\ PHASE AND FREQUENCY MO DULATION sai ieciaaa INTRODUCTION In the previous chapter, we invest chap ‘modulation and frequen it can provide better discrimination against noise and j lation. As will be shown in Chapter 6, however, achieved at the expense of increased transmission eband signal. In this method o izated the effect of slowly varying the amplitude of a sinu- Soidal carrier wave in accordance with the baseband (information-carrying) signal. In this Wwe study a second family of continuous-wave fhumely, angle modulation, in which the angle of the carrier wave is vatied according to modulation, the amplitude of the carrier wave is fed constant. There are two common forms of angle modulation, namely, phase (CW) modulation systems, 42 BASIC DEFINITIONS Let 0(1) denote the angle of function of the information-bearin angle-modulated wave as Where A. is the carrier amplitude by 2m radians, It 4 modulated sinusoidal carrier al ne signal or message sign s(t) = A, cos{@,(¢)} A complete oscillation occurs ‘ses monotonically with time, the averse time f; itis assumed to be a ‘a. We express the resulting (4.1) whenever 0(¢) changes © frequency in Hertz1 interval from #10 ¢ Aris given by eae 2 BASIC OEFINTION aie fiy(t) = OL A) ~ (0) 42) ‘ay thus define the instantaneous frequency of the angle-modulated signal s(0) as AD = im far(t) soni [pe + Ar) -6 an 2nAt (43) _ 1 do) Qn dt svhere, in the last line, we have invoked the definition for the derivative of the angle with cL to time Thus, according to Eq. (4.1), we may interpret the angle-modulated signal s(f) as @ 2 phasor of length A, and angle 6(1). The angular velocity of such a phasor is aby ured in radians per second in accordance with Eq. (4.3). In the simple case ‘of an unmodulated carrier, the angle 0,(1) is (0) = 2a fa be. and the corresponding phasor rotates with a constant angu Yelocty equal to 2rf. The constant ¢ is the value of O(¢) at r= 0, ath ‘ ‘There are an infinite number of : manner with the message monly used methods, phase 1. Phase modulation (PM) is angle 0,(1) is varied time and multiplying the result by 2x, we get 6,(t) = 2a fet + Inky S acERTIES OF ANGLE-MODULATED WAVES uxtulated Waves are charact ed waves in a family of their ow id waves, as illustrated by the w. lulation. Figures 4.ta and 4.15 ively. Figures 4.le, 4.1d, 4 phase-modulated (PM), ani ized by som n, and, important properties that put angle distinguish them from the family of amplitude ‘eforms shown in Figure 4.1 for the case of sinusoi- are the sinusoidal carrier 1d. modulating waves, nd 4.1 display the corresponding amplitude-modulated ul frequency-modulated (FM) waves, respectively "as (4.4) and (4.7), we readily see that the amplitude of PM and aves Js maintained at a constant value equal to the carrier amplitude A, ! time 1, irrespective of the sensitivity factors k, and ky. This property is demonstrated by the PM wave of Figure 4.1d and FM wave of Figure + Consequently, the average transmitted power of ‘angle-modulated waves ‘onstant, as shown by di Pav = 342 (49) it is agsumed that the | is Modulation Process is its nonlinear character. the principle of superposi- al m(t) is made up of two eed by m(r), m,(t), and Tight of this equation, we (0) = Accosnie + ben) 4m] si(0) = Accos!2n fer+ Kym) $2(t) = Ac COs [27 fet + kymz(t)) From these expressions, despite the fact that m(t) = m(1) + m2(t), we readily Sec that the principle of superposition is violated because s(t) # s\(t) + (0) A similar result holds for FM waves. ‘The fact that the angle-modulation process is nonlinear complicates the spectral analysis and noise analysis of PM and FM Waves, compared to amplitude modulation. By the same token, the angles Modulation process has practic its of its own. For example, frequency Modulation offers superior noise performance compared to amplitude mode Aation, which is attributed to the nonlir aracter of frequency modulation.PROPERTY 3 irregularity of Zero-Crossing> ‘ago to become Set my : consequence of allowing the instantaneous angle 449) 7) jc as in n(x) de as eae ‘message signal as in Eq. (4.4) or its integral An ger have # SF eres toe earo-cronsings of « PM or FM wave OWNER are See Topiary in thelr spacing. neross the time-scale. 267 foal ” as the instants of time at which a waveform chars te: Pe if In a Ways from a posit 4 positive to negative value ot the other way around. MT oo the regularity of zero-crossings in angle-modulated waves !6 also att scaliner cht modulation process. To illustrate this prove a oh the PM wave of Figure 4.1d ‘and the FM wave of Figure rea for the corresponding AM wave. or at Hip note that in angle modulation, the informs" fe 6 message signal m(y). resides in the zero-crossings of the moduaes wave, This statement holds provided the carier frequeney Je © large Compared to the highest frequency component of the message S!ET#! ano ion content of the: modulated Jess than 100 pet ion, This is not so in nportant ge of angle modulation over ampli ie realization of improved noise performance. This advant to the fact that the transmission of a message signal by m dof a sinusoidal carrier wave is less sensitive to the presence of add than transmission by modulating the amplitude of the carrier, The performance is, however, attained at the expense of transmission bandwidth requ pedulation. In other words, the use of angle rrodulation oat antes of exchanging an increase in transmission bandwidth for an improvement in noise performance. Such a trade-off is not possible with amplitude modi Iation since the transmission bandwidth of an amplitude-modulated wavy o fixed somewhere between the message bandwidth W and 2W, dependi me the type of modulation employed. The effect of noise on angle odilategn discussed in Chapter 6. ment in nois sponding increase in theEXAMPLE 4.1 Zero-Crossings Jer a modulatin; consid ce ee : a1 = 0, as shown by ) that increases linearly with time f, starting mi) ={% 1=0 RO Sy shere the lope parame Gee Figure 4.24), In wha follows, we iy ‘lowing set of parameters: waves produced by m(r) for the Be factor ky = § radians/volt. Applyin: Eq. (4.5) to the PMiave § radians/volt. Applying kya), 1=0. i) hwo modulated wane 0 10 0 (0) Frequaney modtlatd wave 4= ome fe mater [> FA wove seats fone of Frequany - PM wave Ac cos mfa) A, 008 (mf) GURE 4.3 Illustrating the relationship between frequency modulation and phase modulation for generating an FM wave by using a phase modulator, (B) scheme for generating a 1 wave by using a frequency modulator. 5 ne angle-modulated waveforms of Figure 4.2 should be contrasted with the correspond- "= ones of Figure 4.1. Whereas in the case of sinusoidal modulation depicted in Figure 4.1 \ s difficult to discem the difference between PM and FM, this is not so in the case of Hieure 4.2 In other words, depending on the modulating wave, it is possible for PM and FM to exhibit entirely different waveforms. Comparing Eq. (4.5) with (4.8) reveals that an FM signal may be regarded as a PM enal in which the modulating wave is ff m(r) dz in place of m(#). This means that an FM signal can be generated by first integrating m(#) and then using the result as the input ‘oa phase modulator, as in Figure 4.34, Conversely, a PM signal can be generated by first vitferentiating m(t) and then using the result as the input to a frequency modulator, as in Figure 4.3b. We may thus deduce all the properties of PM signals from those of FM sig- nals and vice versa. Henceforth, we concentrate our attention on FM signals, 4.3 FREQUENCY MODULATION = unlike amplitude modulation, the spectrum of an FM sig manner to that of the modulating signal; rather, its analysis is that of an AM signal. ATE, with AM modulation, that is, we consider modulation.1c of an FM signal is thatthe frequeney deviation ptal characteris ‘modulating signal and is i 1 to the amplitude of the frequency Using Eq, (4.11), the angle (0 of the FM signal is obtained as Ot) ae [ soe = 2nfet + Te sin\ontat) The ratio ofthe frequency deviation A/ to the modulation frequency called the modilation index of the FM signal. We denote it by fi.sam(t) = Ac cos(2afe “3 FREQUENCY MODU! frat) + [23 where jis the modulation ~Sin)t) + cos2n(fe—Sm)t}} (4.20) see that in the €as6 GE Ent . M sign Comparing, Eqs. (4.19) and (4.20), we and a nacrow-handiahniige aa on, the basic difference between an AM signal biked! ‘sign of the lower side frequency in the FM signal requires essentially the with a phasor diagram as shown in reference. We see that the resultant ‘to the carrier phasor. The ef- ow-band FM signal that 0 with re- intentional use of narrow-band FM for analog information sources is not common, unintentional narrow-band phase modulation is quite com- mon. This unintentional phase modulation i commonly referred to as phase noise. Phase noise is often introduced by oscillators in band-pass communi- Cotions and tise o umber of causes. Some causes are deterministic, such as lator temperature, supply voltage, physical vibration, magnetic field, humidity, or output load impedance ‘The phase noise die to these sources may be minimized by good design: Other sourcesbandwidth), say fj. This phas acking effectively nulls those frequency components of the phase noise less than f, 3 Phase rei which is outside the bandwidth of the message bandwidth W, is eliminated when the message is low-pass filtered From observation (1), if Figure 4.5 represents the spectrum of the hoisy O84 a ae Rea then the amplitude spectrum of the modulating anal dal approximately given by Figure 4.5. From the Rayleigh y theorem of Fourier transforms discussed in Chapter 2, time-domain and frequency-domain energies are equivalent, so we write J tocoPar= [ oupear shere @(/) is the Fourier transform of Gp). After the phase-locked loop, > tae : [i aoa | “onear+ [,,onrar where (1) excludes the carrier and frequency components of ¢,(#) below ji. The left-hand side represents the energy of the phase variations, so combining these results with the third observation, we find that the root- ‘mean-square phase error is = 7 Proms 2) |, ( f))? af radians Numerically integrating the phase noise spectrum of Figure 4.5 (see Problem 4.27) for fi = 10 Hz and W = 10 kHz, shows that the rms phase error is Since this value is less than 0.3 radians, the use of the narrowband FM s in the solution is justified. . WIDE-BAND FREQUENCY MODULATION We next wish to determine the spectrum of the single arbitrary value of the modulation index B. In general, an soidal modulating signal, as in Eq. (4.16), is in itself quency f. is an integral multiple of the modulation simplify matters by using the complex representation of band-p Chapter 2. Specifically, we assume that the carrier frequency f. is to the bandwidth of the FM signal) to justify rewriting Eq. (4.16) in the lA expli2afer +JB Sinrfnt))} e[5(r) exp(J2afer)) where §(1) is the complex envelope of the FM signal s(0), defined by 5(0) = Ac expliB sinha) 422) i i e is a periodic function Thus, unlike the original FM signal s(#), the complex envelope s(t) is a periodic of time with a fundamental frequency equal 10 the modulation frequency fy. We may therelore expand s{¢) in the form of a complex Fourier series as follows: a= nexpl 2h!) s(t)= (4.23)ND FREQUENCY, ATION where the complex Fourier coefficient ¢y is given by V2 Cohn | ldeRp(-Panfnt) ! io = fhe fee exp| jh sin(2rfit) — j27 ft] at Define a new variable: X= Inf Hence, we may rewrite Eq, (4.24) in the new form 5 Ff enl(fsins mola The integral on the right-hand side of Eq. (4.26), except for a scaling factor, is re as the nth order Bessel function of the first kind’ and argument fi. This function is e¢ monly denoted by the symbol J,({), as shown by ire pci i = nx)] de J(B)= 5 [explains — m0) Accordingly, we may reduce Eq, (4.26) to on = AcIn{(B) Substituting Eq. (4.28) in (4.23), we get, in terms of the Bessel function J, following expansion for the complex envelope of the FM signal: =A > lf) el ant Next, substituting Eq, (4.29) in (4.21), we get s(t) =Ae + Re bp In(B) expli2n(f. + rn) Interchanging the order of summation and evaluation of the real part in side of Eq. (4.30), we get In Figure 4.6 we have {f for different positive int havior of the inti2, For small values of the modulation index ji, we have 115 Jo(B) ~1 “43 FREQUENCY MODULATION B nip) = § 4.34) In(B) =, n>2 d 216)= (435) Thus, using Eqs. (4.32) through (4.35) and the curves of Figure 4.6, we may make the following observations: 1. The spectrum of an FM signal contains a carrier component and an infinite set of side frequencies located symmetrically on either side of the carrier at frequency separations + 2m 3f--+- In this respect, the result is unlike that which prevails in an AM . sinice in an AM system a sinusoidal modulating signal gives rise to only one pair of side frequencies, 2. For the special case of f small compared with unity, only the Bessel coefficients Jo() and J,(f) have significant values, $0 that the FM signal is effectively composed of a carrier and a single pair of side frequencies at f. + f,,. This situation corresponds to the special case of narrowband FM that was considered previously 3. The amplitude of the carrier component varies with f according to Jo(f). That is, un- like an AM signal, the amplitude of the carrier component of an FM signal is depen- dent on the modulation index . The physical explanation for this property is that the envelope of an FM signal is constant, so that the average power of such a signal developed across a I-ohm resistor is also constant, as shown by io sa (436 P= >A: ) When the carrier is modulated to generate the FM sign quencies may appear only at the expense of the power or the power in the side fre- nally in the carrier, thereby os ost oaNG es eng te power of an FM signal ™ PHASE AND FREQUENCY MODULATION. jon for the average Substituting Ba, (4.35) in 437) the expression iu g Bq. (4.35 6), and so it should. EXAMPLE 4.3 Spectra of FM Signals In this example, we wish to investigate the ways in whi plitode and frequency of sinusoidal odlating sigha affect the speci Of the FM signal, Consider first the case when the frequency of the modulating signal is fixed, but its amplitude is varied, producing a corresponding pee tion in the frequency deviation Af. Thus, keeping the modulation frequency ich variations in the am=fs Voltage controlled coseillator Wide-band ae Bond- | FM wave Frequency |) pass multiplier oe ‘equation (4.45) is suming sinusoidal m we may use the c oscillator, followed | permits the att put frequency ch achieve wide-ban ‘An FM transm tage that the ean generator is and the d whose out taneous fr next sub low-pass fi from its assis FIGURE 4.11 Block diagram of wide- band frequency modulator using a voltage-controlled oscillator. is being frequency-modulated rk, isapproximately given by “a yoltage-controlled This configuration ever, has the disadvan- ‘stable oscillator. It is there- ‘by which a very stable ‘carrier frequency. One - output of the FM sal-controlled oscillator, ne for the frequency stabil- 121122 PHASE AND FREQUENCY MODULATION DEMODULATION OF FM SIGN Frequency demodulation is the process signal from a frequency-modulated si teristic that is the inverse of that of rectly or indirectly. Here we involving the use el‘igure 4.13a. This complex transfer function is related to Hy(f) by AS -f)=2H(f), f>0 jence, using Eqs. (4.46) and (4,47), we get = Br ep Br adage 0, elsewhere «hich is plotted in Figure 4.13, convenience: ; [moa] Let Ai(f) denote the complex transfer function of the slope circuit defined by (447) (4.48) ‘The incoming FM signal s(2) is defined by Eq, (4.8), which is reproduced here for ed to the transmission bandwidth of the | Let 5)(f) denote (4.49) % e response of the slope circuit defined by Figure 4.13b due the theory described in Section 2.10, we may express the 123 ‘4.3 FREQUENCY MODULATION124 PHASE AND FREQUENCY MODULATION output is therefore A 1+ Mm] |si(0)| = aBraAcl1 + Be ‘aA; in the right-hand side of Eq. (4.54) is a of the transfer function of the slope circuit. by subtracting from the envelope-detector output [Fi : ae . detector preceded by the complementary slope circtl n plotted in Figure 4.13c. That is, the respective complex transfer functions slope circuits are related by ‘The bias term mBr A,(f) =AiP) Let s(t) denote the response of the complementary slope circuit p ing FM signal s(2). Then, following a procedure similar to that just the envelope of s2(0) is 3 \a(o = 28104. [1= Hm] where 53(r) is the complex envelope of the signal so(1). The envelopes in Eqs. (4.54) and (4.56) is sal) = 60] ~ 120) = srhyabem(t) which is free from bias, as desired. ‘We may thus model the ideal frequency discrimina with their complex transfer functions related by Eq. (4.55 tors and finally a summer, as in Figure 4.14a. This scher discriminator. The idealized scheme of Figure 4.14a shown in Figure 4.14b. The upper and lower | tuned to frequencies above and below the unm In Figure 4.14c we have plotted the gether with their total response, as ity factor or Q-factor of a resonant Itis formally defined as 27. times the r ‘one cycle to the energy dissipated p onant circuit, the Q-factor is126 PHASE AND FREQUENCY MODULATION ‘The specification of standards for FM stereo transmission js factors: 1. The transmission has to operate within the allocated FM broades 2. It has to be compatible with monophonic radio receivers. The first requirement sets the permissible frequency p deviation, The second requirement constrains the way in which configured. * Figure 4.15a shows the block diagram of the multip| stereo transmitter, Let m(#) and m,(z) denote the signals picked t hand microphones at the transmitting end of the system. ‘matrixer that generates the sum signal, m(t) + m,{t), and th crystal oscillator by frequency doubling) a ducing a DSB-SC modulated wave. In add lated wave, the multiplexed signal m reference for the coherent dete the multiplexed signal is dwhere fi = 19 KHz, and K is the amplitude of the pilot tone. The multiplexed signal m(t) squeney-modulates the main carrier to produce the transmitted signal. The pilot is, uljotted between 8 and 10 percent of the peak frequency deviation; the amplitude K in rq, (4.58) is chosen to satisfy this requirement, iver, the multiplexed signal m() is recovered by frequency demodu= ing the incoming FM wave. Then m(t) is applied to the demultiplexing system shown ) Figure 4.150. The individual components of the multiplexed signal m(1) are separated the use of three appropriate filters. The recovered pilot (using a narrowband filter ned to 19 KHz) is frequency doubled to produce the desired 38-kHz subcarrier. The vilability of this subcarrier enables the coherent detection of the DSB-SC modulated wwe, thereby recovering the difference signal, mr) ~ m,(1). The baseband low-pass fil- in the top path of Figure 4.15 is designed to pass the sum signal, m(t) + m,(0). Finally, the simple matrixer reconstructs the left-hand signal m,(1) and right-hand signal and applies them fo their respective speakers, & then close division/multiplication, Jatter application is the subj Basically, the phase-loch .op filter, and a voltage a feedback loop, as in Figure determined by a voltage modulator may serve as ‘We assume that in is zero, two conditions 1. The frequency of the: where Ag is the iii) by the integral 1 “44 PHASE LOCKED LOOP128 PHASE AND FREQUENCY MODULATION ‘The object of the phase-locked loop is to generate a VCO output same phase angle (except for the fixed difference of 90 degrees) as the si), The time-varying phase angle (0) characterizing s(¢) may be due ‘a message signal m(t) as in Eq. (4.60), in which case we wish to recover estimate m(z). In other applications of the phase-locked loop, the angle (2) of the incoming signal s(¢) may be an unwanted phase ations in the communication channel; in this latter case, we wish 10 produce a signal with the same phase angle for the purpose of chronous demodulation). To develop an understanding of the phase-locked loop. NONLINEAR MODEL OF THE P According to Figure 4.16, the incoming plied to the multiplier, producing two comy 1. A high-frequency component,FIGURE 4.17 Nonlinear model ofthe phase- locked loop. ‘model shown in Figure 4.17 ‘model resembles the phase-locked loop is integrator. ‘increases the diffi to linearizeaeky | dele)n(e= 9) d= tueney domain and solving Fourier transform of ¢i(?).\ dt Transforming Eq. (4.69) into the freq transform of ,(0), in terms: of &,(f), the on ee” ‘The function L(f) in Eq. (4.70) is defined by where H(f) is the transfer function of t open-loop transfer function of inside the baseband we makeFquation (4-77) states that when the loop operates in its phase-locked mode, the output (1) of the phase-locked loop is approximately the same, except for the scale factor kj/ki. 's the original message signal m(1); frequency demodulation of the incoming FM signal 1) is thereby accomplished. significant feature of the phase-locked loop acting as a demodulator is that the andwidth of the incoming FM signal can be much wider than that of the loop filter haracterized by H(f), The transfer function H(f) can and should be restricted to the cband, Then the control signal of the VCO has the bandwidth of the baseband (mes- *) signal m(a), whereas the VCO output is a wideband frequency-modulated signal nose instantaneous frequency Te ‘incoming FM signal. Here we are nerely restating the fact that the bandwidth of a wide-band FM signal is much larger han the bandwidth of the message si ible for its generation. The complexity of the phase-lo is determined by the transfer function /) of the loop filter. The simp! ocked loop is oblained when that is, there ino loop filter, a locked loop is referred to rder phase-locked loop. For ‘a more complex form, Th order of the denominator polyn the output transform Vif) in terms ¢ A major limitation of a first Ko controls both the loop b the hold-in frequency range ref phase-locked to the input signal seldom used in practice. A a second-order phase-loy 2a SECOND-OR 2 To be specific, co transfer functionand the damping factor: ‘ASE AND FREQUENCY ODULATION ®(f) = Assume that the incoming FM \which the phase input is Hence, from Eq. (4.82) weThe Fourier transform of the | loop output is related to @,(f) by Eq, (4.72); hence, ith H(f) as defined in Eq, (4.78), we get ; a vin= Foun 4.87) in light of the definitions given in Eqs. (4.80) and (4.81), we have vo-(Abox( 9) .(f) (4.88) pstituting Eq. (4.82) in (4.88), we get Cif ke) + 26 iF /fn)] v(f)= aif) (4.89) ( + 2CifTh) + it /tey) 1) a wrelore, for the phase input (0) of Eq. (4.83), we find that the corresponding loop L(t) = Ag cos(2nfgt + 2) (4.90) ere the amplitude Ap and phase 2 are, respectively, defined by p= — OLD stale E (91) {11 — Sle)” + 42S /fa)?} | is | 20 fini fa) | 0 tan ‘Ps AG T= Gah ao From Eq, (4.91), we see that the amplitude Ag attains its maximum value of Affk, at \fnify) = 0; it decreases with increasing f,,/f,, dropping to zero at ( f»/f,) = 90 The important feature of the second-order phase-locked loop is that with an incom= ing FM signal produced by a modulating sinusoidal wave of fixed amplitude (corre- sponding to a fixed frequency deviation) and varying frequency, the frequency response hhat defines the phase error de(t) is representative of a band-pass filter [see Eq. (4.85)], but the frequency response that defines the loop output v(t) is representative of a low- pass filter [see Eq. (4.91)]. Therefore, by appropriately choosing the parameters ¢ and fn. which determine the frequency response of the loop, it is possible to restrain the phase error to always remain small and thereby lie within the linear range of the loop, whereas at the same time the modulating (message) signal is reproduced at the loop out- put with minimum distortion. This restraint is, however, conservative with respect to the hold-in capabilities of the loop. As a reasonable rule of thumb, the loop should re- ‘main locked if the maximum value of the phase error 9 (which occurs when the modulation frequency fy is equal to the loop’s natural frequency f,) is always less than 90 degrees. E Phase-locked loop performance is explored experimentally in Problem 4.29. 133 “45 NONLINEAR EFFECTS INFM SYSTEMS 4.5 NONLINEAR EFFECTS IN FM SYSTEMS {In the preceding three sections, we studied frequency modulation theory and methods for its generation and demodulation. We complete the discussion of re Eo by considering nonlinear effects in FM systems.in one form or another, are present in all electrical are two basic forms of nonlinearity to consider: ¥y is said to be strong when itis i i inner for some specific application, Examples of strong law modulators, limiters, and frequency multipliers. cs of a parasitic nature arise due to imperfections. The effect nonlinearities is to limit the useful signal levels in a system and important design consideration, sa In this section we examine the effects of weak nonlinearities on fr. Consider a communications channel, the transfer nl by the nonlinear input-output relation eae vole) = arvi(t) + an03(0) where u,(z) and v,(9) are the input and outp constants. The channel described signal v,(7) is an instantaneous storage involved in the descripfo > 3Af + 2W 4.96) 1s, by using a band-pass filter of mid-band fr ne! output {eae requency f, and bandwidth 2Af + 2W, the 3 at vat (a Act jay Jeostanfet + H(t) (4.97) therefore that the only sie noni ‘Sone Passing an FM signal through a channel with am- Tha ae by appropriate filtering, is simply to modify its ampli- : © amplitude modulation, frequency modulation is not affected by ortion produced by transmission through a channel with amplitude nonlinearities. It is reason that we find frequency modulation widely used in microwave radio and ic communication systems: It permits the use of highly nonlinear amplifiers and Particularly important to producing a maximum power out- 1 this atell power transmitters, which are at radio frequi An FM syst acies. n is extremely sensitive to phase nonlinearities, however, as we would uitively expect. A common type of phase nonlinearity that is encountered in micro- ive radio systems is known as AM-to-PM conversion. This is the result of the phase characteristic of repeaters or amplifiers used in the system being dependent on the instan- taneous amplitude of the input signal. In practice, AM-to-PM conversion is characterized by a constant K, which is measured in degrees per dB and may be interpreted as the peak phase change at the output for a I-dB change in envelope at the input. When an FM wave 1s transmitted through a microwave radio link, it picks up spurious amplitude variations due to noise and interference during the course of transmission, and when such an FM wave is passed through a repeater with AM-to-PM conversion, the output will contain un- wanted phase modulation and resultant distortion, It is therefore important to keep the AM-to-PM conversion at a low level. For example, for a good microwave repeater, the AM-to-PM conversion constant K is less than 2 degrees per dB. 4.6 THE SUPEI 4.6 THE SUPERHETERODYNE RECEIVER In a communication system, irrespective of whether it is based on amplitude modulation or frequency modulation, the receiver not only has the task of demodulating the incoming modulated signal, but it is also required to perform some other system functions: © Carrier-frequency tuning, the purpose of which is to select the desired signal (.e., desired radio or TV station). © Filtering, which is required to separate the desired signal from other modulated sig- nals that may be picked up along the way. © Amplification, which is intended to compensate for the loss of signal power incurred in the course of transmission. ‘The superhererodyne receiver, oF superhet as itis often referred to, is a special type of receiver that fulfils all three functions, particularly the first two, in an elegant and prac- tical fashion, Specifically, it overcomes the difficulty of having to build a tunable high- (and variable-) Q filter. Indeed, practically all analog radio and TV receivers are of the s 4 a ee consists of a radio-frequency (RF) section, a mixer and local oscillator, an intermediate frequency (IF) section, demodulator, and power amplifier Typical frequency parameters of commercial AM and FM radio receivers are listed is Table 4.2. Figure 4.21 shows the block diagram of a ‘superheterodyne receiver for ampli- tude modulation using an envelope detector for demodulation.TABLE 4.2 Typical frequency parameters of AM and FM radio 0,535-1.605 MHz Mid-band frequency of IF section 0.455 MHz IF bandwidth 10 kHz ‘The incoming amplitude-modulated wave is picked t amplified in the RF section that is tuned to the carrier fr ‘The combination of mixer and local oscillator (of « defined by where fio isthe frequency coming RF signal. Wejm. The only practical cure for image inte he RF section (i.e,, between the po th en the antenna and the mixer) in order to favor the desired ped Titan uN the undesired or image signal. The effectiveness of appressing unwanted image signals incre, : er frequency section inerea en ineteases as the number of selective stages in the i “teases, and as the ratio of intermediate to signal frequency ference is to employ highly selective stages The basic difference between AM an FM demodulator such ¢ information is tr \M and FM superheterodyne receivers lies in the use as limiter-frequency discriminator. In an FM system, the 1 cater WR ian of the instantaneous frequency of a si- Pe tee iplituide is maintained constant. Therefore, any variations a an 4 fe at the receiver input must result from noise or interference. An i + Unleae ee ing : eS on, is used to remove amplitude variations by es, : ave at the IF section output almost to the zero axis. The result- rectangular wave is rounded off by a band-pass filter that suppresses harmonics of the rriet frequency. Thus the filter output is again sinusoidal, with an amplitude that is yctically independent of the carrier amplitude at the receiver input (see Problem 4.20). 47 THEME ANALOG ANC CELLULAR. ..7 THEME EXAMPLE—ANALOG AN rAL FM CELLULAR TELEPHONES NG” 1 this example, we consider two applications of an FM modulator that are both related to cllular telephone service. The initial cellular telephone system in North America was nown as the Advanced Mobile Phone Service (AMPS) and went into operation in 1983. he AMPS system uses 30 kHz channel spacing, that is, two 30 kHz channels, one in each jirection is assigned to each user for the duration of call. This method of sharing the radio pectrum is known as frequency-division multiple access (FDMA). The two channels (up- link and downlink) are separated by 45 MHz ina frequency band from 824 to 894 MHz. In AMBS, analog frequency modulation is used for voice transmission, and frequency- hift keying (see Chapter 9) is used for data transmission. As with wired telephone service, he voice bandwidth (W) is limited to approximately 3 kHz before transmission. The FM modulator is designed such that the peak deviation due (o the voice is limited to 12 kHz. ising Carson's rule of Bq. (4.38), with Af = 12 kHz, and replacing f,, by W, the approxi- nate Value of the transmission bandwidth of the AMPS signal is obtained as Br = 2(Af + W) = 2(12 + 3) =30kHz Erhis transmission bandwidth estimate agrees with the assigned channel spacing of 30 kHz. Since FM is a constant-envelope modulation technique, AMPS mobile units can w elficiency power amplifiers. In particular, power amplifiers could be operated in saturation ({aclitating high efficiency) without distorting the output envelope, since it is constant. The constant-envelope property also has advantages for combating the fading that occurs on mo- bile radio links. One serious drawback of the analog FM system was that it provided no pro- tection from eavesdroppers. The AMPS was the first system to introduce the cellular concept for frequency reuse. However, the success of AMPS was the progenitor of its own demise since greater demand for the limited radio spectrum meant that more bandwidth-efficient transmission techniques had to be found. One of the successors to AMPS is the digital cell-phone standard known as GSM (Global System for Mobile Communications). GSM built on some of the FM-related advantages of AMPS but uses a more complex multiplexing strategy and a digital Tepresentation of the data to reduce the bandwidth requirements. To understand theAmpitudeSpectum (8) bon Sock & 22 Spectrum of baseband pulse used in GSM.AMPS. The indiv time-divis In this chapter, continuous-war angle is varied in Angle modulation c Proportion to the t that is varied in pr as the deri cordingly, F one, we can ¢ used in bro voted to FM.140 FM. In particular, we derived an empirical rule known as Carson's rule for an PHASE AND FREQUENCY mate evaluation of the transmission bandwidth By of FM. According to this rul MODULATION controlled by a single parameter: the modulation index ff for sinusoidal FM, or th ation radio D for nonsinusoidal FM. In FM, the carrier amplitude, and therefore the transmitted average power, tained constant. Herein lies the important advantage of FM over AM in combatin fects of noise or interference at reception, an issue that we study in familiarizing ourselves with probability theory and random processes in Cl advantage becomes increasingly more pronounced as the modulation index (de itio) is increased, which has the effect of increasing the transmission bandwidth | sponding way. Thus, frequency modulation provides a practical method ff of channe! bandwidth for improved noise performance, which is not fe plitude modulation, NOTES AND REFERENCES |, Bessel functions play an important roe in the solution ofa certain differential in the mathematical formulation of many physical problems. For a detailed ject, see Wylie and Barrett (1982, pp. 572-625). ‘When « phase-locked loop is used to demodulate an FM wave, the loop incoming FM wave and then follow the variations in its phase, During the lo phase error (between the incoming FM wave and the VCO output will ing the use of the nonlinear model of Figure 4.17. For a full treatment of ‘of phase-locked loop, see Gardner (1979), Egan (1998), and Best (2003). For a detailed discussion of the characterization and system effects of see “Transmission Systems for Communication,” Bell Telephone Labo (Western Electric, 1971). PROBLEMS 4.1. Skeich the PM and FM waves produced by the savwtooth Aeavene. he delays and he wave shown in Figure P4.1 ‘mitted signal. mt ° ae 2 3 Figure P4.1 4.2. In a frequency-modulated radar, the instantaneous fre- ‘quency of the transmitted carrier is varied as in Figure P4,2, which is obtained by using a triangular modulating signal. The i taneous frequency of the received echo signal is shown aaah 3 3 Figure P42, where r is the round-trip delay time. Tena . and received echo signals are applied to a mixer, and the ‘ence frequency component is retained, Assuming | determine the number of beat cycles at the mae ‘over one second,Approximately defined by s(t) = Ac cos(2nfit) = fac Sin(2rfa)sin(2njyt) Determine the envelope of this modulated Signal, What is the rato of the maximum fo the minimum value velope? Plot this ratio versus ao 1o the interval 0 = f = 0,3, Determine the average power of the expressed as a percentage of the avera modulated carrier wave, Plot this resu that f is restricted to the interval 0 By expanding the angle 0,(¢) of the s() in the form of @ power lation index B toa maximum O:(t) = Infet +B is applied to a phase mod modulated carrier wave has Construct a pha compare it with that of the « signal. tas an arbitrary val ‘modulated signal is appl band frequeney fe and quency of the mod time, sensitivity of the modulator? What is the value of Ay for which the carrier component is reduced to zero for the second time? 4.9 An FM signal with modulation index f = 1 is transmitted through an ideal band-pass filter with mid-band frequency f. and bandwidth Sf, where f. is the cartier frequency and fy is the fre- quency of the sinusoidal modulating wave. Determine the tude spectrum of the filter output,4.13 An PM signal with a frequency deviation of 10 kHz ‘modulation frequency of 5 kHz is applied to two frequency mult: Pliers connected in cascade, The first multiplier doubles the fre- {quency and the second multiplier triples the frequency. Determine the frequen: jodulation index of the FM sig- is the fre- es of this FM deviation and the nal obtained at the second multiplier output, quency separation of the adjacent side frequen ignal 4.14 An PM signal is applied to a square-law device with out- Put voltage v related to input voltage v; by where a is a constant. Explain how such a device can be used to obtain an PM signal with a greater frequeney deviation than that available at the input 4.15 Figure P4,15 shows the frequency-determining network of a 4 vol controlled oscillator. Frequency modulation is pro- by applying the modulating signal Ay sin(2nfqf) plus a bias V;, to a pair of varactor diodes connected across the parallel com- bination of @ 200-aH inductor and 100-pF capacitor. The eapaci- tor of each varactor diode is related t0 the voltage V (in volts) applied across its electrodes by C= 100V> pF aH Vi + Ay 5 rth Figure P4.15 The unmodulated frequency of oscillation is | MHz, The VCO ‘output is applied to a frequency multiplier to produce an FM sig- nal with a carrier frequency of 64 MHz and a modulation index of 5. Determine (a) the magnitude of the bias voltage V, and ¢b) the amplitude A,, of the modulating wave, given that f,, = 10 kHz, 4AG/ “Te FM sign o=csabiggeae al detector does not load the filter. Determine the resulting i ‘envelope detector output, assuming that ky |m(2)| Je £0 4,17 In the frequency discriminator of Figure 4. quency separation between the resonant frequenci parallel-tuned LC filters be denoted by 248, where bandwidth of either filter and & is a scaling factor. both filters have a high Q-factor. 4 (a) Show that the total response of both filters hi to 2/B(1 + #)S? at the center frequency fo. Let D denote the deviation of the total respon with respect to a straight line passing: this slope. Plot D versus 6 for k = 1.5 and where 5 = f= f. ) 4.18 Consider the frequency demodulation sel Figure P4.18 in which the incoming FM signi through a delay line that produces a phase-shift of the cartier frequency f,. The delay-Hine output is the incoming FM signal, and the resulting s(t) = Ac cos nf + f sin( analyze the operation of this nol 7 index fi is less than unity and the delay line is sufficiently small to justify maki| Freq = [0:FS/FFTsize:Fs/2}, Subplot(2,1,1), plots), xlabel( Time (ms), label Amplitude); axis(10 0.5 ~1.5 1.5), grid on subplot(2,1,2), stem\Freq sqrt S/682)) xlabell'Frequency (kHz), ylabel(’Amplitude Spectrum’); axis(!95 105 0 1), grid on (a) t= (0:Ts:10*TSI; %— FIR differentiator (Fs = +1024 kHz. oe FIRGiff = [1.60385 0.0 0.00.0-0.00.00.0-0 , and 10, determine the | — 1.603851; oe) Power in the harmonics of the modulating frequency about BP_diff = FiRaiff * exp(j*2“pi f0*t; the carrier (ignore sidelobes). How many side-frequencies Butterworth filter -Fs = 1024 are required for 90% of the power in each case? see (b) For what minimum modulation index is the power at the cartier frequency reduced to zer0? For modulation indices of 1, 2, 1£-4710.0706 0.2117 0.2117 01 1.0000 ~2,9223 2.8476 ~0.925 4.27 Use the following Matlab script to numerically estimate the | D1 = feta a 3 ms phase error for a signal having the spectrum of Figure 4.5. | D2= EnvDetect(D1); D3 = filter(LPF_B,LPF_A, D2); %—One-sided phase noise spectrum @ t 1101001000 10000); %He SdB = [-30-40~50~65~70}; % Spectrum (dBc) What differences are observed bet and the original? Why? (b) Modify the modulating signal to the (0) = sin(2zfyt) + 0.5 %-—interpolate spectrum on linear scale— aod % integration step size (He) Peer eee: (10: det: 100001; eH Ann peed $1 = interp1(f, 10.(SAB/10), fl); % absolute power eae | s%—numerically (Riemann) integrate from 10 He to 10 kHz— tion in the output appear? Why? | ints = sum(s1)* del_f; Theta_rms = sqrt(int_S); % in radians 4.29 The following Matlab seriptis‘a di Theta_rms = Theta_rms*180/pi __% in degrees ing the behavior of a phase-locked loop 3 following: Repeat for the phase noise spectrum of Figure P4.27 where the range ‘he Following: of frequencies of interes is from 100 Hz to 1 MHz. (0). Compare hee stantae 30 VCO, and loop filter to the a g the text. Justify the d ) © @) © “Amplitude Spectrum (ac)