Communication Systems/Frequency Modulation
Contents
Frequency Modulation[edit]
If we make the frequency of our carrier wave a function of time, we can get a generalized function that looks like this:
We still have a carrier wave, but now we have the value ks(t) that we add to that carrier wave, to send our data.
As an important result, ks(t) must be less than the carrier frequency always, to avoid ambiguity and distortion.
Deriving the FM Equation[edit]
Recall that a general sinusoid is of the form:
Frequency modulation involves deviating a carrier frequency by some amount. If a sine wave is used to deviate the carrier, the expression for the frequency at any instant would be:

 where:



 instantaneous frequency
 carrier frequency
 carrier deviation
 modulation frequency


This expression describes a signal varying sinusoidally about some average frequency. However, we cannot simply substitute this expression into the general equation for a sinusoid to get the FM equation. This is because the sine operator acts on angles, not frequency. Therefore, we must define the instantaneous frequency in terms of angles.
It should be noted that the modulation signal amplitude governs the amount of carrier deviation while the modulation signal frequency governs the rate of carrier deviation.
The term is an angular velocity (radians per second) and is related to frequency and angle by the following relationship:
 To find the angle, we must integrate with respect to time:
 We can now find the instantaneous angle associated with the instantaneous frequency:
 This angle can now be substituted into the general carrier signal to define FM:
 The FM modulation index is defined as the ratio of the carrier deviation to modulation frequency:
 Consequently, the FM equation is often written as:
Bessel's Functions[edit]
This is a very complex expression and it is not readily apparent what the sidebands of this signal are like. The solution to this problem requires a knowledge of Bessel's functions of the first kind and order p. In open form, it resembles:

 where:



 Magnitude of the frequency component
 Side frequency number (not to be confused with sidebands)
 Modulation index


 As a point of interest, Bessel's functions are a solution to the following equation:
Bessel's functions occur in the theory of cylindrical and spherical waves, much like sine waves occur in the theory of plane waves.
It turns out that FM generates an infinite number of side frequencies (in both the upper and lower sidebands). Each side frequency is an integer multiple of the modulation signal frequency. The amplitude of higher order side frequencies decreases rapidly and can generally be ignored.
The amplitude of the carrier signal is also a function of the modulation index and under some conditions, its amplitude can actually go to zero. This does not mean that the signal disappears, but rather that all of the broadcast energy is redistributed to the side frequencies.
A plot of the carrier and first five side frequency amplitudes as a function of modulation index resembles:
The Bessel coefficients have several interesting properties including:
One very useful interpretation of this is: represents the voltage amplitude of the carrier, represents the amplitude of the 1st side frequency, the 2nd side frequency etc. Note that the sum of the squares (power) remains constant.
FM Bandwidth[edit]
FM generates upper and lower sidebands, each of which contain an infinite number of side frequencies. However, the FM bandwidth is not infinite because the amplitude of the higher order side frequencies decreases rapidly. Carson's Rule is often used to calculate the bandwidth, since it contains more than 90% of the FM signal.

 Carson's Rule
In commercial broadcast applications, for a purely monaural station, the maximum modulation index () = 75/15 = 5, coming from: the maximum carrier deviation () = 75 kHz, and maximum modulation frequency () = 15 kHz. The total broadcast spectrum according to Carson's rule is 180 kHz, but an additional 20 kHz guard band is used to separate adjacent radio stations. Therefore, each FM radio station is allocated 200 kHz.
For stereo stations, the maximum modulation index is significantly reduced because the information needed to separate the channels has to be transmitted along with the mono audio signal. This increases the required bandwidth to 53 kHz. Hence, the max. modulation index is = 75/53 = 1.41509434... Radio Data System (RDS) information, increase this further to ~60 kHz, reducing the max. modulation index to 75/60 = 1.25.
How FM Stereo works[edit]
The mono signal is M = L + R, with the stereo difference being S = L  R. Adding both simultaneous equations together gives M + S = 2L + (RR), recovering the left channel, while subtracting them recovers the right channel. This is transmitted as a double sideband suppressed carrier (DSBSC), which is essentially just an AM "station" going along, but without it's carrier being sent when there is nothing being transmitted on it. ("Stations" sent along with the main program (usually in ultrasonic frequencies) are known as subcarriers.) A stereo "pilot" tone is used to let the receiver know that there is a stereo signal being received, and also to allow the suppressed carrier to be regenerated (by doubling the pilot tone's frequency) so the stereo difference signal can be demodulated just like a normal AM station and the resulting signals used to separate the audio into two channels.
RDS information is yet another "AM station" being sent along with the main program, but at 3x the pilot frequency (19 kHz x 3 = 57 kHz). It's contents are not audio, but analogue values meant to represent a digital signal which carries the station name and many other info like it's alternate frequencies, time of the day, program info, etc.
Noise[edit]
In AM systems, noise easily distorts the transmitted signal however, in FM systems any added noise must create a frequency deviation in order to be perceptible.
The maximum frequency deviation due to random noise occurs when the noise is at right angles to the resultant signal. In the worst case the signal frequency has been deviated by:
This shows that the deviation due to noise increases as the modulation frequency increases. Since noise power is the square of the noise voltage, the signal to noise ratio can significantly degrade.
To prevent this, the amplitude of the modulation signal is increased to keep the S/N ratio constant over the entire broadcast band. This is called preemphasis.
Pre & Deemphasis[edit]
Increasing the amplitude of high frequency baseband signals in the FM modulator (transmitter) must be compensated for in the FM demodulator (receiver) otherwise the signal would sound quite tinny (too much treble).
The standard curves resemble:
In commercial FM broadcast, the emphasis circuits consist of a simple RC network with a time constant of 75 Sec and a corner frequency of 2125 Hz.
The magnitude of the preemphasis response is defined by:
FM Transmission Power[edit]
Since the value of the amplitude of the sine wave in FM does not change, the transmitted power is a constant. As a general rule, for a sinusoid with a constant amplitude, the transmitted power can be found as follows:
Where A is the amplitude of the sine wave, and R_{L} is the resistance of the load. In a normalized system, we set R_{L} to 1.
The Bessel coefficients can be used to determine the power in the carrier and any side frequency:



 is the power in the unmodulated carrier.
 is the total power and is by definition equal to the unmodulated carrier power.

As the modulation index varies, the individual Bessel coefficients change and power is redistributed from the carrier to the side frequencies.
FM Transmitters[edit]
FM Transmitters can be easily implemented using a VCO (see why we discussed Voltage Controlled Oscillators, in the first section?), because a VCO converts an input voltage (our input signal) to a frequency (our modulated output).
Signal >VCO> FM Signal
FM Receivers[edit]
Any angle modulation receiver needs to have several components:
 A limiter, to remove abnormal amplitude values
 bandpass filter, to separate the outofband noise.
 A Discriminator, to change a frequency back to a voltage
 A lowpass filter, to remove noise added by the discriminator.
A discriminator is essentially a differentiator in line with an envelope detector:
FM >Differentiator>Envelope Filter> Signal
Also, you can add in a blocking capacitor to remove any DC components of the signal, if needed.