Frequency modulation (FM) is a form of radio frequency modulation in which the message varies the instantaneous frequency of an RF carrier while its amplitude stays constant. The peak frequency change is the deviation, and its ratio to the message frequency sets the modulation index, the sidebands and the bandwidth. FM trades extra bandwidth for better output SNR, and it is used in broadcasting, two-way radio and FMCW radar.
This article explains how frequency modulation works, with numbers for broadcast and narrowband FM. It is a companion to amplitude modulation, phase modulation and the broader survey of continuous wave modulation.
What Is Radio Frequency Modulation?
Radio frequency modulation means imposing a message on an RF carrier. The purpose of RF modulation is practical: antennas need to be a sizeable fraction of a wavelength, and audio frequencies would need antennas many kilometres long; moving the signal to its own carrier frequency also lets many users share the spectrum. A carrier A cos(2πfct + φ) has three things to vary. AM uses A, PM uses φ, and FM varies the rate of change of φ.
How Frequency Modulation Works
The instantaneous frequency of an FM signal follows the message m(t):
f(t) = fc + kf m(t)
Since phase is the integral of frequency, the signal is:
s(t) = Ac cos(2πfct + 2πkf ∫m(τ)dτ)
The amplitude Ac never changes, so a saturated class C amplifier can carry the signal, and any amplitude noise can be clipped away in the receiver. Figure 1 shows how frequency follows the message.

FM and PM are closely related: FM of a message equals PM of that message’s integral. A discussion of the phase side is in the phase modulation article.
Frequency Deviation and Modulation Index
Two numbers describe an FM signal. Peak frequency deviation Δf is the largest shift from the carrier, set by the message amplitude. For a tone at fm, the modulation index is:
β = Δf / fm
Note that a louder tone raises Δf, while a higher-pitched tone at the same loudness lowers β. β is also the peak phase swing in radians.
FM Spectrum and Bessel Sidebands
A tone-modulated FM carrier expands into a carrier and an infinite set of sideband pairs at multiples of fm, with amplitudes set by Bessel functions:
s(t) = Ac Σ Jn(β) cos(2π(fc + n fm)t)
Total power is constant whatever the modulation, because the amplitude is constant. Modulation only moves power from the carrier into the sidebands. For β = 5, the first coefficients are:
| Order n | Jn(5) | Level (dB) | Cumulative power |
|---|---|---|---|
| 0 (carrier) | −0.178 | −15.0 | 3.2% |
| 1 | −0.328 | −9.7 | 24.6% |
| 2 | 0.047 | −26.6 | 25.0% |
| 3 | 0.365 | −8.8 | 51.7% |
| 4 | 0.391 | −8.2 | 82.3% |
| 5 | 0.261 | −11.7 | 95.9% |
| 6 | 0.131 | −17.7 | 99.4% |
| 7 | 0.053 | −25.5 | 99.9% |
| 8 | 0.018 | −34.7 | 100.0% |

FM Bandwidth and Carson’s Rule
Carson’s rule estimates the occupied bandwidth:
BT ≈ 2(Δf + fm) = 2fm(β + 1)
| System | Δf | Highest fm | β | Carson bandwidth |
|---|---|---|---|---|
| Broadcast FM (typical) | 75 kHz | 15 kHz | 5 | 180 kHz |
| Two-way radio, 25 kHz channel (typical) | 5 kHz | 3 kHz | 1.67 | 16 kHz |
| Two-way radio, 12.5 kHz channel (typical) | 2.5 kHz | 3 kHz | 0.83 | 11 kHz |
Types of frequency modulation are usually split on this basis. Narrowband FM keeps β small, with a bandwidth not far above AM, and is used for voice. Wideband FM has a large β and gains SNR. Digital FM is frequency shift keying, and a linear frequency sweep is the chirp used in FMCW radar.
Worked Example: Broadcast FM
Take Δf = 75 kHz and fm = 15 kHz.
- Modulation index: β = 75/15 = 5.
- Carson bandwidth: 2(75 + 15) = 180 kHz, which sits inside the usual 200 kHz channel spacing.
- Power within the band: ±90 kHz covers six sideband pairs, and the table gives 99.4%.
- SNR gain over baseband: 3β²/2 = 37.5, or 15.7 dB.
- Output SNR against CNR in 180 kHz: 3β²(β + 1) = 450, or 26.5 dB above CNR.
A CNR of 20 dB should therefore give 46.5 dB. The simulated discriminator gave 46.8 dB, which agrees within the statistical spread of the run. For narrowband FM at β = 1.67, the SNR gain over baseband, 3β²/2, is only 6.2 dB. SNR gain grows with the square of the bandwidth spent, and that is the core exchange of FM.
FM Demodulation and the Threshold Effect
Common FM demodulators are:
- Limiter-discriminator. A limiter removes amplitude variation, then a slope or Foster-Seeley style circuit converts frequency to voltage.
- Quadrature detector. A phase-shifted copy of the signal is multiplied with the original, giving an output proportional to frequency.
- Phase-locked loop. The loop’s control voltage follows the input frequency, and some designs lower the threshold.
- Digital. The phase difference between consecutive I/Q samples gives frequency directly, as in many software receivers.
FM’s gain holds only above a threshold. Near that CNR, noise occasionally spins the phasor around the origin, which produces audible clicks, and output SNR falls fast.

Here the knee sits near 9 dB, in line with the usual rule of thumb of about 10 dB. Below it, 6 dB of CNR gives 19.5 dB against a linear prediction of 32.5 dB. A link budget for FM should keep CNR comfortably above this point. See noise figure and the RF link budget for how to find CNR.
Pre-Emphasis and De-Emphasis
FM demodulator output noise rises with frequency, so the top of the audio band is the noisiest. Broadcast transmitters boost treble with a pre-emphasis network, and receivers cut it again. The time constant is typically 75 µs in the Americas and 50 µs in Europe, with corners at 2122 Hz and 3183 Hz. A 75 µs network applies 17.1 dB of boost at 15 kHz, which restores SNR at high audio frequencies.
Capture Effect and FM Advantages
When two FM signals share a channel, the discriminator locks onto the stronger one. A difference of a few dB is typically enough to suppress the weaker signal, which is the capture effect. AM has nothing similar, and interference simply adds to the audio.
| Property | AM | FM |
|---|---|---|
| Envelope | Follows message | Constant |
| Transmitter amplifier | Linear for low-level AM | Saturated allowed |
| Bandwidth for 15 kHz audio | 30 kHz | 180 kHz at β = 5 |
| SNR above threshold | No gain | 15.7 dB gain at β = 5 |
| Co-channel interference | Audible | Capture effect |
Applications of Frequency Modulation
- Broadcast radio in the VHF band, with stereo added by a 19 kHz pilot and a difference channel around 38 kHz.
- Land mobile, marine and amateur two-way radio, using narrowband FM.
- FMCW radar, where a chirped carrier gives range from the beat frequency.
- Telemetry, where a constant envelope allows efficient amplifiers.
- Digital links that use FSK, GFSK and MSK.
Frequently asked questions
What is frequency modulation in simple terms?
FM changes how fast the carrier oscillates in step with the message, leaving its amplitude alone. A louder signal gives a larger frequency swing, and a higher pitch gives a faster swing.
What is an example of FM modulation?
Broadcast FM radio is the standard example: a VHF carrier with up to ±75 kHz deviation. A 15 kHz tone at full deviation has β = 5 and fills about 180 kHz.
What are the advantages of frequency modulation?
FM gives better output SNR than AM above threshold, tolerates amplitude noise, allows efficient amplifiers and rejects weaker co-channel signals. Its costs are wider bandwidth and the threshold effect.
What are the types of frequency modulation?
Narrowband FM keeps β small and is used for voice. Wideband FM uses a larger β for high SNR. FSK, GFSK and FMCW chirps are digital or radar forms of the same principle.
Related topics
See amplitude modulation and demodulation for the AM receiver side, frequency shift keying for the digital form, and also amplitude shift keying, pulse amplitude modulation and continuous wave modulation. The RF fundamentals hub collects related material.

