+1 ,

Radio-FrequentrU Eircuits

2.1 2.2
2.3

ktroduEtion High-Frequencg Effects

Radio-Freqrrencg

OBJECTIVES After studying this chapter,you should be able to: . Explain the differencesin operation and constructionbetweencircuits that operateat low frequenciesand those that operateat radio frequencies, . Describe the characteristicsof ClassA, B, and C amplifiers and decide which type is the most suitable for a given application, . Analyze radio-frequency amplifier circuits, both narrowband and broadband, and choosethe correct configuration for a given application, . Describe, draw circuits for, and analyzethe most common types of radio-frequency oscillator circuits and discusstheir relative stability, . Explain the operation of varactor-tunedvoltage-controlled oscillators and calculate the variation of frequency with tuning voltage, . Describe the operation of crystal-controlled oscillators and explain their comparedwith LC oscillators; perform and disadvantages advantages frequency-stability calculations for crystal oscillators, . Describe the function of a mixer and analyzeseveraltypes of circuits, explaining how and where they are used and calculating output frequencies, . Analyze frequency synthesizersemploying phase-lockedloops. . Explain the usesof integrated circuits in radio-frequency systems.

Amplifiers s ve
rlse at )f

2.1

Radio-Frequencg OsEillators

2.5 2.6

Mixers Frequencg Sgnthesizers

IS,

rn

any of the circuits in this chapter predatethe inventionof the transistorin 1948.Vacuumtubes can do almost anythingthat
solid-state devices can do, though their size and power requirements make complex circuits difficult to construct economically-Even before vacuum tubes, however, relatively high-power radio-frequency signals could be generated using specialized mechanical ac generators called alternators. The first practical alternator transmitter was designed in 1906 by General Electric engineer Ernst Alexanderson for a station at Brant Rock, Massachusetts. lt generated a power of 2 kW at a frequency of 100 kHz.

Alternatrrrs Oscillators

as

2.1 Introduction
In the previous chapter, we looked at some of the signals found in a communication system.In particular, we saw the need to modulate a signal using an information (baseband)signal. The carrier must be much higher in frequency than thc basebandsignal. Carrier frequenciescan be as low as a few kilohertz but are typ ically much higher: megahertzor hundreds of megahertz.Microwave communications use carrier frequenciesin the gigahertzrange. You are probably already familiar with amplifier and oscillator circuits that operate at audio frequencies.In this chapter, we will explore some of the differences in design and construction that permit these circuits to work at radio frequencies.We will also look at some techniquesthat are used in radio-frequency (RF) circuits but impossible or impractical to implement at lower frequencies.In addition, we will discussdevices such as frequency multipliers and mixers, which allow the frequency of a signal to be changed. The purposeof this chapter is to give some insight into RF circuit design and "building btocks" that can be used in to provide the teadet with some electronic later chaptersas we look into the design and construction of practical transmitters and receivers.

Effects 2.2 High-Frequency

When you first began to study electronics, you probably divided the frequency spectruminto two parts: ac and dc. A capacitoq for instance,would be considered an open circuit for dc and a short circuit for ac. This simplifying assumption works well in arriving at a generalunderstandingof an audio amplifier circuit and in calculating its gain at midband frequencies. Later on, you needed to be more careful in considering the effects of frequency. The bypass capacitor that looked like a short circuit at 1 kHz was no longer so simple at2OMHz.In order to find the low-frequency responseof a simple audio amplifier, you found it necessaryto consider capacitive reactance.Similarly, other capacitanceshad to be taken into account to calculate the highfrequency response.Some of these capacitancesinvolved actual components in the circuit, while others were incidental parts of components,for example,juncin tion capacitances transistors.
44

Section ?.? O Hlgh-Freqrencg Effects ne rat rts rre tls ed tn at Perhaps amplifier containsa transformer or other inductive element.Again, an i nrst approximation might simplify an inductor as an open circuit for ac and a -rrrrt circuit for dc. A transformerwould be represented ideal. Further investigaas :rrn. however,especially at the extremesof the amplifier's frequency range, might =real the need to consider not only inductive reactancebut also lossesin the iron :!re of the inductor or transformer. As we extend our study of electronic circuits to higher frequencies,we have r- be more careful to include reactive effects, not only those that are included de:'nerately as circuit elements but also the "stray" reactancesin components and :.en within and between wires and circuit board traces.As we get still higher in :iquency, into the UIIF range, we find that conventional devices and construc--,xr methods become inefflcient and innovative approachesto circuit design be:'me important. At microwave frequencies,many circuits seemto bear very little :nt sical resemblanceto those used at lower frequencies. \[icrowave-circuit design will be discussedin a later chapter.For the present, le Jrall look at more conventional circuitry operating in the range between approxm;uelv 300 kHz and 300 MHz, that is, in the MF, HF, and VF{F frequency bands.

Imcaorlnam the

The Effect of Frequencyon DeviceCharacteristics

{ electronic devices, whether active or passive, have capacitancesand induc:rm-esthat are not included intentionally but are an inevitable result of the design rcrl e'onstructionof the component.A capacitor, for instance,will exhibit induc:rn-e and resistanceas well as capacitance.It can be representedby the equivamt circuit in Figure 2.1(a). The seriesinductive component Ls is mainly due to m leads.The resistive component can be divided into two parts, a small series :r:mponent R5 due to lead resistanceand a large parallel resistanceRp representn._:dielectric losses. As the frequency increases,so does the inductive reactance.Meanwhile, the :rgncitive reactance of the component decreaseswith increasing frequency. r;enrually, a point will be reachedwhere the two reactances equal and the caare rL-rtor becomes a series-resonant circuit. This point is called the self-resonant ftiquency. Above this point, the magnitude of the inductive reactance becomes seater than that of the capacitive reactance,and our so-called capacitor behaves llre an inductor. Similar effects occur in transistors. Consider an ordinary bipolar transistor *.r-h as the 2N3904, for instance.Each of the two junctions has capacitancethat ::,n be represented by capacitors drawn in between base and emitter and between ::ilector and base, as shown in Figure 2.1(b). The size of these capacitors demrls partly on the physical structure of the transistor and partly on its operating lrqnt. As frequency increases,the capacitive reactanceswill decreaseuntil the gtbrmance of the transistor is degraded.The base-to-collectorcapacitance,for :smnce, will causefeedbackfrom output to input in an ordinary common-emitter r-mplifier circuit. The feedback can lower the gain of the amplifier or cause it to =L-ome unstable. As the frequency increasesinto the gigahertzrange, transit-time effects also -R*orDeimportant. The transit time is the time it takes a charge carrier to cross a ivice. In an NPN transistor,it is the time taken for electronsto cross the base;in ; PNP transistor, the holes exhibit transit time. In general, free electrons move nore quickly than holes, so NPN transistors are preferred to PNP for highrequency operation.

e typmuni:s that liffero freuency Io rS. *'hich n and ;ed in utters

?
I

aR" 1 I

lency dered ption it and f treas no r simSimhighntsin junc-

5z * g
/
l

?--" +1 5>R ,
?*
I

I I
I

a
(b) Tlansistor (a) Equivalent junction circuit for a capacitance capacitor Figure C.l High-frequency effects

fhapter

? | Radlo-Frequencu

fircults

Transit time can be reduced by making devices physically small, but this causesproblems with heat dissipation and breakdown voltage. It is difficult to remove large quantities of heat from a small area, and of course the dielectric strength of insulators is proportional to their thickness. There are limits to how small devicescan be made, particularly if considerablepower must be dissipated. For this reason, some devices include transit time as part of their design. Rather than simply making transit time as small as possible, for instance,it might be designed to be one or a number of periods at the operating frequency. Techniques such as these are used mainly in the microwave region of the spectrum and are in discussed Chapter 17.

Lumped and Distributed Constants

At low frequencies,we generally assumethat capacitorshave capacitance,resistors have resistance, and short sections of good conductors (for example, the traceson a circuit board) have neither.We saw in the previous section that this assumption is really a simplification that becomesless accurateas the frequency inFor instance,a circuit board trace has a small amount ofinductance in adcreases. dition to resistance.There will also be capacitancebetween this trace and every other trace on the board. A little thought will show that the inductance and capacitancerelated to this trace cannot be shown exactly by a single capacitor and a single inductor. In fact, "lumps" of capacitanceand inductancewill do, since they are no finite number of both distributed along the entire length of the fface. We can often approximate these distributed constantsby a few lumped constantswith reasonableaccuracy, but you need to remember what is really happening, becausesome problems can be solved only using distributed constants.Figure 2.2 givesyou an idea of how an ordinary circuit board trace might look when analyzed this way. Of course, there is really no accurateway to sketch this trace, since the number of capacitors,resistors, and inductors should be infinite. At frequenciesin the UHF range and up, conductors even a few centimeters in length can no longer be ignored or have only their lumped capacitanceand inductance taken into account. They must be analyzed as transmission lines. This processincludes the distributed constantsthat are really there all the time but can usually be ignored at lower frequencies.At microwave frequencies, almost all conductors must be analyzed in this way. Transmission lines will be covered in detail in Chapter 14.

A circuit-board trace Figure ?,e using distributed constants

High-FrequencyConstructionTechniques

"stray" capacitanceand inIt is possible to design circuitry to reducethe effect of ductance resulting from the wiring and circuit board traces themselves.In general, keeping wires and traces short reduces inductance, and keeping them well separatedreducescapacitancebetween them. Inductive coupling can be reduced by keeping conductors and inductors that are in close proximity at right anglesto each other. The use oftoroidal cores for inductors and transformers also helps to reduce stray magnetic fields. There may seem to be contradictions here. For instance, how do we keep componentsfar away from each other while simultaneouslykeeping connections short? Obviously, compromise is necessary. Another way to reduce interactions between componentsis to use shielding. Coupling by way of electric fields can be reduced by shielding sensitivecircuits

Sectlon 2-2 f High-Frequencu Effects t this lo re:ctric how rated. ather e derques d are rith any good conductor, such as copper or aluminum. Ideally, the shielding *rould form a complete enclosureand be connectedto an earth ground, but even i piece of aluminum foil glued to the inside of a plastic cabinet and connectedto fr circuit common point will sometimesprovide adequateshielding. This "cheap ind dirty" technique is often found in consumer electronics. When double-sided :rrcuit boards are used, most ofthe copper is often left on one side and connected :,r sround. This ground plane can provide useful shielding. Figure 2.3 shows a lpical VHF circuit, in this case a cable-television converter, that uses the techrrques outlined here.
F-gur* H"3 VHF circuit layout

47

ESlS-

. the
F AS-

i inn ad1ery r this tact, I are mate 13cy, ; can

Fan

here

:ters I infhis
! dll

t all din

I inlPn-

r ell t.ced sto tsto ieep |t]ns trlg. uits

Generally, conductors in RF circuits are kept as short and as far away from rthers as possible. The neat right-angle circuit-board layouts common at low rnequencies often avoided in favor of more direct routing. Such layouts may are nrt be as pretty, but they can avoid problems of inductive or capacitive coupling. Anyone attempting to service RF equipment should remember that in highrrequencycircuits the placement of componentsand wiring can be critical. Movilrq one component lead may require a complete realignment. In fact, sometimes .i length of solid insulatedhookup wire (a few centimeterslong) will be seen,conrccted at only one end. This wire is called a gimmick; it provides a small capacrnnce to ground that is adjusted during circuit alignment by bending the wire .lightly in one direction or another. ln order to prevent RF currents from traveling from one part of the circuit to .irnother, careful bypassing is necessary. For instance,RF energy may travel from ,:xreslageto another stagethat sharesthe samepower supply. At first glance, this Decoupling Amp i rnav seem unlikely, since power supplies typically contain large electrolytic caI vcc 100f,) pacitorsthat should have very low reactanceat radio frequencies.Thus it would be I 0.001pF erpectedthat the power supply would look like a short circuit at high frequencies. Unfortunately,this assumptionignores the inductancein the leads from the circuit iLrthe power supply and in the electrolytic capacitorsthemselves.To prevent en;ryy from traveling from one circuit to another via this route, it is necessaryto provide small capacitors to ground right at the power connection to each stage. Small capacitorsare actually better than large electrolytics for this application, be;ause they have less inductance.In difficult cases,either an inductanceor a resistFigure t"4 Decoupling circuit .rncecan be added in the lead from the power supply to further discourage the u'ansferof RF energy. Using an inductor reducesthe dc voltage drop, of course, rnd the inductor must be chosenwith care becauseinductors, like capacitors,can erhibit self-resonance. Figure 2.4 shows a typical decoupling circuit between an emplifier stageand a power supply. If the power leadspassthrough a shieldeden;losure, a feed-through capacitor can be used to maintain the shielding.
I

Ehapter

e a Radio-Frequenrg

Elrcults

SEETION E.E REVIEW OUESTION

Why is it not possible to assume that a power supply is a short circuit at high frequencies?

2.3 Radio-FrequencyAmpliliers
Amplifiers for RF signals can be distinguished from their audio countelparts in several important ways. Wide bandwidth may or may not be required. If it is not, gain can be increased and distortion reduced with the use of tuned circuits. Depending on the type of signal to be amplified, linearity of output with respect to input amplitude may or may not be required. If linearity is not necessary, efficiency can be improved by operating amplifiers in Class C, which will be described shortly. Impedance matching is likely to be more important than at lower frequenciesbecauseofthe possibility oftrouble causedby signal reflections.

NarrowbandAmplifiers
Often the signals in an RF communication system are restricted to a relatively narrow range offrequencies. In such circumstancesit is unnecessary and, in fact, undesirable to use an amplifier with a wide bandwidth. Doing so invites problems with noise and interference. Consequently, many of the amplifiers found in both receivers and transmitters incorporate filters to restrict their bandwidth. In many casesthese filters also increase the gain of the amplifier. The simplest form of bandpass filter is, of course, a resonant circuit, and these are very cornmon in RF amplifiers. Consider the bipolar common-emitter amplifiers shown in Figure 2,5, for instance.Figure 2.5(a) shows a conventional amplifier using RC coupling. As you will recall, this amplifier is generally biased so that the emitter voltage is about lOVoof Vs6. T}lrecollector resistor will drop another 40Vo or so, leaving the voltage between collector and emitter, Vg6, at about one-half the supply voltage. The voltage gain is given approximately by

. ^r:
where

- (Rc Rr) ll I

(2.r)

Au : voltage gain as a ratio: Au: v6lv; Rc ll R, : parallel combination of the collector resistance and the load resistance rL : ac emitter resistanceof the transistor

Do not confuse r!, a transistor parameter, with R6, which is part of the external circuit. The value of rj depends on the bias current. It is given, very approximately, by ' ', e where -26mY
IE

r'": emitter resistance ohms in Is : dc emitter current in amperes

The tuned amplifier in Figure 2.5(b) is similar but not identical. For instance, the bias point is different. In the absence of a collector resistor, the dc collector

section

2.3 | Radio-Frequencg

Amplifiers

49

I Vc"
-.- "---i---i

: l . l i : 3 r 3 | c , . ' i i , l i t l ' t !
l i f l r
. TJf

^ v"" + ;1T: [.;i

"',1' Rr
.a

l :

r t r

, _|*_____\_,
.11 t l

l l

ruln s not, . Derct to , effie deLOWer

:, o"
+:

t-

- - ' - - . ' ' - - - r i " - - - - - " ' ' 1 Uo

D . tLB1

*---t I
Eu,

i r"'

CC"
I

>
.,., Rr,

CC'

.t

- =__.1 u; :)__ - ,.i i,."-______+_ , ik { \ p C C ,. . t.B2

i,/ /

l 1

nu ill
t l

'r' ca :
i l

- l l

t l

t - c o

..'

R o ,

, . . . . - '
Jvely fact, rlems both nany , and mtter ional iased drop 'E, at ry
a t Low-frequency,BC-coupled amplifier Figure 8.5 Common-emitter amplifiers

, . : "
(b) High-frequency transformer-coupled tuned amplifi er

, ',nge will obviously be equal to the supply voltage, less the very small drop due rhe dc resistanceof the transformer primary. The gain is still equal to the ratio : .-ollector-circuit impedance to emitter-circuit impedance, but the collector-':-'uit impedanceis now very much a function of frequency. to Once again, it is necessary recall somebasic electrical theory.The collector -,aed circuit will be parallel resonantat a frequency that is given approximately by ^ t" . :lir .f, : L1 : C1 : 1 2nf Lp,

(2.2)

(2.r)

resonantfrequency in hertz primary inductancein henrys primary capacitancein farads

xterroxi-

Equation (2.2) is reasonablyaccurateas long as the loaded quality factor (Q) : the tuned circuit is greater than about ten. At resonance,the impedanceof the :-ned circuit is resistive and its magnitude is at a maximum. On either side of ::ionance, the impedanceis lower in magnitude and is reactive: inductive below :i>onance and capacitive above. The higher the impedance, the higher the gain. lherefore, the gain is greatestat resonance. Once the Q has been found, the bandwidth at 3 dB down from the resonant:::quency gain is given very simply by

B -*{t
;here nce, rtor bandwidth : resonantfrequency /4 0 : loaded Q at resonance B:

(2.3)

fhapter

2 | Radio-FrequenEg

flrcuits

The gain goes down away from resonancebecausethe impedanceof the resonant circuit is greatest at resonanceand drops rapidly at higher and lower frequencies. The loaded Q depends on the load resistance,the transformer turns ratio, and the coupling between the transformer windings. The more heavily loaded the amplifier, the lower its Q and the wider its bandwidth. An example may clarify this analysis.

EXAMPLE

Z.f

An RF amplifier has the circuit shown in Figure 2.6. Find

(a) the operating fiequency O) the bandwidth Assume the loaded Q of the transformer primary is 15.
r Vcc=15 v

: i 50pF
. ' D

.-, 300 ct

{56kO

t)i'-

j-

-1 -

*-*tt'/
t

! r", .a'10ko

i
l

0.001pF

.-"'
Figure 8.5

+7multistvr
5oluUon (a) We can find the resonantfrequency from Equation (2.2):

, t" -

1 2n\/rrc,

x urt/(zs x 10-5(50 ro-12)

: 4.50MHz

I
le resk tie'turns eavily lb) The bandwidth can be found from Equation (2.3).

Sectlon

?.3 a Radio-Frequenql

Ampllflers

n:L

4.5IedIIz 15 :300kHz
ind Tlrereare some problems with the simple circuit in Figure 2.6.T\e transistor outrt impedanceis connectedacrossthe tuned circuit. For a bipolar transistor oper{ing at radio frequencies, this impedance may be only a few thousand ohms. [r-xrsequently, the transistor is usually connected across only part of the coil, as iho*n in Figure 2.7.T\e transformer p.i-ary acts as an autotransformeqincreasrg the effective impedance of the transistor and increasing the Q of the circuit. \trte that while the transistor is connectedacrossonly part of L1, the capacitor c, rnes the whole inductor. Remember that the top end of the coil is effectively :onnected to ground for ac through the filter capacity of the power supply. A de:rrrpling network consisting of Rp and C12 also been added. has The capacitancebetween the collector and baseofthe transistor is also likely rrr csuStrouble. By feeding back some of the output signal to the input, it re.fi-re'es gain of the circuit at high frequencies. In fact, for the common-emitter the :rrcuit shown in Figure 2.7, the effect of this capacitance,often called the Miller cfect is the same as if a much larger capacitance had been connected across the nput. The influence of the Miller effect can be reduced somewhat by transformer:trupling the input as well as the output and tuning the secondary of the input
vcc Figure P,7 hactical corlmonemitter amplifier with tapped primary

>Ro

'
i -

Crr<
i _

iR",
u;+ -*----- ir------------+---,-----! CC,
.

t./ r
|
\

j z,jf,iIl J _ - , . . 43 l i n , ii i 3 l r 7 14 - "

,-tsi-(,-

. :_,: u f i
v ,

n", ,;:
--t__

'
i--i
-\

4 n
-

--

i
:

-' -

>

1R, ;ts c" l

= l

5?

fhapter

2 a Radao-Frequencu

flrcuits

Figure ?.8 Narrowband RF amplifier with tuned input and output

V""
i

*
RD

Cp

"".;

:C, i------l i
I : . 1 r

R",

,t

,,i3,S

i : ,I + : i ti t i_____-i_l

+ ...'..--

Rn,

CB

RE

transformer.The Miller capacitancethen becomespart of the capacitancenecessary to tune the input circuit to resonance.Such a circuit is shown in Figure 2.8. The input is tapped down on the transformer secondaryto reduce the loading effect ofthe transistor and its biasing circuit on the Q ofthe tuned circuit. C1 tunes the input tuned circuit, and in this circ:uitC2 is the output tuning capacitor. The practice of tuning both the input and output of an amplifier can lead to instability. This instability can be corrected by neutralization, which will be discussedlater in this chapter. Another way to avoid the problem of the Miller effect is to use a commonbaseamplifier, as shown in Figure 2.9. This circuit showstransformer coupling at both input and output. Common-baseamplifiers are rare in low-frequency applications but quite common at radio frequencies.The capacitance between collector and base appearsacrossthe output tuned circuit, where it simply reducesthe exFigure P.9 amplifier Common-baseRF
Vcc

BD

CD

Lr
-*-T*l't l v1

'

Its r F l
i+1

ti ,l'r
-_*L l l

E- !l . ; i t - _-.-L_l

--*

j i--'-' L " i - -i

"rot vEE

ti c" ....

Sectlsr 2-3 a Radio-Freqrencg :lrnal capacitanceneededto achieveresonance.This will geatly extend the use. "rl frequency range of the transistor. In Figure 2.9,the input tuned circuit consists of L1 and C1, and the output ::rnedcircuit is 12 and C2. Emitter bias is shown, but voltage-divider bias is also .,..ritable does not require a negative supply voltage. and The loading effect of the transistor impedances on tuned circuits can be :reatly reduced by using field-effect, rather than bipolar, transistors.Field-effect :usistors (FETs) are famous for their very high input impedance,of course,but -:eir output impedanceis also much higher than for bipolar transistors.

Ampliflers

\\'ideband Amplifiers \t'rt all amplifiers used in communications have tightly restricted bandwidth. For -il'stance, amplifiers usedfor the basebandpart of the system are usually wlZethe :--nd (also called broadband). You are no doubt already familiar with audio i-rnplifiers.Those used for basebandvideo are similar, though their bandwidth is .;rger (about 4.2MHz for broadcasttelevision signals). In this section,we will consider amplifiers designedfor higher-frequencyop:ration, where the responseis required to extend over a relatively wide range of requencies. For instance,an amplifier for a cable television system might be reruired to amplify frequenciesfrom 50 MHz to approximately 500 MHz. This is a :eirly difficult requirement, especially if the system is expectedto be flat, that is, ,(r haveequal gain acrossthe entire bandwidth. (An actual cable television amplirer would be designed for more gain at higher frequencies to compensatefor :Teaterloss in the cable.) In most cases,it is not required that a broadbandRF amplifier have response Jr-rr*'n dc or even to audio frequencies. In fact, this is likely to be undesirable. to \t'e would not want our hypothetical cable television amplifier to amplify 60 Hz :lum. for example. In fact, we would prefer it to ignore both AM radio broadcasts r about I MHz and CB radio transmitters at 27 MHz as well. Broadband ampli:ren. then, generally incorporate some form of filtering so that the frequency :.\ponse, while broad, is restricted to the range ofinterest. Wideband RF amplifiers, like their narrowband counterparts, typically use -:,rnsformercoupling. This technique was once popular for audio amplifiers as :ell. but the size, the weight, and especially the high cost of audio transform:n has led to their virtual elimination from audio circuitry. For RF amplifiers :rl retain some advantages, however. Transformers are very convenient for -mpedance matching, which is likely to be more important in RF than in audio de:rgns.The isolation betweeninput and output is useful in helping to keep spurious .rrrnals at frequencies greatly different from the desired signal frequency from ropagating through the system.Transformer coupling also makes it easy to cou:le balanced inputs or loads to the amplifier. Balanced lines have equal impedoc-e from each conductor to ground. They are often used with antennas; ordinary :elevision twin-lead is an example of a balancedline that should be transformer:oupled to the first stagein a television receiver. The transformers used in wideband amplifiers need careful design to avoid .elf-resonanceand to maintain relatively constantgain acrossthe frequency range rf interest.In particular, the reactanceof the windings must be large compared to tre impedances that are connected to them. Toroidal transformers with ferrite :L)resare common. In the case of an amplifier that must have a very wide band; idth, the ferrite, having much higher permeability than ar, allows for sufficient :eactance at the low end of the frequency range. At higher frequencies, the ferrite

tecese 2.8. rg efTUNES

ad to r dismonng at rppliector e ex-

Chapter

2 o Radlo-Fret|r.lentu

Eircults

is likely to be less effective, but the reactance of the winding will be sufficient, since reactance increases with frequency. Figure 2.10 is the circuit of a broadbandRF amplifier using a bipolar transistor. Notice that it is the sameas Figure 2.5(b),except that Cr, the tuning capacitor, is gone and the input as well as the output is transformer-coupled' coupling beIt is also possible to use conventional resistance-capacitance tween broadband amplifier stages. Such circuits can be analyzed in the same way as their low-frequency counterparts. Whatever design is used, if an amplifier operates over a sufficiently wide frequency range, it will tend to have higher gain at the low end of its range than at the high end because the gain of the transistor falls off with increasing frequency. Negative feedback is often used to overcome this, with the amount of feedback decreasing as the frequency increases. Another method uses a high-pass filter at the input, so that as the frequency goes down, the amount of input signal reaching the amplifier is reduced.

Amplifier Classes
Amplifiers are classified according to the portion of the input cycle during which the active device conducts current. This is called the conduction angle and is expressed in degrees.The active device may be a bipolar transistor, FET or tube; for now we will assurne it is a bipolar transistor. For a push-pull amplifier, the conduction angle for one of the two amplifying devices is usedSingle-ended audio amplifiers are generally operated in Class A, where the transistor conducts current at all times for a conduction angle of 360'' This is the only way to achieve linear operation, in which the output is a reasonably faithful copy of the input, except for amplitude. The small-signal RF amplifiers described earlier in this chapter were also biased for Class A operationPush-pull amplifiers can be linear if at least one of the two transistors is conducting at all times. In a Class B amplifier, each transistor is biased at cutoff, so that each conducts for 180" of the input cycle. This results in greater effrciency

Figure P.lo amplifier

Broadband RF

-F-

tri u" .**

l+l

2 Ru,
l

_r-uilH
i

t--.-:-j

-1-

SeEtion 2-3 o Radlo-Frequencll

Ampllflers

;rent, rnsiscitor, s be)way e frean at ency. lback ter at :hing

rhich s ex:; for cone the s the thful dbed con[, so ency

rhan for Class A amplifiers, but the distortion is larger due to nonlinearity of the ransistors near cutoff. Most audio power amplifiers use Class AB, which is a compromise between rllass A and Class B, in terms of both distortion and efficiency. Each transistor :r-rnductsfor slightly more than 180" of the input cycle, thus reducing the -.-rossover" distortion that occurs as the input signal passesthrough zero, where rrrh transistors are near cutoff. Figure 2.11 shows a simple Class B amplifier for RF operation. Like the ClassA amplifiers shown earlier, this amplifier usestransformer coupling, which, rhile rare in modern audio amplifiers, is still the most cornmon way to design "implifiers for high-frequency operation. The example shown is broadband, but jx circuit couldjust as easily be tuned. Both transistorsare biased near cutoff by the voltage drop acrossD1. When te top end of the input transformer secondarygoes positive, Q1 turns on and Q2 -: c-utoff. The reversehappensin the other half of the input cycle; thus each tranilstor conducts for half the cycle. Audio signals, as well as some radio-frequency signals,have complex wave:r-rnns and require linear amplification to avoid distortion. Suppose,however,that re have to amplify an RF signal that we know is a sine wave. It might be possible :rr do this with an amplifier that created a great deal of distortion, provided that we r-xl some means to remove that distortion after amplification and restore the signal :o is original sinusoidal shape.This extra trouble would be worthwhile if the re'ulting amplifier were more efncient than either ClassA or Class B amplification. Such an amplifier class exists: it is called Class C. In it the active device conjucs for less than 180" of the input cycle. The amplifier can be single-endedor rrsh-pull; either way, it is apparentthat the output for a sinusoidal input will re'emble a series of pulses more than it does the original signal. These current rulses can be converted back into sine waves by an output tuned circuit. The re'nlting gain in efficiency can be quite worthwhile, as can be seenin Table 2.1. ClassC amplifiers achievetheir maximum efficiency when the amplifying de'.rcealmostsaturates peaksof the input cycle. It is, of course,cut offfor most of the at :1cle. This type of operationminimizes power dissipationin the transistor,which is zerowhen the transistoris cut off, low when it is saturated,and much higher when rt is in the normal, linear operating range. Since Class C amplifiers are biasedbe)ond cutoff, they would be expectedto have zero power dissipation with no input. This is indeed the case,provided the bias is independentofthe input signal. There remains the problem of distortion. Figure 2.I2(a), a simplified circuit tor a Class C amplifier, showshow the distortion is kept to a reasonablelevel. The Frocesscan be explained using either the time or the frequency domain. In the ume domain, the output tuned circuit is excited once per cycle by a pulse of coliector current. This keeps oscillations going in the resonant circuit. They are dampedoscillations, of course, but the Q of the circuit is high enough to ensure 6at the amount of damping that takesplace in one cycle is negligible, and the outi\BLE Cllss Conduction angle Maximum efficiency Likely practical effi ciency e.l Review of amplifier classes A

Flgure

?.ll

Class B RF amplifier

3600 50Vo 25Vo

180" 78.57o 60Vo

<180' lNVo 757o

5E

fhapter

2 | Radio-FrequenEu

fircuits

Figure

2,1?

Class C amplifier

c1

t .t'

:.

* _-.*i
f,
i \

+
D ! ;

l -l I'n

(a) Circuit

: .

(i) Input voltage

(ii) Base voltage

(iii) Collector current (b) Waveforms

(iv) Collector voltage

section 2-3 o Radlo-Freqrenqg Ampllflers F F a reasonably accurate sine wave. The frequency-domain explanation is even qln the resonant circuit constitutes a bandpass filter that passes the fundafrequency and attenuates harmonics and other spurious signals. In many Mal cz*s- especially when the amplifier is the final stage of a transmitter, additional Sierrng after the amplifier further reduces the output of harmonics. From the foregoing description, it might seem that Class C amplifiers would Lmr to be narrowband. Many of them are, but this is not a requirement as long as fu urput is connectedto a low-pass filter that will attenuateall the harmonics dnrr 56 generated. [.et us look at the circuit in FigweZ.l2(a) alittle more closely.We begin with the trx 'ircuit, whichis unconventional.Farfrombeing biasedin the middle of the lincr operatingrange,asfor ClassA, a ClassC amplifier must be biasedbeyondcutoff. F:r a bipolartransistor,that can meanno basebias at all, sinceabase-to-emittervolt.u of about0.7 V is neededfor conduction.With no signal at its input, the transistor rll baveno basecrurent and no collector current and will dissipateno power. lf a signal with a peak voltage of at least 0.7 V is applied to the input of the rrnfrlifier, the transistor will turn on during positive peaks. It might seem at first girnr.:ethat an input signal larger than this would turn on the transistor for more mJ more of the cycle, so that with large signals the amplifier would approach C;ss B operation, but this is not the case.The base-emitterjunction rectifies the ulrt signal, charging C6to a negative dc level that increaseswith the amplitude :r de input signal. This self-biasing circuit will allow the amplifier to operatein Crss C with a fairly large range of input signals. The collector circuit can be considerednext. At the peak of each input cycle, tue transistor turns on almost completely. This effectively connects the bottom :ol of the tuned circuit to ground. Current flows through the coil L1. When the input voltage decreasesa little, the transistor turns off For the rest x de cycle, the transistor represents an open circuit between the lower end of the umed circuit and ground. The current will continue to flow in the coil, decreasing grdually until the stored energy in the inductor has been transferred to capacitor f . s'hich becomes charged. The process then reverses, and oscillation takes :i.r.r. Energy is lost in the resistance of the inductor and capacitor, and of course trryy is transferred to the load, so the amplitude of the oscillations would gradotlll be reducedto zero, except for one thing: once each cycle, the transistorturns :n- anothercurrent pulse is injected; becauseof this, oscillations continue indefintely at the samelevel. Figure 2.12(b) shows some of the waveforms associatedwith the Class C rrylifier. Part (i) shows the input signal, and part (ii) shows the actual signal ap3hed to the base.Note the bias level that is generatedby the signal itself. Part (iii) $ows the pulses of collector current, and part (iv), the collector voltage. It should rat be surprising that this reachesa peak of almost 2Vg6.T"hepeak voltage across te inductor must be nearly V6g, since, when the transistor is conducting, the top eod of the coil is connectedto Vcc and the bottom end is almost at ground poten:al ("almost" because,even when saturated,there will be a small voltage across te transistor).At the other peak of the cycle, the inductor voltage will have the 'ame magnitude but opposite polarity and will add to Vrc to make the peak collector voltage nearly equal to 2Vg6. This description implies that the output tuned circuit must be tuned fairly ;losely to the operating frequency of the amplifier, and that is indeed the case. Srncethe transistor must swing between cutoff and something close to saturation for Class C operation, it is also implicit in the design that this amplifier will be

Chapter

2 | Radlo-Frequenru

EarEuits

nonlinear; that is, doubling the amplitude of the input signal will not double the output. These two limitations restrict the use of class c amplifiers to RF signals, and in fact, only some RF applications can make use of this circuit. Class c amplification can be used with either field-effect or bipolar transistors. It is also quite commonly used with vacuum tubes in large transmitters.

Neutralization
Device and stray capacitancetend to reduce gain and cause instability as frequency increases.care must be taken to separate inputs and outputs to avoid feedback, but often a transistor or tube will itself introduce sufficient feedback to causeoscillations to take place. Sometimesthis type of feedbackcan be cancelled by a processcalled neutralization. Neutralization is accomplishedby deliberately feeding back a portion of the output signal to the input in such a way that it has the same amplitude as the unwanted feedback but the opposite phase. since the device capacitancesvary from component to component,careful adjustmentis necessary. Figure 2. 13 shows one type of neutralization. The basic circuit is that of Figure 2.8, a narrowband RF amplifier with tuned input and output. As mentioned earlier, such an amplifier has a tendency to oscillate because of the feedback through the base-collectorcapacitanceof the ffansistor. The method used in Figure 2.13 to provide feedback for neutralization is to rearrangethe primary of the output transformer so that the center is connectedto vgs. The power supply is, of course, at ground potential for ac (especially when the decoupling network consisting ofRp and cp is used). The transistor is still connectedacrossonly part of the output transformer, to reduce its loading effect as before, and there is no difference as far as the operation ofthe circuit is concerned. The reason for rearranging the transformer connections is that now the top and bottom ends of the winding have oppositepolarities with respectto ground for the
Figure ?.t3 amplifier Neutralized RF ucc

i
i 'jR"

'i

?-.'--r C' '---. I

t
I

Fc,
:l-

-t---i

l. -7---*---i

,,lg i
l

#C"
I

l i i I '

-+,,---..---- -'

ii A -l------1 t J t t i i S i i - + -kitFi,, ; 5 i ; F t ' " lgi l--i --t-'__'__,i B

SeEtlon 2.3 I Radlo-FrequenrlJ Ampllflers u .tgnal. The voltagesmay or may not be equal in magnitude: there is no require:cnt that the transformerbe tappedin the centeras shown.The internal capacitance ---,- showndotted, feedsa signal from the top end of the coil, pointA, to the transisr:r base. small variable capacitor,C1g, adjustedto feed a signal of equal magniA is -::e but oppositepolarity from the bottom of the coil, point B, to the base.When Cry : :,,ljustedconectly, the signalswill cancel and the amplifier will be stable.

59

ble the ignals, ransis)-

u freI feedack to rcelled of the as the

Frequency Multipliers
l":e output circuit of a Class C amplifier can be tuned to a harmonic of the input .--:ral. The amplifier will still operate,though with reducedefficiency. By provid-: energyto the circuit during alternatecycles the pulsesofcollector current will .-.ain oscillations in an output tuned circuit that is tuned to twice the input ::quency. Another way to look at the process is to recognize that the collectorj--:r-entpulse in a Class C amplifier is rich in harmonics. Any one of these har:rrorcs can be chosenas the output by tuning the output bandpassfilter to the ap:r:,priate frequency.This harmonic operation can be very useful when a signal is :q:quiredat a higher frequency than can be conveniently generated.Figne 2.14 {.}r s input, output, and collector-current waveforms for a frequency doubler.
Figure ?.14 Doubler waveforms

s vary
rf Fig:ioned dback n Figof the is, of i conart of o difp and or the

Input Sigrral

- - l I

I . i l, . . " o

Output Signal

CollectorCurrent

Ehapter

Z a Radlo-Frequencg

Elrcults

FrequencyMultipliers In 9 MHz xB x2 x3 L62MHz

Figure e.t5 tion

Frequency multiplica-

Because frequency multipliers operate at lower efficiencies than straightthrough amplifiers, they are used at low power levels. Most multipliers operateat the second or third harmonic of the input frequency and are known as doublers or triplers, respectively. They are more efficient than multipliers operating at hi gher-order harmonics. Multipliers can be used in cascade greatermultiplication is required. Figure if 2.15 shows one way to get a multiplication of 18 times. This is a cornmon iurangementinVFIFtransmitters, thoughitis notusedinnewdesigns muchasitonce was. as

SEETION ?.3 QUESTION

REVIEW

Sketcha graphshowinghow the outputvoltagevarieswith the input voltagefor Class andClassC amplifiers useit to showwhy Class amplifiers deA and A are scribedas "linear" and ClassC amplifiersarecalled "nonlinear."

2.4 Radio-Frequency Oscillators

RF oscillators do not differ in principle from those used at lower frequencies,but the practical circuits are quite different. While low-frequency oscillators usually use RC circuits in the frequency-determiningsection, tC circuits are more common at radio frequencies.In addition, many RF oscillators are crystal controlled. Any amplifier can be made to oscillate if a portion of the output is fed back to the input in such a way that the following criteria, known as the Barkhqusen criteria, are satisfied: 1. The gain around the loop must be equal to one. (If it is initially greater than one, it will become equal to one when oscillations start, due to some process such as transistor saturation; otherwise the output voltage would continue to increasewithout limit.) 2. The phaseshift around the loop must total either 0o or some integer multiple of 360o at the operating frequency (and not at other frequencies). Taken together, these statements simply mean that at the operating frequency, an input signal will be amplified then fed back in phaseand with suffrcienramplitude that it will maintain its value at the output without any further input. The initial signal needed to start the process can be noise or a transient caused by switching on the power to the oscillator circuit. Figure 2.16 shows a "generic" oscillator consisting of an amplifier with gain A and a feedback network with "gain" B. B will usually be less than 1, and only part of the output will be fed back to the input. B is often called,thefeedbackfraction. Accordrngto the Barkhausencriteria, we will have:

{L| --'

r-* output

iEi
Figure ?,15 Generalizedoscillator

AB: I

(2.4)

when oscillations are in progress.The phaseshift around the loop will be zero or some multiple of 360'. This means that if the amplifier is an inverting amplifier, B must also invert the signal phase. The foregoing must be true at only one frequency, so B has to have some form of frequency dependence. Either its phaseshift or its amplitude responseor both must vary with frequency. There are many types of networks that meet this requirement, but one of the simplest (and the one most commonly used in radiofrequency oscillators) is the LC resonant circuit. Both series- and parallel-reso-

section Z-zl a Radio-Frequencg Osclllators r.rnt circuits have amplitude and phaseresponsesthat are functions of frequency, rlrl both have applications in oscillators.

aight?te at rblers Lngat ;igure angeI WaS.

LC Oscillators
nrrsnow time to look at somepractical oscillatorconfigurations.We will begin with whose frequency is controlled by a resonant circuit using inductance rn:r-illators rnJ capacitance. Crystal-controlled oscillators will be describedonpage72. All the circuits shown in this section have been in use for many years and :;rve been irnplemented with vacuum tubes, bipolar and field-effect transistors, rnJ integratedcircuits. In each case,we will begin by showing the amplifier as a gain block. All of these oscillator types can be implemented with either ;a-rnple rrerting or noninverting amplifiers and will be shown both ways. Hertley Oscillator This oscillator type can be recognizedby its use of a tapped *luctor, part of a resonantcircuit, to provide feedback. Figure 2.17(a) shows a Hartley oscillator using a noninverting amplifier and Frgure 2.17(b) shows the same oscillator using an inverting amplifier. In either ::,r-<e. resonant frequency is that of the tuned circuit, including the whole inthe i.r.-tor, that is , t'nhere f : L : C : 1 2r\/LC

s. but ually comrlled. ak to t crithan

]CESS

(2.s)

frequencyofoscillationinhertz inductanceof the whole coil in henrys capacitanceacross the coil in farads

ue to rle of ency, npliThe dbv galn only frac(a) Using noninverting amplifier

(2.4) ro or ifier, ;ome seor this rdiotso-

:
(b) Using inverting amplifier Figrre ".17 Hartley oscillators
(c) Using inverting amplifier withA, < 1 (butA; >1)

Ehapter

Z a Radio-Frequenru

f.lrEuits

The feedback fraction is easy to find if the assumption is made that the inductor is a single coil with unity coupling. It is then given simply by the turns ratio (input turns to output turns). Thus there are two different equations,depending on whether the circuit usesan inverting or noninverting amplifier. For the noninverting circuit of Figure 2.17(a), B: N, AL + N, Q'6)

This circuit can be implemented with transistorsconnectedin the common-base or common-gateconfiguration. In the oscillator using the inverting amplifier, as in Figure 2.ll(b), -N, B : N2

(2.7)

The negative sign indicates that the feedback has the opposite polarity to that of the output signal. This is required for an oscillator using an inverting amplifier, such as would be provided by a transistor connected for common-emitter or common-sourceoperation. Both of these circuits assume that the amplifier has a voltage gain greater than one. It might seemthat this is a necessary condition for oscillation, but this is not the case.AB can be equal to one ifA is less than one, provided that B is greater than one. The real amplifier requirementfor oscillation is power gain, not voltage gain. The tapped coil of the Hartley oscillator can be used as an autotransformer to give a voltage feedback fraction greater than one. of course,there will be a requirement for current gain from the amplifier in this case,a requirement that can be filled by using a common-collector or common-drain amplifier. Figure 2J7(c) demonstrates this configuration, in which the operating frequency remains as before, but the feedback fraction is now

B :

N , + N ,' Nr

(2.8)

EXAMPLE ?.E

Two practical Hartley oscillators are shown in Figure 2.18. The one in Figure 2.18(a) usesajunction field-effect transistor (JFET) in the common-source configuration, and in Figure 2.18{b) usesthe sametransistorconnectedas a common-drain amplifier (source follower). Find the operating frequency and the required gain for the amplifier part of each oscillator. 5olfiion (a) The operating frequency is given by Equation (2.5):
f ro : I

zn\/rc

zrrt/toxlol(looxlo-3
: 5.03MHz

Section

?.4

a Radio-Freqrelrra

llsclllators

the inI turns epend-

(2.6) n-base

100oi ?--_: 1 ;rco

i i

t nrt

Vao= 15 V

Flgure

3.18

.i 0.0rpF 100pF :

*multisrtvt

100 T\rrns
N2

Zr= 10 pH Q=50

(2.7) rhatof plifiea [er or

10 T\rrns N,

Voo= 15Y
I
j

reater thisis reater oltage ormer tareaI can . I 7(c) bea-s

100o>

C c =0 . 0 1 P F
l-..

?multislvt
MPF1O2

--l
t r S0Turnsi
!=

Cr= 1oopF .{ :
j

Rol 100ko< --

r2.8)

l i

s--...-.---

2o r\'ns X * IrO:
l

--1,

..

(b)
t

rd The feedbackfraction is given by Equation(2.7): B: -N, -------'

N2

:-

-10

100 : -0.1

Ehapter

Z a Radio-Freq.rencu

Elrcuits

To determine the minimum amplifier gain for oscillation, begin with Equation (2.4):

- 1 I B -0.1 : -1{)

As usual,the negativesign denotes phaseinversion. a O) The operatingfrequencyis the sameasin part (a), but now the feedback fraction is given by Equation(2.8): B : iy', + N, ' N,

_80+20 20 - 5 This requiresan amplifier voltagegain of at least0.2. Sincethe actualgain will be only slightly lessthanone,the circuit will work.
colpitts Oscillator The colpitts oscillator uses a capacitive voltage divider instead of a tapped inductor to provide feedback. once again, the configuration of the feedbacknetwork dependson whether the amplifier is noninverting, as in Figure 2.19(a), or inverting, as in Figure 2.19(b). The operating frequency is determined by the inductor and the series combination of Cl and C2. fo:

(2.e\
2r-\/LC,

c2a

_t______{Uo
I I
i l )

,--,+
cr --iI !

,n,'
i

4: :

FI

_i

(a) Using noninverting Figure ?.19 amplffier

Colpitts oscillators

Section

?.4

O Radlo-Frequerrcg

Oscillatss

C,C, r r - - --------:----:Cr+C,

(2.10)

The feedback fraction is given by the ratio of reactancesbetween output and -rfrtt circuits. This is, ofcourse, the reciprocal ofthe ratio ofcapacitances,because

xc:

znfc

(2.rr)

From this it follows that the ratio of the reactancesof two capacitors C1 and i;is

xc, _ 2rfCt
Xc,

I ZrfC2 _ 2rfCz 2nfC1 _c2 c1

(2.r2)

in

For the noninverting version of the oscillator, shown in Figure 2.19(a), the ,rtput is acrossthe seriescombination of C1 and Cz, which correspondsto C7 in Equation (2.9), and the input is the voltage across C1. It is easy to see that the :eedback fraction is
XC, B :
AC,

ider nof Fignbi-

:C, cl CtCz

(2.r3)

z.e)

:ct+c2

cr

- C , cr+c2
The feedback fraction is even easier to determine for the inverting circuit of Frgure2.19(b). Since the output is applied acrossC2 and the input is taken across C,. the feedbackfraction is

B:
-

xc, X''''..

(2.r4\

_Cz c1

.\s before, the negativesign indicatesthat the feedback signal is 180" out ofphase n'ith the output. Since changing either Cl or C2to tune the oscillator will changethe feedback traction, it is quite cornmon to use a variable inductor for tuning instead.

Eh4ter

Z a Radlo-Freqrcntu

Clrcuits

Like the Hartley, the Colpitts oscillator can be configured for an amplifier with power gain but no voltage gain. The derivation of the circuit and the feedback fraction is left as an exercise for the reader.

Deterrninethe feedbackfractionsandoperatingfrequenEXAMPLE 23 ciesfor the oscillatorswhosecircrritsareshownin Figure 2-20.

vcc

(a) Co-mon-emitter Colpitts oscillator

Vcc

:'-

-'

] roopn
ij

t ARFC

100 pF
'' C t'l c 'uo

-+-

> R" C".:,i'-

;/

'--n uz

::- -tt*

1"nF :

1*' _

-"' o"|i

I -"*^ilooPFl

(b) Common-base Colpitts oscillator Flgure ?.?O

Figure use Our Colpitts oscillator examples bipolar transistors. Solrthn is connected circuit, while the transistor 2.20(a)usesa cofiunon-emitter use two exarnples invertingand in common-base Figure2.20{b).These both with voltagegain greaterthan noninvertingamplifiers,respectively, circuit is alsopossible. one.The useof a common-collector In Figure 2.21(a),C6,is a coupling capacitorthat preventsa dc short circuit from occurring betweencollector and basethrough the coil. The radio-@uency chokeRFC takesthe place of a collector resistorand keeps the ac at the collector from being sho*-circuited by the power supply.A of collector resistcrcan be use4 but the choke,because its lower dc the increases output voltageand improvesthe efficiency of the resistance,

SeEtion Z-4 a Radlo-Frequencg Oscillators

67

nplifier refeed-

.ircuit. R6, and R6,?re bias resistors, of course. This leaves the frequency of fre oscillator to be determined by Lt Ct and C2. The effective capacitance futr determining the frequency of operation is given by Equation (2.I0):

"'- c, + c" rox roo : -nF

C,C,

l0 + 100r' : 9.09pF

Tbeoperatingfrequencycan be found from Equation(2.9):

f :

2"-t/4
2tt

(1 x10-1(9.09x10-12)

: 52.8MHz Tbe feedback fraction is given approximately Equatioa(2.14). by

fi:
Co ----:

cr

10 100 : -0.1 For the common-base circuit, the operatingfrequencywill be the samebut tu feedback fraction will be different.Of course,the sign will be positive the hecause amplifier is noninverting,but the magnitudeis also slightly different.From Equation(2.13),

,:#o
10 100+ 10 : 0.0909

tps

Clepp Oscillator The Clapp oscillator is a variation of the Colpitts circuit, de*-:ned to swamp device capacitancesfor greater stability. In the oscillators of 7 4ne 2.21, the frequency of oscillation is determinedby the inductor and the se:,'ssc'ombinationof C1, C2, and C3.In practice, the total capacitanceis determined rrnr--rst entirely by C:, which is chosen to be much smaller than either Cy or C2. lle total effective capacitanceof the three capacitorsin seriesis given by
C,T

l Ct

1 I T C2 c3

(2.rs)

-->
(a) Noninverting Figure 2.Pl

fh+ter

Z a Radio-Fre(ilrenrg

fircuats

-*--_?____*u,
I
j

-d

- :ur q

\
i
i I

#c"
I

Clapp oscillators

After finding C7 from Equation (2.15), the operating frequency can easily be found from Equation (2.9), as for the Colpitts oscillator. The feedback fraction is found in the same way as for the Colpitts oscillator. One circuit will serveto illustrate the Clapp oscillator, since it is so similar to the Colpitts.

EXAMPLE e.zf Calculate the feedback fraction and oscillating frequency of the circuit inFigwe2.22.

i C , -l 100 pF

1"""
I I I
!:
Flgure E.??

t- C, -. 1000pF :
I

Sohrtisr Figure 2.22is the sameasFigure 2.2Wa)exceptfor the addition of C3in serieswith the coil and the increase the valuesof C1and Cz.My in transistorcapacitances appearacrossC1 aaird wherethey will have will C2, little effect on the frequency.

Sectlon ?-4 t Radlo-Frequencl; Osclllators

Before calculating the oscillating frequency, it is necessary to calculate the effective total capacitance from Equation (2.15):
CT

-1+ - +1c1 c2

1 c3 pF 1 10

1 1 100 1000 : g.01pF

- T

ily be llator. ilar to

Yote the relatively small effect of Cl andC2. Now the operatingfrequencycanbe found in the usualway from Equation 12.9).
Jo

{ -

2r\/rcr

z"r@
:53.O2MtIz
ncy \ aractor-Thned Oscillators

The frequency of an LC oscillator can be changedby varying, or tuning, either the rductive or the capacitive element in a tuned circuit. Inductors are typically :rned by moving a ferrite core into or out of the coil; this is known as slug tuning. \-ariable capacitors usually have two sets of plates that can be interleaved to a jreater or lesserextent. Mechanical tuning tends to be awkward. The components are bulky, expen-r"rle. subject to accidentaldetuning, for instance,in the presenceof vibration. and \ariable capacitors and inductors are mechanical devices that have to be moved :ir1'sically.This makesremote or automatic frequency control rather cumbersome. Varactors are a more convenient substitute for variable capacitors in many :1rylications.Essentially,a varactoris a reverse-biased silicon diode.As the reverse ""trltage increases,so does the width of the diode's depletion layer. As a result, the -unction capacitancedecreases. this junction If capacitanceis made part of a resorut circuit, that circuit can be tuned simply by varying the dc voltage on the var.u-tor.This can be done in many ways and is well adapted to remote or automatic ;ontrol. The resulting circuit is often called a voltage-controlledoscillator (VCO). It is, ofcourse, necessary separate dc control voltage from the ac signal to the '.oluges. This is quite straightforward: Figure 2.23 shows one way to do it. The roninverting Clapp oscillator of Figure 2.21(a) has been adaptedfor use as a VCO :1 using a varactor for C3. Resistor R prevents the RF in the circuit from being ;bort-circuited by the circuit that provides the tuning voltage, and the extra --apacitor, keepsthe dc control voltage out of the rest of the circuit. Cais made Ca, nuch larger than C3 so that its reactancewill be negligible and C3 will still conml the operating frequency.

Ehapter

? a Radlo-Frequencg

Elrcults

Figure e,a3 oscillator

Varactor-tuned

't -*--

i'\ , A-;rl-"-

. ---'-{

uo

':i. cz 'l

.=-, C,
'

F Zt
.V: C4 R
}----""';""'r------L rr

i-

, varac[or

vB

Tuning -.:
Vottage

The variation of capacitancewith voltage is not linear for a varactor. It is given approximately by C : where co

\h+zv

(2.16)

C : capacitanceat reversevoltage V with no reversevoltage Co: capacitance

From this equation it can be seen that for relatively large reverse-biasvoltages, the capacitance is approximately inversely proportional to the square root of the applied voltage. A number of varactors, with maximum capacitancesvarying from a few picofarads to more than 100 pF are available. From Equation (2.23) it can be seen that the minimum capacitance will be limited by the breakdown voltage of the diode and, of course, by the tuning voltage available. In practice, a variation of about 5: 1 in capacitanceis quite practical. The magnitude of control voltage required for this can be found by letting C/C : 5 in Equation (2.16): C^

y1+ 2V

vl +2v:;

C^
L

t/-t + zv: s
| + 2V:25 2V:24 V: l2Y

of EXAMPLE U.5 A varactorhasa maximumcapacitance 80 pF andis usedin a tunedcircuit with a 100 p,Hinductor. (a) Find the resonant frequencywith no tuning voltageapplied. at for O) Find the tuning voltagenecessary the circuit to resonate doublethe part (a). frequencyfound in

Section

U-4 a Radio-Frecprencg

Osclllators

S.lrfign

t a) The maximum capacitance of the varactor occurs for zero bias voltage, so the capacitance in this case will be 80 pF. The resonant frequency can be calculated from Equation (2.2).

fo-

zrrt/rC

2tr ( 1 0 0 x l 0 - 6 x 8 0 x 1 0 - t 2 ) : 1.78MHz rr. It is r b ) From Equation(2.2) it is apparent that the resonant frequencyis in-

(2.16)

verselyproportionalto the square root of capacitance, that doubling so the frequencywill requle reducingthe capacitance a factor of four. by Alternativelywe can useEquation(2.2) dtrectly.
-

Jo

2rrY LC
| 4n'LC I

'ltages, 'oot of i varyluation by the Lilable. r mag=5in

i2t" . :

+n'f"t
4n212x 1.78x to6)2(1oox to-6)
= 2Ox 10-12 F :20pF

Now we can find the requiredtuning voltagefrom Equation(2.16\.

C: co

\/r + 2v
co C

\/--r+ 2v:

t+2v:(?)'

-' (7)' v2

\a,l

/soY-

2 :7.5Y

7?

fhapter

e a Radlo-Frequencg

fircuits

Crystal-Controlled Oscillators
The frequency stability of any LC oscillator depends on that of its resonant circuit, including any stray or device reactances that may be present.Even with careful design, LC oscillators are subject to frequency change from such diverse sources as voltage variations, changes in load impedance, temperature changes, and mechanical vibration. All these problems can be reduced, but only at great expense. Crystal oscillators achieve greater stability by using a small slab of quartz as a mechanical resonator, in place of an LC tuned circuit. Quartz is a piezoelectric material: deforming it mechanically causes the crystal to generate a voltage, and applying a voltage to the crystal causesit to deform. Like any rigid body, the crystal slab has a mechanical resonant frequency. If it is pulsed with voltage at that frequency, it will vibrate. From the outside, the crystal will have the appearance of an electrical resonantcircuit. Figure 2.24(a) is the schematic symbol for a crystal, and Figure 2.24b) shows the equivalent circuit. Note that a crystal does not actually have this circuit A crystal is just a slab of quartz with electrodes.This circuit simply models its behavior, as seenfrom outside. The only part ofthe equivalent circuit that resembles the actual device is C", the parallel capacitance, which actually is the capacitance of the holder and the crystal itself. The series inductor and capacitor are electrical analogs of the mechanical properties of the crystal. In electrical terms, the inductor in the equivalent circuit is very large (on the order of henrys), while both capacitors are very small (picofarads or less), with the parallel capacitance much larger than the series. The resistance is small, and the Q of crystals is very high (in the range from 104to 107).From the equivalent circuit, it is apparent that the crystal will actually have not one but two resonant frequencies: one series and one parallel. Figure 2.25 shows graphically how the reactance varies with frequency. The two resonances will vary by an amount that is on the order of l%oof the operating frequency, and when ordering a crystal, it is necessaryto know at which of these frequencies the circuit operates. For parallel resonance, it is also necessary to have a constant, known circuit capacitance across the crystal. It is also possible to operate the crystal at a harmonic, or overtone, of the fundamental frequency. This is necessary when frequencies above approximately 40MHz are required. Crystal oscillators offer great accuracy and stability at the price of fixedfrequency operation. The operating frequency can be accurate within a few parts per million (ppm) at the reference temperature at which the crystal is calibrated The stability over a typical range of operating temperatures varies with the design of the crystal and the oscillator. For simple oscillators without temperature compensation, the variation can be approximately 10 ppm. When greater stability is required, temperature compensation can be used This usually involves using thermistors to sense temperature variations, and ap plying a variable capacitance across the crystal using a varactor. More advanced schemes use a microprocessor and a look-up table that relates required varactor voltage to temperature. For the ultimate in stability, as required in demanding applications like test equipment and some transmitters, a crystal oven can be used. This device keeps the crystal at a temperature slightly grcater than the highest expected ambient temperature. A crystal oven can improve the temperature stability of a crystal oscillator by a factor of about 100 000.

i__--i i i
i

F
:
i

i
i I

u.q

/-j\

;i: C,
i
1 I

(-

_l

D <

I -T(a) Symbol for a quartz crystal

F--':----i
i

(b) Equivalent circuit for a crystal Crystals

Flgure

?.?,rl

s
. o

f
Figure ?,85 Variation of crystal reactance with frequency

SeEtion 2-4 a Radlo-Frequencg Osclllators The temperaturedependenceof the frequency of an oscillator can be given '.heequation :,^

mt crrh careJiverse ranges, [ $eat Brtz as lectric re, and

fr: f" + V"(T - T")

L iere f7 : f : t : operating frequency at temperature Z operating frequency at referencetemperatured" temperaturecoefficient per degree

(2.r7)

crysat that nrance 2.24(b) circuit. , its beembles citance rctrical (on the l. with rli, and ivalent fsonant ow the nt that al, it is nrallel cltance tr overabove 'fixedr parts brated. design e com: used. nd apvanced aractor ke test : keeps mbient crystal

BAMPLE 2.5 A portable radio transmitter has to operate at temperaruresfrom -5oC to 35"C If its signal is derived from a crystal oscillator rith a temperature coefficient of + 1 ppm/degree C and it transmits at exactly lJ6 MHz at20"C, find the transmitting frequency at the two extremes of the Jperating temperature range. .*ttttttttttttttttion Equation (2.17), the frequency at a temperature of 35oC From rrll be fr:

f*

: l46MHz + (146 MHzXI x 10-1G5- 20) : l46.AA2l9MHz

- T") fo + kf,(T

By similar reasoning,the operating frequency at the lower temperature rmit will be f^;,: 146MHz * (146 MHz)(l x 10-6X-5 - 20) : r45.99635MH2

ln other words, varying the temperature over a total range of 40"C caused Ssrtransmitter frequency to changeby (2.19 + 3.65) : 5.84 kHz. This rould not usually be acceptable in a practical transmitter, so some form of nequency compensationwould be needed.

Figve 2.26 shows two typical circuits for crystal oscillators. The Colpitts ::;uit in Figure 2.26(a) usesthe parallel-resonancemode of the crystal, with ca:icitors C1 and C2 providing feedback. R, and C, decouple the drain from the yr er 5upptt. The bias is generatedfrom the signal itself using the gate-leaktechru:ue previously described for FET oscillators. The circuit in Figure 2.26(b), r,:',n n as the Pierce circuit, uses the crystal in place of the inductor in a series-:rrnant circuit consisting of Cr, Cz, and the crystal. It operatesin the inductive :::ron ofthe reactance curve ofFigure2.25, closeto but slightly abovethe series -:!rnant frequency. The frequency of a crystal oscillator can be adjusted slightly by placing a ..:uble capacitancein seriesor in parallel with the crystal, dependingon the type :i;ircuit. An oscillator of this sort is known as a variable crystal oscillator ,fol. If the capacitor is actually a varactor, the frequency can be adjusted by a :ermistor for temperature compensation.

fhapter

Z a Radio-Frequencu

Elrcults

vDD

t*----r r
I

vDD

-r\ '-* Sarc ? i

*C"

: 9Ro

uo tt__-,,,_-,o

(b) Pierce Flgure ?.?6 Crystal oscillator circuits

SEETION A.4 REVIEW OUESTION

Compare lC oscillators and crystal oscillators. What are the advantagesand disadvantagesof each type?

2.5 Mixers
In communications, a mixer is a nonlinear circuit that combines two signals in such a way as to produce the sum and difference of the two input frequencies at the output. Sometimesthe original input frequencies,and perhapsother frequencies as well, are also present.Mixers provide a way of moving a signal, complete with any modulation that may be present, from one frequency to another. It is important to distinguish mixing from linear summing, which produces only the two input signal frequencies.Unfortunately the term mixer is often used "mixer" that regulatesthe in audio electronics to designatea linear summer.The sound in a recording studio is really a good example of a summer. Figure 2.27 illustrates the difference between mixing and summing. Any nonlinear device can operate as a mixer. Familiar examples include a semiconductordiode and a Class C amplifier. In general, a nonlinear device proby ducesa signal at its output that can be represented a power series,for example
vo

:Avi*nv?+cv?+
output voltage instantaneous input voltage instantaneous constants

(2.18)

where

vo :
v i -

A, B, C,

76

fhapter

Z o Radlo-Frequencu

Elrcults

and v": Av; -f Bv! : A(sin o1/ * sin l,r,zt) B(sin o1r * sin o2l)2 + * 28 sin tolt sin o2l The first two terms in Equation (2.21) are simply the input signals multiplied by A, which is a gain factor. The next two terms involve the squares of the input signals, which are signals at twice the input frequency (plus a dc component). This can be seen from the trigonometric identity

: A sin<n(t Asinco2l r ,int rrr i i itf .r, +

(2'2ll

sin'A:

t-

tcos24

This meansthat the third term in Equation (2.21) becomes B -, B cos2toll , and the fourth term is t B - B 2o2t tcos

The final term is the interesting one. We can use the trigonometric identity sinA sinB : to expand this term. 28 sin altsin ar2l : ? - a2)t - cos (<,r1 a2)tl * t"o, (a1

1 lcos(A - B ) - c o s ( A + B ) l ;

Q.n)

: B [cos (o1 - to2)r- cos (o1 * a)tl As predicted, there are output signals at the sum and difference of the two input frequencies, in addition to the input frequencies themselves and their second harmonics. In a practical application, either the sum or the difference frequency would be used, and the others would be removed by filtering.

EXAMPLE 2.7 Sine-wave signals with frequenciesof 10 MHz and l1 MHz are applied to a square-law mixer. What frequencies appear at the ouFut? Solrtbn I*t ft : 11 MHz ffid fz : 10 MHz. Then the output frequencies are as follows:

fi: llMllz fz: l0Nftlz Diode Mixers

2fi:22MHz 2fz:ZOMHz

fi + fz:ZlMHz fi - fz: 1 MHz

There are two senses which we can say that a diode is a nonlinear device. First, in consider an ideal diode, which has zero reverse current for any reverse voltage and zero forward voltage for any forward current. This voltage-current relationship is shown in Figure 2.28(a). There is a very obvious nonlinearity atzero voltage.

s:ectltrr

2.5 a Mixers

(2.21)

Itiplied e input onent).

v
,at Ideal diode Figrre ?.?A Diode curves

(b) Tlpical small-sigrral silicon diode

ntlty (2.22)

\ow consider a real diode with its forward voltage-currentcharacteristics. A -r-J ,-urvefor a typical silicon signal diode is shown in Figure 2.28(b).It is obviruslr not a straight line; that is, a real diode is nonlinear even when operated in tu tbrward-bias region. Of course, the larger the forward current, the more the .1i,"-rJe approacheslinearity. Dode mixers can use either type of nonlinearity. In the first type, one of the s5nrls is strong enough to switch the diode between the reverse-biased and forrrr-Gbiased states.The diode ring balancedmixer, to be describedlater, operates n fris way. There is also another possibility, where the diode operates with a gr.rfl forward bias, in the "knee" of the curve of Figure 2.28(b). The input signal x 'mall, so that the diode stays in this highly nonlinear region of its characteristic crre. Figure 2.29 shows a circuit for a mixer of this type. Such mixers are not rsoi very often becauseof their poor noise figure.
Figure ?.?9 Diode mixer

he two tlr secrce fre-

|l

Innsistor Mixers
{ transistors,both bipolar and field-effect types, have nonlinearities. In fact, mrh of the work in an introductory electronics course usually involves biasing trc:e devices in such a way as to reduce the nonlinearity to a minimum. The suceis of these efforts is then measured in terms of percent distortion. On the other hand, nonlinearities are desirable if a mixer is required. Either a nrrolar transistor or an FET can be operated in such a way that the input causes fre device to enter nonlinear regions. The FET is especially convenient for use as r nxer because parabolic shapeofits transconductance the curve gives it approxrrn;rrelya square-lawresponse.The bipolar transistor will produce more spurious

:. First, rgeand rshipis ge.

fhapter

Z a Radio-Frequencu

Elrcuits

Figure

?.3O

Transistormrxers

,l

V""

.l R^
-------l}-.----." ..{

a i rt---:I

r, la $]', *'", i"r{ tr

i Ft c
.r,R,
:i,

i----_1_.

t--:-- --

. i * Bd

,{ c,

h F i o n 1 ' * /6 1 i------r

bi f,-f,

ttt

I2 ln

(a) Bipolar mixer vDD

Rn. .

--a
vD -:

!
i

r:l_--.'-^-'.'.- - -- -----+.1 I

Lrl
r.--

"* E--f#.r)

;1-'C, .'
+

It

In
l

:-c,
In (b) JFET mixer

D r t

f, ,.

;:,-.c5
l

":-,.

...

frequenciesat its output, frequenciesthat will probably have to be removed by filtering. However,both typesof transistorare in common use in mixer circuits. Figure 2.30 includes circuits for three typical transistor mixers. Figure 2.30(a) uses a bipolar transistor, and Figure 2.30(b) uses a JFET. The circuit in Figure 2.30(c) usesa dual-gate metal oxide semiconductorFET (MOSFET) with one of the two signals to be mixed applied to each gate. The bipolar-transistormixer of Figure 2.30(a) resemblesa conventionaltuned RF amplifier, except that it has two inputs, one at the base and one at the emitter One of the two signals to be mixed will be applied to each input, with the stronger of the two going to the emitter. The output can be tuned either to the sum or the difference ofthe two input signal frequencies,whichever is required. In this circuil the resonantcircuits L;Cy and l2-C2 ne tuned toi and the required output frequency,respectively. Rtu R",, and R6 form the usual voltage-divider bias circuil C3is a coupling capacitorthat keepsL1 from short-circuiting the basebias, and Cp

Sectlon Z-5 a Mixers

vDD

Figure

?.3O

Continued

RD

f"rn

"'', ,f$ ,;it':.,,)o,, U

i*l + - t F vt i

- . - a

: .
Rc

*
p rrs

L, ;;;l g,

: Dual-gate MOSFET mixer rmJ R, form a decoupling network that keeps RF out of the power supply line. It is r-<o possibleto applyf2, as well asfi, to the base,summing them at this point. The JFET circuit of Figure 2.300) illustrates the technique of summing the =o input signals at a single input terminal. Bothfi andf2are applied to the gate. Th tuned circuits L;C1 and l2-C2have the samefunctions as before, and so do 'f_-and R2. Cacouples the/2 input into the circuit, C5bypassesthe sourceresistor, ^j Rs provides self-bias. In some JFET mixers, one of the input signals is apprled to the source. Dual-gateMOSFETs are very easyto use as mixers, as one input signal can be {f'tred to eachgate.This is demonstratedin Figure 2.30(c). Dual-gate MOSFETs ulke excellent mixers. They have better dynamic range than mixers using bipolar rzmsistors;that is, they can operatesatisfactorily over a wider range of signal amptrnrdes. They also produce fewer unwanted output frequencies.

\ *^ i Out

lelenced Mixers
:d by filuits. Figure ircuit in ;T) with raltuned emitter. stronger r the dif; circuit, put fre; circuit, .andCp .{ balrnced mixer is one in which the input frequenciesdo not appearat the outgu, Ideally, the only frequencies that are produced are the sum and difference of fu mput frequencies. -{ multiplier circuit, where the output amplitude is proportional to the prodrr of two input signals, can be used as a balanced mixer. This is easy to show mdematically. Let the multiplier have the equation
vo: uo : Av;,v;,

(2.23)

the instantaneous output voltage

v;, and v;, : A :

tha instantaneousvoltages applied at two input terminals aconstant

[-Er fr input consist of two sine waves of different frequencies, as before. For rmplicity, assumeeach signal has a peak amplitude of 1 V.
u;, : SlIl {D1l Sin @zt viz :

fhapter

? r Radlo-Frequencu

Elrcutts

Thenthe oueut will be

vo: Av;rvi,

: A sin <oltsin o2t We can use the same trigonometric identity, Equation Q.n), earlier with the square-lawmixer sinA sin B ::[cos to show that the output is ," : l[cos (o1 - az)t - cos (ar1-l a2)t] L As predicted, this mixer produces only the sum and difference of the input fte quencies.Thus it is an excellent choice for many applications.In fact, the usual bloctdiagram representationof a mixer is the sameas for a multiplier (seeFigure 2.3I ). Though general-purpose integrated-circuit multipliers exist, there are a nurF ber of balanced mixers designed especially for communications use. One of thesc is the MC1496, whose abridged data sheetis shown in Figure 2.32.T\ese deviccs are capable of operating over a very wide frequency range and are extremely weil balanced; that is, the input signal frequencies are reduced to very low levels. It is also possible to build a balanced mixer using discrete components. Fig: we 2.33 shows a traditional circuit using four diodes in a ring configuration. Fr best results, the diodes should be carefully matched.The input signals should nc have the same amplitude: the input atf2 should have an amplitude large enough b turn the diodes on completely, and the other input signal should be of much lowcr voltage. In order to understand the operation of this mixer, first consider an insterr when the /2 signal causes the secondary of transformer Z, to have the polarity shown in Figure 2.33, that is, the left side is positive. This will cause diodes D1 and D2to turn on and D3 and D4 to turn off. Dy and D2will then connect the muct weaker signal at the secondary of T1 directly to the primary of Z2 without aty polarity change. Now consider what happens when the polarity of the secondary of T3 re verses. D1 and D2wlll now be off, and D3 and Da will be on. The signal from Tt will still be connected to 72, but this tirne the polarity will be reversed. From this description, it can be seen that the circuit multiplies the signal atfi by a factor of * 1 or - 1, alternating atthe f2 rate. This is equivalent to multiplying the signal atfi by a square wave with frequency/2 and amplitude 1 V peakThe result will be the sum and difference frequencies described earlier plus -t higher-frequencycomponentsat3f2 ! fr,5f, fi, and so on. Generally,/2 is tb higher frequency, and these spurious signals will be at such high frequencies compared to the desired frequency that they will be very easy to filter out. Balanced mixers are also called balanced modulators and will be seen again under this name when we discuss the generation of suppressed-carrier signals in Chapter 5. (A - B) - cos (A + B)] that we usod

f,
Figure e.3l Mixer symbol

SECTTON?.5 REVTEW OUESTION

"mixer" is actually a summer. Explain what would happen if a real mixer An audio were used to combine two audio sisnals.

Order this document by MC1496/D

O
we used

*tcr'(oRcrLA

nput frenl block2.3r). e a numof these : devices rely well 'els. nts. Figtion. For ould not noughto ch lower r instant polarity iodesD1 he much rcut any rf 13 refrom T1 pal atfl rultiplyV peak. ier plus /2 is the ;uencies :n again gnals in

-arcse devices were designed for use where the output voltage is a rr':o.,c'tof an input voltage (signal) and a switching function (carrier).Typical cc"cations include suppressed carrier and amplitude modulation, $rrctrronous detection, FM detection, phase detection, and chopper -clications. See Motorola Application Note AN531 for additional design rq!:rnation. o, =rcellent Carrier Suppression +65 dB typ @ 0.5 MHz t50 dB typ @ 10 MHz tr r"qustable Gain and Signal Handling tl:ajanced Inputs and Outputs t! iih Common Mode Rejection +85 dB typical -"!s -vice contains8 activetransistors.

BALANCED MODULATORS/DEMODULATORS
SEMICONDUCTOR TECHNICALDATA

D SUFFIX PLASTIC PACKAGE CASE751A (sor14)

P SUFFIX PLASTIC PACKAGE cAsE 646

.14

#
Vee
N/C Oulput N/C Carrier Input N/C Input Carrier

PIN CONNECTIONS

/ib'

iln iln tt tlF , \dr, T ilr \v/ \l

/.dir.

rt{t
Figure 1. Suppressed Carrier Output Waveform

Signal Input Gain Adjust Gain Adjust Signal Input Bias Output N/C

ORDERING INFORMATION Figure 2. Suppressed Carrier Spectrum

Device
MC1496D MC1496P

Operating Temperature Range TA = OsC +705C to

Package

so+14
Plastic DIP

M C 1 4 9 6 8 P Ta = t+OsO +1255C Plastic DIP to

Figure 3. Amplitude Modulation Output Waveform mrxer

--------{
O Motorola, Inc. 1996

Figr.re

E.3P

Integrated-Circuit Balanced Mixel (Copyright of Motorola. Used by permission)

BZ

fhapter

e a Radio-Frequencg

Clrcults

R.tlng

Symbol

Vrls. 30

Unit
Vdc

Applied Voltage lv5 - V7,V6 - Vr, V9 - V7,V9 - V8,V7 - V4,V7 - V1, Vg - V4, V6 - vg, VZ - V5, V3 - v5l
Oifterential Input Signal

Vz-V8

+ 5.0 1(5+l5R) 10

Vdc

MaximumBiasCurent
Th6rmal Resisiance, Junction to Ail Ceramic Dual In-Line Packsge Plastic Dual In-Line Package Metsl Package Operating Temperalure Range MC1496 MC1596 Storage Temperature Range Tstq

RoLl 100 100 160 T4 0to +70 -55to +125 -65 to + 150

mA "c,^lv

'c

= = = V 2V E L E C T R I C A L C H A R A C T E R | S i l C S ' ( t 1C C d c , V e = - 8 . 0 V 4 " , 1 5= 1 0 m A d c . R g 3 9 k O ' R e l 0 k O ' unless otheNise notedl are unless otheMisenot"Oiiltt input and outputcharacteristics single-ended Ta = + 2E.C MCt596
Charlctari3tic Carrier Feedthrough Vc = 60 mV(rms) sine wave and offset adiusted to 2ero Vc = 3oo mvP-P square wave: oftsel adiusted to zero oftset not adjusted fc = t 0 kHz lC = 10 MHz fC - t.0 tHz fC = t.O ttiz

MCr496
Min TYp N 140 mV{rms) Mrx Unh FV(rms)

Fis

Not.

Symbol VCFT

Min

Tvp 40 140 0.04 20

Mrx

0.2 100

0.04 20

0.4 200
d8

Suppression Carrier mv(rms) tS = 10 kHr,30O fC : 500kHz,60 mv(rms)sinewave l. = 10 MHz,60mv(rms)sinewave

Transadmittance Bandwidth {Magnitude) (RL = 50 ohms) Car.ier Inpui Porl, VC = 60 mv{rms) sine wave lS = t.0 kHz, 300 mv{rmsl sine wave Signal Input Port, VS = 300 mV(rmsl sine wave lVCl = 0.5 Vdc Signal Gain Vs : 100 mv(rms), f = 1.0 kHz; lVCl = 05Vdc Single-Ended Input lmpedance, Signal Port, I = 5.0 MHz Parallel Input Resistance Parallel Input Capacitance Single-Ended Output lmpedance, I = 10 MHz Parallel Output Resistance Parallel Outout Capacitance Input Bias Currenl

vcs
65 50 40
o5

50 300 80

t
MHz

BW3og 300 80

t0

AVs

2.5

,tp

200 2.0 40 5.0 l2 25 0.7 0.7 2.O

l4

200

ko pF
k{t pF 30 30 7.0 1.0
nA/C

rop

,lt) 5.0 12 12 0.7 0.1 2.0 50

14

,or=5!,,0.=L#
Input Offset Currenl l;q5 = l1-14; lioc = lz-18 Average Temperature Coef{icient ol Input Oftsel Current (Tr = -55'Cto +125'C) 7

lbs
t6a

r,to5l

liocl 7 7
7 4

5.0

lrcriol
ilool iTCtool CMV
ACM

Output OffsetCurrent (15-19)

Average Temperalure Coefticient ol Output Offset Cu.renl {Tr = -55'Cio +125'C) = t 0 lHz Common-Mode Input Swing, Signal Port, fS Common-Mode Gain, Signal Port. tS = 1 0 kHr.

80
nAfC Vp-p

90

90

9 10 t0

-85 8.0 8.0 2.0 3.0 3.0 4.0 8.0 8.0 2.O 3.0 33 4.0 5.0

dB vp-p vp-p mAdc

lVc= osvdc

Outpulvoltage(Pin6 or Pin 9) Ouiescent Common-Mode DilterentialOutput VoltageSwing Cspabilitv Power Supply Current 16t19 DC Power Dissigation

Vout Vout lcc

PD

mW

Figure

2.3?

Continued

sectlilr

2.6 a Fre$lencg

Sgnttresizers

A3

Figure ?,33 diode mixer T2

Double-balanced

,i f,*f,\
Fi

+ f,-f,)

and

) Out

,A
:.6 Frequency Synthesizers
[: section 2.4, we saw that a free-running lC oscillator can be easily tuned to dif:erent frequencies becauseits operating frequency is usually determined by tuned :rcuits. For that reason, they are often referred to as variable-frequency oscilla-r;rs (VFOs). Unfortunately, they also exhibit undesired frequency changes as a :rnlt of vibration, voltage or temperature changes, component aging, and so on. : addition, accurately setting these oscillators to a particular frequency is tricky, =luiring precision-built variable capacitors or inductors and expensive dials :srng complicated arrangements of gears and pulleys. Crystal oscillators, on the other hand, have very good stability. With voltage :-:sulation and temperature control or compensation, frequency ddft can be rerred to a few parts per million over long time periods. The disadvantage, for T-rnv spplisafions, is that the frequency of a crystal oscillator can be changed :ol1- a very small amount by adjusting series or parallel capacitors. This makes r-; use awkward or impossible for any application that requires continuous freJtsncy variation or even operation at more than a few discrete frequencies. Most receivers and many transmitters require the frequency agility of theVFO :'"rupledwith the stability and accuracy of the crystal oscillator. For many years, it ras customary to use VFOs for applications where tuning had to be continuous or rhere there were large numbers of frequencies in use and to use crystal oscillators r:th switchable crystals when operation was required on only a relatively small rmtrer of different frequencies. In recent years, however, the phase-locked freJrency synthesizer has become very popular. It is now the preferred method of freJEncy generation in most modern receivers and transmitters. In fact, it is often rr:sible to savemoney as well as improve performance when a synthesizer is used nstead of a VFO becauseof reduced requirements for mechanical precision.

Realr\ucllor

Phase-Locked Loops
Before looking more closely at synthesizers, it is necessary to know something -uhout phaseJocked loop @LL), becauseit is the basis of practically all modthe -n synthesizer design. Though the PLL was actually invented in 1932, it is only mce 1970-when it was first produced on an IC-that it has been much more ::an a laboratory curiosity. A PLL synthesizer can certainly be consEucted using lbes or discrete transistors, but so many devices would be needed that the tech:lque is not practical.

14

Chapter 2 | Radio-Frequencg f'rtuits

+''* H*#ll$;:l$fl:i+
Figure2.34showstheessentialsofasimplePLL.Theloopconsistsofe (VCO), and a low-pass filter phase detector, a voltage-controlled oscillator ..loop filter." An external reference signal is compared with the G_pD cJled the voltage is produced whose amplitudc vco slgnal in the phLe detector. An error two signals. After filtering, this errr narier.Jith the phase difference between the VCO' signal is applied as a control voltage to the

Fisure ?.34 Phase-lockedloop

BHffi*

LpF

Hfho".

"9"

n*T#:,'lil'#r"::#:r*",1-T,.5:"I:3;l1,:l
frequency(thatiS,thefrequencyut*t'i.t'itoperateswhenthecontrolvoltage iszero).ThiswillprobablynotbethesameaSthereferencefrequency.Thc will generate a control voltagg loop is said to be uilocked. The phase detector w h i c h w i l l c a u s e t h e V C o f r e q u e n c y t o c h a n g e u n t i l i t i s e x a c t l y t h a t o f tphase he of phase lock Once external inpui .ignut. This is caUea ine acquisition locked indefinitely. Any tendency lock has u""n u""qui'"d, the loop will remain oftheVCotodriftinfrequencywillresultinachangeinthecontrolvoltage

; ;iffi'"T ll*'r"*:::3 ''* #:-#':'J**1,"':* ;ffi' I

purposeoftheloopamplifiershowninFigure2.34.Therewillalsobesomelirnitsonhowfarapartthefree-runningVCofreqlgncyandthereferencefrequency of course, it will also take a small can be for lock to be acquired or riaintained.

ff -1 #-dffi;ffi ffi ;irTr*::f :"*ml:*m,Lilig1:T: n::{y;ll*j**;"t"ffi *ru"";nril"",-r,:il#"r ru::n:*ri:ffi |

itdetermineshowfarapartthe"^t"'nutandinternalfrequenciescaninitiallyb

ffilfi#ruff*fifi.ff*qr
/ -uiTtT:';:"lf:JtTffJ,fm
:i-eiy^:;c; ture range. Fisure ?.35
specifications

pLL frequency

he lock andlock ,-g"' Tlt" lock

betwee' crear. rhe ailrhis Note difference largerthantrt" "oi I r'1r,sr uror urv vsr ls range is almost arways range almostalways
II

Setrtion 2-6 a Frequencg

Sgnthesizers

;ts of a ;s filter lith the rplitude error Lis

-\ phase-lockedloop has a VCO with a free-running frequency of tZMHz. -\s the frequency of the reference input is gradually raised from zero, the loop Iocks at 10 MHz and comes out of lock again at 16MHz. ' a I Find the capture range and lock range. -bl Suppose that the experiment is repeated, but this time the reference input begins with a very high frequency and steadily moves downward. Predict the frequencies at which lock would be achieved and lost.

EXAMPLE

Z-A

Output

t artion

rat The capture range is approximately twice the difference between the free-running frequency and the frequency at which lock is first achieved. For this example, That is, the two lnstant, unning voltage cy. The voltage, l of the e phase :ndency voltage is is the me lim)quency a small capture range : 2(12MHz - 10 MHz) : 4MHz The lock range is approximately twice the difference between the frequency r-here lock is lost and the free-running frequency. Here, lockrange :2(l6MHz :8MHz - l2MHz)
Loek Range

'b) The PLL frequency responseis (at least approximately) symmetrical; that is, the free-running frequency is in the center of the lock range and capture range. The frequency at which lock will be acquired, moving downward in frequency,is l2MHz 1 2MHz: L.ock will be lost on the way down at l2MHz-4MHz:8MHz Egure 2.36 shows theserelationshipsfor this example. Figure ?.36 l4MHz 8 1 0

Capture Range

/(MHz)

rquency radually CO freill lock. achieve Ition, as tially be :d to the moving CO will :ntually, ring frecan be between the cap-

SimpleFrequencySynthesizers
": may not be obvious how a PLL can fulfill our original goal of creating an oscil.uor with crystal-controlled stability and VFO agility without using a great num=r of crystals. To achieve this goal, we need to add one more component to the ,,rop: a programmable divider. See Figure 2.37 for a sketch of a frequency syndsizer reduced to its simplest terms. In this circuit, the phase detector still compares two frequencies and pro:uces a control voltage that results in the two becoming locked together. Howrer, while one of theseis still an external referencefrequency that could be genrated by a crystal oscillator, the other is no longer the VCO frequency itself. That frequency is divided by some integer number N and then compared with the :eterencefrequency. Using a programmable divider allows N to be varied. It is 3rsy to seethat, assumingthe PLL is locked, Phase Loop Detector Filter
, a

VCO
1 ' ' i

'

>

+.V * Programmable Divider Figure ?.37 synthesizer A simple frequency

fvco . I*r: N fvco: Nf,"t

Ehapter

? o Radio-Frequencg

Elrcults

the Generally, the VCO generates output frequencyf. Then

fo : Nf,"t

(2.41

This meansthat a large number of different output frequencies,all locked to a single crystal-controlled reference frequency, can be generatedsimply by changing the modulus (the value of M). The modulus can be changedby altering the voltageson someof the pins of the divider chip. Thereforethe techniquelends itself very easily tocomputercontrol and/orremotecontrol. Expensivevariablecapacitorsandinductors are also eliminated. The resulting cost savingscan actually make a frequency synthesizercheaperthan a conventionalVFO, in spiteof its greatercomplexity. just any frequeng/ There is one problem, however.This circuit cannotgenerate but only thosethat aremultiples offl For instance,iff,yis 100kHz, the circuit can, at leastin theory generateany multiple of 100kHz. This would be satisfactoryin an FM broadcastreceiver,sinceFM broadcastchannelsare200 kHz apart,but it would haveits limitations in AM broadcasting,where the channelspacingis only 10 kHz The minimum frequencystepis called the resolution of the synthesizer. The obvious solution is to reducef,1. Crystals with frequenciesmuch below 100 kHz are impractical, but a fixed-modulus divider can be used to divide down the referencefrequency,as shown in Figure 2.38.
Frequency syntheFigure ?.38 sizer with divider for the reference freouencv
r

Phase Detector

Loop Filter

VCO

.{,,}.1":-Fr -',ei
:
r
l

losc

f,"r

/';-..

;8,

- - J G- . * r . ,
l

:
I + N t Progtammable Divider

usinga l0 MHz a EXAMPLE e.g Configure simplePLL synthesizer from 540 to frequencies the crystal so that it will generate AM broadcast 1700kHz. will Solutlon Our synthesizer havethe block diagramshownin Figure is 2.38. Sincethe channelspacingin AM broadcasting 10kHz and all channels areat integermultiples of 10 kHz, it would be logical to usethis value by or fot f,"6 in that case,eachtime N is incremented decremented 1, the output frequencywill moveto the adjacent channel. to Sincewe areusing a 10 MHz crystal,it will be necessary divide it by a factor(shownasQ in Figure2.38)to get 10kHz. This factorcaneasilybe found asfollows: { 6 -J3E fr"r : 10MHz 10kH, :1000

seEtion ?-E I Frequencg Sgnthes,zers

(2.24\ a single nng the ageson y easily linducquency iry. quency :uitcan, rryin an twould l0kHz. r below Le down

Next, we should specify the range of values of Nthat will be required. We hrte already seenthat changingNby 1 changeschannels.All we needto do ben is findNat eachend of the tuning range.We can reiurangeEquation(2.24)

f" : Nf,"r
Er get

f^ fy' : :-" f'"t Ar tk low endof the band,we have

A _ " I *-

54OWlz lokH, :54

-{r the high end,

kHz ^, - 1700 " _ lokH" : l7O

lttscaling Ttrre is a problem with the basic synthesizer when output frequencies in the lHF range and higher are required. Programmable dividers are simply not rnr:Jable at frequencies much above 100 MHz. With current technology, it is n:r possible to build a UHF synthesizer with the simple topology of Figure ' -: i The simplest way to get a synthesizerto work at frequenciesbeyond those 5 rhich programmable dividers operate is to add a fixed-modulus divider in lfr"rnt of the programmable one, as illustrated by Figure 2.39. This divider ,xu1d employ emitter-coupled logic (ECL), a digital technology that can be c*ed at frequencies above I GHz. For even higher frequencies, discrete trann*".rs using gallium arsenide could be used. This will work well into the mi"rr--rsave region.
Phase Detector
e f ,'-/ ! L -1 -J

*f.

lanlue e out-

.-

l --l

r I

Loop Filter

VCO

Figure ?.39 Frequency synthesizer with fixed prescaler

Fixed-Modulus
Divider

I ,

i Prescaler ;

bya b Programmable Divider Fixed-Modulus Divider

There is one drawback to this idea. The VCO frequency is now divided by the fx;rJ modulus M, then by the programmable modulus N, before it is compared u:A the reference frequency. Thus

f,
J rel

MN

fhapter

Z I Radio-Frequencg

fircuits

f,:

MNf,.r

(2.2s1

Since only N can be changed,the minimum amount by which the frequency can be changedis now Mf*y. For example, if f*yis 10 kHz, and a 10 : I prescale is used,the minimum stepby which the frequency can be changedis 100 kHz. We seemto have taken one step forward and one back. A truly elegant solution to the problem, one that is often used in synthesizers for VHF and UHF, is to use a two-modulusprescaler This is a divider that can bc programmed to divide by either of two consecutiveintegers (for instance, 10 and 11 or 15 and 16). We can let these integersbe P and (P + 1). The frequency lim. itations of such dividers are not nearly as severeas for fully programmable dividers: using ECL, they can work up to at least I.2 GHz. In addition, they are much more flexible than single-modulus counters. Figure 2.40 shows a synthesizer with a two-modulus prescaler.The main counter divides by N, as before, but the prescaler can divide by either P r (P + 1). In addition, one more programmable counter is needed.Let it divide by M. The output of this counter switches the modulus of the two-modulus counter between P andP * 1. Remember that M, N, and P are all integers.M and N can be changed by writing to registers on the counters, but P cannot be changed.In addition, N must be greater tharrM, as will shortly become obvious.
FrequencYsYntheFigure 2.4O sizer with two-modulus prescaler Reference Oscillator Fixed-Modulus Divider Phase Detector Loop Filter

Output

To understand the operation of this synthesizer, suppose that initially the prescaler is set to divide by (P + 1) and the divide-by-M counter has just been reset.The two-modulus counter will switch to the P mode when the divide-by-if counter changesstate(after it has receivedM transitions from the VCO). Then the two-modulus counter will divide by P until the divide-by-M counter changesstate again. The trick here is that the M counter remains inactive until reset; that hap pens when the divide-by-N programmablecounter reacheszero. That is, on reacF ing zero, the divide-by-N counter resetsboth itself and the divide-by-M counter, setting each to the value stored in its respectiveprogramming register. Since initially the two-modulus counter divides by (P + 1), it will take (P + 1) transitions from the VCO to apply one count to the divide-by-M counter. Therefore, this counter will reach zero after M(P + 1) transitions on the VCO output line. The same pulses that go to the divide-by-M counter are input to the divide-by-N counter. By the time the divide-by-M counter has reached zero, the divide-by-N counter has reached (N - M). This value must be greater than zero, which accountsfor the condition imposed abovethat N must be greater than M.

Sectaon ?.6 O Frequencu sgntheslzers -\t this point, the prescalerswitches to its other mode. The next P transitions :'{n the VCO produce one transition at the prescaleroutput. This continuesuntil te dir-ide-by-Ncounter reacheszero. Since this counter begins this section ofthe rr,{ess with a count of (N - M), it must receive (N - M) output transitions :,s-lrrthe prescaler-this accounts for P(N - M) statechangesof the VCO out:u: -\t this point, both programmable countersreset to their programmed values, rcd the two-modulus counter switches back to its (P + l) mode. If we look at the whole sequence, seethat the total number of pulses emwe rn.:1i1o from the VCO for one output pulse to the phasedetector is M(P + l) + P(N - M): MP + M + NP - MP :M+NP

a9

(2.2s)
quency tscaler }lz. We resizers t can be , 10 and rcy limable dihey are re main erPor ir ide by counter d l{ can rged. In

ll;t is. the whole system including both the prescaler and the main counter has n'rded the VCO output by a factor of (M + NP). Therefore

f":

(M + NP)f,"f

(2.26)

The reader who has followed this rather convoluted argument may feel that tu:! is not a very spectacularresult from so many counters and so much head,c:rtching. It is actually worth all the trouble, however, as an example will
iliTr-'*.

Output

EAMPLE Z-lO The synthesizerin Figure 2.40 has P = 10 and -'.q : 10 kHz. Find the minimum frequency step size and compare it with fi't obtained using a fixed divide-by-l0 prescaler. cdrtirrn rould be With a fixed-modulus prescaler, the minimum frequency step step size : Mf,x : 10 x 10kHz : 10 kHz To find the step size with the two-modulus system,let the main divider nodulus Nremain constantand increasethe modulus M to (M * 1) to find how nrch the frequency changes. For the first case,the output frequency would be f" -* (M + NP)f.4 : (M + NP)10 kHz

ially the ust been de-by-M Ihen the ges state hat haprn reachcounter, rill take counter. he VCO ,ut to the zero,the lan zero, han M.

ff *'e now leave Nalone but changeM to (M * 1), the new frequency is

fL:(M + 1+ NP)f,"r : (M + 1 + NP)l0kHz

Thedifferenceis fi f" : (M + I + NP)l0YJlz - (M + NP)r0kIlz :(M+1+NP-M-NP)l0kHz : 10kHz

Thisis the samestepsizeaswouldhavebeenobtained withoutprescaling.

Ehapter

2 O Radio-Frequencg

Earcuats

Thus, the two-modulus prescaler achieves the benefit of prescaling (increased high-frequency capability) without the disadvantageof poorer resolution.

Frequency Tlanslation
All of the synthesizersshown so far are capable of generatingvery low frequencies, right down to f,r.. When such a synthesizeris used to generatea high frequency,a very large value of N is required. Dividing the VCO output by N divides the loop gain by the sameamount, and very large valuesof Ncan causeinstability. In many practical applications, it is completely unnecessaryfor the synthesizer to operate at very low frequencies.For instance, an FM broadcastreceiver must tune at200kllzintervals, but only between 88.1 and 107.9MHz. One way !o produce the local oscillator signal for such a receiver would be to generateit at a relatively low frequency and then raise it by mixing. Such a system is shown in Figure 2.4l.The final output does not cover fhe FM band but rather a range of frequencies 10.7 MHz higher, since it is a local oscillator signal that is being generated and lO.7 MHz is the usual IF in FM broadcastreceivers.Local oscillators arc discussedin detail in Chapter 6. The VCO output varies from 20 MHz to 39.8 MHz in 200kJIz steps.It is mixed with a fixed-frequency oscillator at 78.8 MHz, and the sum of the two frequenciesis used for the output, which varies from 98.8 to 118.6MHz, as required.A bandpass filter is usedto remove the difference component as well as the VCO and crystal oscillator frequenciesfrom the mixer output. This movement of a block of frequenciesis called a frequency translation.

Figurre E.4l Synthesizerwith frequency shifting

MIIZ 98.8-118.6
Ij

Phase Detector 2OOkHz

Loop Filter

-I5 iBPF ,
j

VCO

-i/-: ;d \-.',7

- -l

f *-t L &\__:

{'t- ) \v

20_39.9 MHz

-r X )Mixer _i

tLo

78.8MHz
Programmable Divider

Figure ?.zlP Synthesizerwith mixer in the loop

Phase Detector

Loop Filter

VCO

fr, *@F{E

*Gt"-z^r '' *-Mixer ( X \--, l

I

-r.
frc

Programmable Divider

BPF

Setrtlon ?.E I FrequenEg Sgnthes|'zers Another possibility is to include the mixer and local oscillator within the rrr4). as shown in Figure 2.42. In this case, the VCO provides the output frer-Ency directly, but its output is mixed down before being applied to the prog'r.rnmable divider.

9l

rcreased

frequenugh freI divides ;tability. synthereceiver e way to tertata hown in e of fres generrtors are to 39.8 8 MHz, om 98.8 ce com-\er outlation.

EXAMPLE Z.ll A synthesizerof type showninFigure 2.42has the 2OkHz anda local oscillatoroperatingat 10 MHz. Find the frequency i-r: rage of the outputasthe valueof Nranges from 10 to 100.Also find the nrinimulnamountby which the frequencycanbe varied. e*tlon The frequencyat the input to the divider mustbe N/4 by the rme logic that wasusedwith the simplesynthesizer. This is the outputof a nixer that subtracts fixed 10 MHz frequencyfrom theVCO frequency. the Ober mixer outputsareremovedby the bandpass filter. Thereforethe VCO frequency, which is alsothe outputfrequency, is f": fur ftn : lOMHz,f,4: f": Nf,"r+ fut

Z0kllz, andN : 10,we have I0 x 20kJlz * 10MHz : I0.2MHz

Hi{changesto 100,the outputfrequency be will

6 MHz

f" = lO0 x Z0kllz + l0 MHz : lZMHz S-now we know that the stepsizewill be constant,so we can simply divide fu total frequencyrangeby the numberof stepsto find it. l2MHz - 10.2MHz step size = 100- 10

BPF

Mixer

1.8 MHz 90
= 20kllz As you would expect,the stepsizeis simply equaltofoy.

lllz

"tr-: not always necessaryto use a secondcrystal oscillator for fys; with careful .n-'ice of frequenciesit may be possible to use the local-oscillator frequency,suitrnll divided down, as the referencefrequency.This technique is shown in Figure - -l-1. ln this case,the only additional stagerequired is the mixer itself.

Figure P.43 Derivation of fleand fo$rom the same source

Explainhow frequency synthesizers combine manyof the advantages l,C andcrysof nl oscillators.

sEfTlON e-E OUESTION

REVIEW