critical for safety: This is why aviation VHF uses AM rather than FM — in an emergency, multiple aircraft and ground stations may transmit simultaneously on the same frequency with safety-critical information. With FM, only the strongest would be heard. With AM, all transmissions add at the receiver and all are audible.

Pre-Emphasis and De-Emphasis

FM has a characteristic noise spectrum that creates a practical problem: the noise at the FM discriminator output increases linearly with audio frequency. High audio frequencies suffer more from FM noise than low audio frequencies. This is an inherent property of the FM discriminator — the noise floor after demodulation rises at 6 dB per octave as you go from low to high audio frequencies.

The standard engineering solution is pre-emphasis at the transmitter and de-emphasis at the receiver:

Pre-emphasis: Before the audio reaches the FM modulator, it passes through a high-pass filter (technically an RC network that boosts high frequencies). This intentionally over-emphasizes (amplifies) the high audio frequencies before transmission, boosting them relative to the low frequencies. The standard pre-emphasis time constant in the United States is 75 microseconds (abbreviated 75 μs), which is defined as the RC time constant of the pre-emphasis network. (Europe and most of the rest of the world uses 50 μs for mono broadcast, but US broadcast uses 75 μs.) The pre-emphasis begins rising at 1/(2π × 75×10⁻⁶) ≈ 2,120 Hz.

De-emphasis: At the receiver, after FM demodulation, the audio passes through a complementary low-pass filter with the same 75 μs time constant, which rolls off the high frequencies back to their natural levels. Because the FM noise spectrum also rises at high frequencies, and the de-emphasis filter rolls off at exactly the same rate, the de-emphasis filter cuts the noise as well as cutting the boosted high frequencies — resulting in a flat audio response with significantly reduced high-frequency noise compared to a system without pre-emphasis/de-emphasis.

Pre-emphasis (75 μs) boosts high audio frequencies at the transmitter by up to 17 dB at 15 kHz. De-emphasis at the receiver applies the complementary roll-off, restoring a flat audio response while also reducing high-frequency FM noise. The combined effect of pre-emphasis + de-emphasis provides a flat audio response from 20 Hz to 20 kHz.

View Larger

Pre-emphasis time constants are standardized: 75 μs in the Americas and South Korea, and 50 μs in Europe and most other regions, for FM broadcasting. Two-way radios use a similar 6 dB/octave pre-emphasis over the speech band. If the transmitter and receiver use different time constants, the system still works but the audio sounds slightly dull or slightly bright. FM receiver chips apply the correct de-emphasis automatically, and it is rarely something the user needs to adjust.

FM Demodulation

FM demodulation — recovering the audio from the frequency-modulated signal — requires a circuit that converts frequency variations into voltage variations. Several circuit types accomplish this:

The FM Limiter

Before the detector, an FM receiver passes the signal through one or more limiter stages. A limiter is a high-gain amplifier that clips its output at a fixed amplitude regardless of the input amplitude — it acts like a hard clipper. Because FM encodes information only in frequency (not amplitude), clipping the amplitude does no harm to the information content but dramatically reduces amplitude noise, especially impulse noise from lightning and electrical equipment. After limiting, the signal has constant amplitude with only frequency variations — the pure FM signal the detector was designed for.

The Slope Detector

The simplest FM detector tunes an LC circuit slightly off the FM center frequency, on the slope of the resonance curve. As the FM signal's frequency varies above and below center, the signal moves up and down the slope of the resonance, converting frequency variations into amplitude variations. An envelope detector then recovers the audio from the resulting AM-like signal. The slope detector has poor linearity and low efficiency, but illustrates the basic principle of FM-to-AM conversion.

The Foster-Seeley Discriminator

The Foster-Seeley discriminator uses a special transformer arrangement to produce two voltages whose difference is proportional to the frequency offset from the center frequency. At center frequency, the two voltages are equal and their difference is zero (no audio output). Above center, one voltage rises and the other falls; below center, the reverse occurs. The differential output, after rectification and filtering, gives the recovered audio. The Foster-Seeley discriminator requires a limiter stage before it.

The Ratio Detector

The ratio detector is similar to the Foster-Seeley but includes a large storage capacitor that suppresses amplitude variations — it provides some of its own AM rejection without requiring a separate limiter stage. Ratio detectors were common in high-quality FM broadcast receivers from the 1950s through the 1980s because they combined amplitude rejection and FM detection in one stage. They have slightly higher harmonic distortion than the Foster-Seeley discriminator.

The Phase-Locked Loop (PLL) FM Detector

Modern FM receivers, including the single-chip FM radios in phones and the receivers in two-way radios, use either a phase-locked loop or a digital demodulator after an analog-to-digital converter. In a PLL detector, the VCO tracks the incoming FM signal's frequency. As the FM signal's frequency swings above and below center, the error voltage that drives the VCO to track it is proportional to the frequency deviation — this error voltage is the demodulated audio. PLL detectors offer excellent linearity, low distortion and wide bandwidth, and need no tuned coils.

Where FM Is Used

FM is the standard for high-quality analog broadcasting and for local voice communication:

• FM broadcasting (88–108 MHz): wide-band FM with ±75 kHz deviation, stereo carried on a 38 kHz subcarrier with a 19 kHz pilot tone, and RDS station-name data on a 57 kHz subcarrier.
• Two-way radio: walkie-talkies, business, public-safety and marine VHF radios use narrow-band FM. Many systems switch between talk groups using sub-audible tones (CTCSS, 67–254 Hz) or digital codes (DCS) added to the audio, so that a receiver only opens its squelch for the right users.
• Wireless microphones and in-ear monitors: wide-deviation FM with companding gives near-broadcast audio quality.
• Analog TV sound, baby monitors, cordless phones and FM transmitters for car stereos all used or use FM.
• Data: frequency-shift keying (FSK) and Gaussian FSK (GFSK) — the digital form of FM — carry Bluetooth, many 433/868/915 MHz sensor links and pagers, as described in the Digital Modulation Overview lesson.