HAND-OUT 4: FREQUENCY TRANSLATION

Suggested Reading: Chapters 3 and 4.6 of Simon Haykin and Michael Moher, Communication
Systems (5th Edition), Wily & Sons Ltd; or Chapter 3.7 of R. E. Ziemer and W. H. Tranter,
Principles of Communications: System Modulation and Noise (5th Edition), John Wiley and Sons,
Inc.

From a Fourier transform viewpoint, the processes in amplitude modulation schemes can be
viewed as those of frequency-translating a certain signal to a desired frequency. This handout
describes some more aspects that are related to the concept of frequency translation.

1 Frequency Division Multiplexing

Multiplexing is generally referred to as a technique for transmitting a multitude of message signals
over the same channel. In frequency division multiplexing (FDM), a frequency band is divided
into a number of frequency subbands, and we assign each message signal to occupy one subband.
Figure 1 shows a generic system diagram for FDM. At the transmitter, each message signal is
modulated to its assigned subband by a suitably chosen carrier frequency. Figure 2 illustrates the
spectrum of an FDM transmitted signal. To avoid interference between subbands, the widths of
the subbands must be chosen such that the modulated signals are non-overlapping in frequency.
At the receiver, the modulated signals of the various message signals must first be separated from
one another. This can easily be done by applying an appropriate bandpass filter for each subband
(at least in principle). Then, a standard demodulator can be applied to recover the message signal
for each subband.
The key concepts we wish to emphasize here are i) frequency translation of the various message
signals to spectrally non-overlapping subbands, and ii) the idea of dividing a channel into inde-
pendent resource blocks in frequency domain. In that regard, the specific modulation scheme used
within each subband is not as crucial, as far as the modulation scheme modulates the message
signal to the assigned subband properly and the bandwidth of the resulting modulated signal is no
greater than the subband bandwidth. In fact, in principle there is flexibility for each subband to
employ a different modulation scheme.
FDM has been used in early long-haul voice telephone systems (obsolete today), where a number
of modulated voice signals are multiplexed and transmitted over a trunk line1 . The principle of
FDM has also been used in 1G mobile systems under the name of frequency division multiple access
(FDMA), where mobile users share the same channel via the same frequency division concept as
discussed above.
1
A specification is as follows. The transmission line has a transmission band of 60 kHz to 108 kHz. SSB is used
to modulate voice signals of bandwidth 4 kHz. A simple calculation reveals that we can multiplex at most 12 voice
signals over the line.

1
 Figure 1: Block diagram of an FDM system.

Figure 2: Spectrum of the FDM signal.

2 Frequency Translation
Consider the following problem: Given a bandpass signal s1 (t) whose center frequency is f1 , trans-
form s1 (t) such that the transformed signal, denoted by s2 (t), has its center frequency at f2 . This
process is called a frequency translation process, and is a generalized form of some modulation and
demodulation processes we encountered previously. For example, in DSB-SC modulation, we have
a message signal s1 (t) = m(t) and we wish to frequency translate s1 (t) to the carrier frequency fc
so that a modulated wave s2 (t) = m(t) cos(2fc t) can be obtained. Also, in DSB-SC demodulation,
we have a modulated wave s1 (t) = m(t) cos(2fc t) and our task is to shift the center frequency
down to 0 Hz such that s2 (t) = m(t).
Frequency translation can be accomplished by the mixer system shown in Figure 3. The idea
is similar to that of the modulation and demodulation processes for DSB-SC. The key issue here

2
 Figure 3: Block diagram of a mixer.

is to choose the frequency of the local oscillator, denoted by f , and design the bandpass filter. To
simplify our study, we will concentrate on the special case of DSB-SC modulated signals, i.e.,

s1 (t) = m(t) cos(2f1 t),

although we should note that the same idea is applicable to general bandpass signals (e.g., QAM).
The product modulator output of the mixer is

s (t) = s1 (t)[2 cos(2f t)]

= 2m(t) cos(2f1 t) cos(2f t)
= m(t) cos(2(f1 + f )t) + m(t) cos(2(f1 f )t). (1)

There are two cases to consider. The first case is when the target center frequency f2 is greater
than the original center frequency f1 , i.e., f2 f1 . In this case, we choose f such that f2 = f1 + f .
The result is simply
f = f2 f1 .
By designing the bandpass filter such that the term m(t) cos(2(f1 + f )t) in (1) is kept and the
term m(t) cos(2(f1 f )t) in (1) is rejected, we obtain the desired result s2 (t) = m(t) cos(2f2 t)
at the bandpass filter output (Note that we have made a subtle assumption, namely, that f W
where W is the message signal bandwidth. Draw the spectrum of s (t) and you will see why).
The second case is when f1 f2 . The idea is the same as above. We choose f such that
f2 = f1 f , or simply
f = f1 f2 .
Then, we design the bandpass filter such that the term m(t) cos(2(f1 + f )t) in (1) is rejected and
the term m(t) cos(2(f1 f )t) in (1) is kept.
Frequency translation is important in enabling any desired frequency up-conversion and down-
conversion of a signal. Particularly, practical modulation and demodulation systems may have more
than one frequency up-conversion and down-conversion stages, owing to practical implementation
constraints. One well-known example is the superheterodyne receiver architecture, to be described
next.

3
3 The Superheterodyne Receiver
The superheterodyne receiver is a special type of receiver that is quite widely used in communi-
cation systems, especially, in broadcast radio receivers. In DSB-SC, for example, coherent demod-
ulation is nothing more than a one-stage frequency translation processbut only in theory. The
superheterodyne receiver, shown in Figure 4, has two stages. In particular, we perform frequency
down-conversion of the received modulated signal to a predetermined immediate frequency (IF), and
then demodulation is carried over the IF band. The reason of doing so has several practical factors
taken into consideration. Simply speaking, a receiver is generally supposed to be tunable with the
carrier frequency fc . It may be difficult to build a sharp bandpass filter for a very high and tunable
center frequency. The superheterodyne receiver overcomes this difficulty by down-converting the
carrier frequency to the IF. Specifically, let fc and fIF denote the carrier frequency and the IF,
respectively. The IF fIF is chosen to be low, and is fixed irrespective of any change of fc . A mixer
with a local oscillator frequency fLO is used to carry out the frequency down-conversion. By the
frequency translation concepts discussed in the last section, the value of fLO should be chosen as

fLO = fc fIF .

Note that fLO varies with the desired fc ; in other words, fLO is tunable.2 Since the IF is low
and always fixed, good amplification and bandpass filtering (in terms of high selectivity) can be
performed over the IF band. Note that the RF section still performs bandpass filtering (in a tunable
and less selective manner), although the idea is to let the IF section do the main task.

Figure 4: Block diagram of a superheterodyne receiver.

The superheterodyne receiver architecture is elegant and practical. However, the two-stage
architecture also results in the so-called image interference effects. This issue may be discussed in
class.

2
As an additional note, there are practical receivers that would use a local oscillator frequency different from the
abovementioned, namely, fLO = fc fIF . For example, receivers in AM broadcast may use fLO = fc + fIF for some
rather practical circuitry reasons. It can be proven that the choice fLO = fc + fIF works at least for AM signals.