AN INTRODUCTION TO ACTIVE ANTENNAS

An antenna is a structure that converts electromagnetic en­ergy propagating in free space into voltage and current in an electrical circuit and/or vice versa. In a transceiver system, the antenna is used both to receive and to transmit free-space waves. At minimum, a transceiver then must consist of a sig­nal source that serves to drive the antenna as well as a re­ceiver circuit that reads out the signal from the antenna. Un­til recently, practically all antenna systems operating in the microwave frequency regime (operation frequencies greater than 1 billion cycles per second, or 1 GHz) were mostly de­signed to isolate the antenna from the circuits—that is, to find ways to make system operation independent of the an­tenna’s electrical characteristics. In contradistinction, an ac­tive antenna is one in which the antenna actually serves as a circuit element of either the driver or the readout circuit. To understand why this is different from conventional antenna driving or readout will require us to take a bit of a historical trip through the last century or so.

Actually, the first antenna was an active one. Heinrich Hertz, back in 1884 (2a), was the first to demonstrate that one could generate radio waves and that they would propa­gate from a transmitter to a receiver at the speed of light. The apparatus used is schematically depicted in Fig. 1. The idea of the transmitter is that, by discharging an induction coil (a wire looped about a magnetic core such that the com­posite device can store significant amounts of magnetic en­ergy) into a spark gap, one can generate a current in the 5 mm diameter wire. The voltage in the spark gap induces a current in the wires, which in turn induces a voltage in the wires, and this voltage in turn induces current, so that the voltage and current propagate along the two pieces of the wire

J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering. Copyright © 1999 John Wiley & Sons, Inc.

AN INTRODUCTION TO ACTIVE ANTENNAS

Induction

Figure 1. Hertz apparatus for (a) transmitting and (b) receiving ra­dio waves, where the transmitting antenna serves to choose a specific frequency of the spark gap voltage to transmit to the receiving an­tenna, which also serves to pick out this special frequency from the free-space waveform and turn this electromagnetic disturbance into a voltage across the receiver antenna gap.

AN INTRODUCTION TO ACTIVE ANTENNAS

to either side of the gap as waves, appearing much like a one­dimensional slice through a water wave propagating away from the point where a pebble has struck the water’s surface (the spark gap). A wave will propagate rectilinearly until it encounters an obstruction, at which point it can suffer reflec­tion from or transmission into the barrier that the obstruction presents. There will be reflections then off the metal spheres on the ends of the wire. The spark will generate a broad spectrum of frequencies or wavelengths. The reflections off the two ends, though, will tend to cancel each other except at certain special frequencies. The effect at these wrong frequencies is much like the effect ofthrowing a handful ofpebbles into the pond and not­ing that, in between the points where the pebbles struck, the waves are much more indistinct than they are far from where the handful struck the surface. The special frequencies are ones which just fit into the region between the spheres. The current needs to be zero at the two ends in order to fit, whereas the volt­age needs to be maximum at the ends. The current and voltage waves at the right frequencies may appear as depicted in Fig. 2.

The Hertz transmitter is the archetypical active antenna. The source is the spark gap, which is actually placed in the antenna. The antenna then acts as a filter to pick the right frequency out of a large number of frequencies that could be launched from the gap. The receiver is picked to be of a length to also select this primary frequency.

Hertz-style spark gap transmitters, after further develop­ment and popularization by Marconi, were in use for fifty years after Hertz. However, such transmitters exhibit some rather severe drawbacks. The main problem is that the sim­ple resonant dipole antenna (that is, a straight wire antenna with a gap or a feeder cable used to feed in current) is a filter with a poor frequency selection. Namely, if one increases the

AN INTRODUCTION TO ACTIVE ANTENNAS

Figure 2. Current and voltage waveforms for the lowest-order (least number of zeros) waveform for the Hertz transmitter of Fig. 1(a). The current must go to zero at the points where the wire ends, whereas the potential will be highest there.

AN INTRODUCTION TO ACTIVE ANTENNAS

Figure 3. A sketch of what the transmission as a function of fre­quency might look like for the Hertzian dipole antenna of Figs. 1 and 2.

frequency by 50%, there is 75% as much power transmitted at this frequency as at the first resonance, which is called the fundamental. There is a second resonance at twice the frequency of the first resonance, and another at each integer multiple of the fundamental. With increasing frequency, the transmitted power decreases a little and then flattens out around the second resonance, decreases a little, flattens out at the third resonance, etc., as is illustrated in Fig. 3. If the spark discharge is really broadband (that is, if it generates a large number of frequencies where the highest frequency may be many times the lowest), then what is transmitted by the antenna will also be broadband, although with somewhat higher transmission at the fundamental frequency and its harmonics than in between. In the very early days of radio, this was somewhat acceptable, although any information im­pressed on such a broadband carrier would be rather severely degraded upon reception. However, the demise of the spark gap transmitter was really instigated by the early success of radio, which caused the available frequency bands to begin to fill up rapidly. This band filling led to the formation of the Federal Communications Commission (FCC) in 1934, which was charged with allocation of frequency bands. The alloca­tion by nature led to a ban on spark gap transmitters, which were needlessly wasting bandwidth.

In a later experiment, Hertz noticed that the waves he was generating would tend to have a component that hugged the ground and could therefore travel over the horizon and, in fact, across the Atlantic Ocean, skimming along the surface of the water. Other researchers noticed that the effect became more pronounced at wavelengths longer than the roughly 2 m wavelength that Hertz originally used. (For the frequencies and wavelengths of some important frequency bands, see Ta­ble 1.) In order for wave transmission to be useful, however, the transmitted signal needs to carry information. Impressing information on the wave is called modulating the carrier. One can modulate the height (amplitude), the frequency, and so on. The discovery of a technique to amplitude-modulate the waves coming off an antenna (in 1906) then led to the incep­tion of AM radio in bands with wavelengths greater than 300 m, which corresponds to roughly 1 MHz. AM radio became commercial in 1920. By the 1930s, other researchers noted that waves with frequencies around 10 MHz, corresponding to a wavelength around 30 m, could be quite efficiently propa­gated over the horizon by bouncing the wave off the iono­sphere. This led to the radio bands known as short-wave. In 1939, a researcher realized a technique to modulate the fre­quency of the wave. This realization led in the 1950s to FM radio, which was allocated the band around 100 MHz with

Table 1. A Listing of the Allocated Microwave and Millimeter-Wave Bands as Defined by the Frequency and Wavelength Range Within Each Band

Band

Designation

Frequency

(GHz)

Wavelength

L

1-2

15-30 cm

S

2-4

7.5-15 cm

C

4-8

3.75-7.5 cm

X

8-12

2.5-3.75 cm

Ku

12-18

1.67-2.5 cm

K

18-26

1.15-1.67 cm

Ka

26-40

0.75-1.15 cm

Q

33-50

6-9 mm

U

40-60

5-7.5 mm

V

50-75

4-6 mm

E

60-80

3.75-5 mm

W

75-110

2.7-4 mm

D

110-170

1.8-2.7 mm

G

140-220

1.4-2.1 mm

Y

220-325

0.9-1.4 mm

a corresponding wavelength around 3 m. However, the FM technique was used first during World War II as a radar mod­ulation technique. Radars today are at frequencies above roughly 1 GHz or wavelengths below 30 cm.

There is a fundamental difference between circuits that op­erate at frequencies whose corresponding wavelengths are less than the maximum circuit dimension and those that are large compared to the carrier wavelength. The effect is closely related to the concept of impedance. As was mentioned above, in the wire antenna, the voltage and current reinforce each other and thereby travel on the antenna as waves. The same effect takes place in a circuit. At any point along the path (line) in a circuit, one defines the ratio of voltage at one fre­quency to the current at the same frequency as the impedance at that frequency. For a sinusoidal waveform, if the imped­ance tends to preserve the phase relationship (where the wave peaks lie, relatively), then we say that the impedance is resistive. If the impedance tends to drive the voltage peaks forward with respect to the current peaks, we say that the impedance is capacitive; in the opposite case we say that the impedance is inductive. In a small circuit (small compared to a wavelength), one generally tries to carefully design passive components—resistors, capacitors, and inductors—so that

R

о—vw-

L

C

Battery

Ground

Figure 5. A circuit with lumped elements connected by wire seg­ments.

they exhibit large local impedance, that is, large impedance within their physical dimensions. When the circuit is small, one would like to control the phase and amplitude of the wave at discrete points by using lumped elements and thereby min­imizing line effects. The lines (wires) between the components have little or no effect on the electromagnetic disturbances passing through the circuit, then, as the impedances in the wires are small and reasonably independent of their lengths. When the circuit is large, the lines themselves effectively be­come circuit elements, and they themselves must be carefully designed in order to exhibit the proper impedances. To illus­trate, consider the parallel plate capacitor of Fig. 4. The ca­pacitance is maximized by maximizing the permittivity e (a material parameter equal to the ratio of electrial displace­ment to applied electric field) and area A while minimizing the plate spacing d. However, the fact that the capacitance depends on the plate spacing d is the important point here. Consider the circuit of Fig. 5 as an example. The only ground in the figure is the one on the battery, but the wires connect­ing the circuit elements together in essence form at each point a capacitor, with a point on the wire that is carrying charge as the upper plate and the ground as the lower. This capaci­tance changes as a function of position along the wire. For a small enough circuit (relative to the wavelength of the highest frequency carried by the circuit), the effect is not too impor­tant, as the wire-ground pair has small capacitance and the position-varying effect is small. For a large circuit, the effect is disastrous, as we shall consider below. The effect is identi­cal to the effect of Fresnel coefficients in optics.

Consider the circuit of Fig. 6. We will now discuss what happens when impedances are not carefully controlled. This leads to the concept of impedance matching. Let us first say that the circuit is short (compared to a wavelength). If the load resistor, RL, is not matched to (that is, is not equal to, or, one could say, not impedance matched to) the resistance of the source, RS, some amount of reflection will occur at RL, propa­gate back to RS, be reflected with a reversal of sign at RL,

Area A

X

Rs

AW

— + + + + + + + + + + ++T^^^ —i-

е d

T

©

•R,

Figure 4. Schematic depiction of a parallel plate capacitor in which the flow of a current will tend to change the upper plate, causing a voltage difference between upper and lower plates. The capacitance is defined as the ratio of the amount of change of the upper plate to the magnitude of the voltage this change induces between the plates.

Figure 6. A circuit in which one is trying to supply power from a source with internal resistance RS to a load with resistance RL. The power transfer is maximized only when RS and RL are equal, in which case half the power supplied by the source is supplied to the load, the other half being dissipated in the source and causing it to heat.

AN INTRODUCTION TO ACTIVE ANTENNAS

(ground potential)

Figure 7. A coaxial cable in which signals are carried on an inner conductor and in which the grounded outer conductor serves to carry the ground plane along with the signal in order to give a constant impedance along the line.

propagate back to RL, etc. The reflections add up perfectly out of phase (that is, simply subtract from one another) at the source and load, and the amount of power supplied to the load is less than optimal. In this limit of a small circuit, it is as if the load will not allow the source to supply as much power as it is capable of. Let us now say that the line is ‘‘well-designed’’ but long compared to the wavelength used. Then the same argument applies to the reflections, but in this case the source does not know that the load is there until several wave peri­ods have passed (several maxima and minima of the wave­form have left the source), so the source supplies all the power it can. The power, though, is not allowed to be fully absorbed by the load, and some of it will rattle around the line until it is radiated or absorbed. As we mentioned above, in a long enough circuit the wire itself becomes a distributed element— that is, one with an impedance of its own. If the distance to the nearest ground is not kept fixed along the line, the induc­tance and capacitance become dependent on the position. In this case, we have distributed reflections all along the line and the circuit will probably not work at all. This spatial vari­ability of the line impedance is remediable, though, as illus­trated by the drawing of a coaxial cable in Fig. 7. The idea is that, if the line brings along its own ground plane in the form of a grounded outer conductor, the characteristic impedance of the line can be kept constant with distance. Such a line, which carries its own ground plane, is called a transmission line. The problem becomes the connection of the line to the source and load (i. e., impedance matching).

AN INTRODUCTION TO ACTIVE ANTENNAS

Figure 8. Schematic of a conventional RF microwave trans­mitter in which each individual element of the transmitter is matched to each other element.

Before going on to discuss the conventional solution versus the new active antenna solution, perhaps we should summa­rize a bit. In AM, short-wave, and FM applications, the wave­lengths are of order greater than meters. If one considers typ­ical receivers, the whole circuit will generally be small compared to the carrier wavelength. This is also to say that in all of these cases, the antennas will be active in the sense that the antenna presents an impedance to the circuit. (Recall that an active antenna is any antenna in which an active ele­ment lies within a wavelength of the antenna and is used as an element to match the antenna impedance to the decoder impedance.) To passively match an antenna to the receiver circuit, one needs pieces of line comparable to a wavelength. However, from here on we shall not be interested in the low- frequency case but rather in the well-above-1-GHz case, as AM, FM, and TV technologies are mature technologies. Dur­ing World War II, radar was the application that drove the frequencies above 1 GHz (wavelength less than 30 cm). In a radar, one sends out a pulse and, from the returned, scattered wave, tries to infer as much as possible about the target. Tar­get resolution is inversely proportional to wavelength. There has been a constant drive to shorten wavelength. Therefore, as is indicated by Table 1, bands have been allocated out to hundreds of gigahertz. Presently, however, there are a pleth­ora of nonmilitary drivers for pushing to higher-frequency communication systems that are compact and have lower power dissipation. However, the conventional solution, which was developed originally for radars, is really not conducive to compactness nor to the pressures of cost minimization of the commercial market.

A typical conventional transmitter is schematically de­picted in Fig. 8. A main concept here is that the transmission lines and matching networks are being used to isolate the oscillator from the amplifier and the amplifier from the an­tenna, in contrast to the situation in an active antenna. There were a number of reasons why the conventional solution took on the form it did. Among them was the urgency of World War II. Radar was developed rapidly in both Great Britain and the United States in the 1930s and 1940s. Rapid develop­ment required numerous researchers working in parallel. When operating frequencies exceeded 1 GHz (corresponding to 30 cm wavelengths), passive matching networks, whose main requirement is that they must consist of lines of lengths comparable to a wavelength, became convenient to construct (in terms of size) for ground-based radar. In this case, then, the oscillators could be optimized independently of the ampli­fiers, which in turn could be optimized independently of the antennas and the receiver elements. The impedances of the individual pieces didn’t matter, as the matching networks could be used to effectively transform the effective imped­ances looking into an element into something completely dif­ferent for purposes of matching pieces of the network to each other. There are costs associated with such a solution, though, such as total system size as well as the tolerances that compo­nents must satisfy. However, once the technique was in place, the industry standardized on the conventional solution and perfected it to the point where it was hard to challenge. The reemergence of the active solution owes itself to two indepen­
dent technologies, the emergence of high-frequency solid-state devices and the development of planar circuit and planar an­tenna technology.

A single frequency of electromagnetic energy must be gen­erated in a so-called oscillator—that is, a circuit that converts dc electrical power to ac electromagnetic power at the proper frequency. The basic operation of an oscillator can be de­scribed with respect to Fig. 9. What is shown here schemati­cally is an amplifier in which a portion b (<1) of the output is fed back to the input with either a plus or a minus sign. When the feedback is off (b = 0), then the signal out will be just G times the input. When the feedback is negative, the output will be less than G times the input. However, in the negative feedback mode, the stability to noise increases, since fluctuations will be damped. That is, if the output fluctuates up, this lowers the effective input, whereas if the output fluc­tuates down, the output is driven up. The opposite is true in the positive feedback case. In the positive feedback case, if there were no fluctuations, any input would cause the output to increase until all of the dc power in as well as all of the input signal showed up at the output. (This is all of the power that can show up at the output. Such behavior is typical of unstable operation.) This would not be such an interesting case; however, there are always fluctuations of the input, and the positive feedback will cause these to grow. If there is a delay from output to input, then fluctuations with a period corresponding to this delay will be favored, as a rise in the input will show up as a rise in the output one period later, and rapidly all of the dc power in will be converted to power out at this magic frequency.

A real circuit operates a bit more interestingly than our ideal one. In a real circuit, as the fluctuations build up, the gain is affected and some elements absorb power, but the os­cillations still take place, although perhaps with a different frequency and amplitude from what one would have predicted from nondynamic measurements.

The transistor was first demonstrated in 1947, with publi­cation in 1948 (3), and the diode followed shortly (4). Al­though the field effect transistor (FET) was proposed in 1952

(5) , it was not until the mid 1960s that the technology had come far enough that it could be demonstrated (6). The FET (and variations thereof) is presently the workhorse microwave three-terminal device. Two-terminal transfer electron devices

Metal

AN INTRODUCTION TO ACTIVE ANTENNAS

Figure 10. Views of (a) a microstrip and (b) a coplanar waveguide line. In the microstrip, the ground plane is the lower electrode, whereas in the coplanar waveguide the ground plane is placed on the surface of the dielectric substrate.

Dc

power

in

I

AN INTRODUCTION TO ACTIVE ANTENNAS

Figure 9. Schematic depiction of a feedback system that can operate as an oscillator when G is greater than 1, the feedback is positive, and there is a delay in feeding back the output to the input.

(TEDs) were used before the FET for microwave applications and are still in use, but tend to have a much lower wall plug efficiency (dc to ac conversion), especially as the amplifying device of an oscillator. Radar systems, however, were already in use in the late 1930s. Essentially all of the microwave sources in radars up until the 1970s operated on principles that required that the source have physical dimensions larger than a wavelength, and perhaps many wavelengths. This fact almost required the conventional solution to be used. Transis­tors, though, can have active areas with dimensions of mi­crometers; even packaged hybrid devices can have complete packages of dimensions smaller than a millimeter. The tran­sistor can therefore act as an amplifier with dimensions much smaller than a wavelength and does not, therefore, need to be placed in a conventional (passive) solution design.

The last piece of our story of the new active antenna era involves the development of printed circuit technology, along with slot and patch antennas. The two most common planar ‘‘open waveguide’’ designs are microstrip line and coplanar waveguide (CPW). Depictions of these waveguide lines are given in Fig. 10. The idea behind the microstrip line is to propagate electromagnetic energy along the lines by confining the electric field between the upper signal line and a lower ground plane. As the upper line carries current, a magnetic field encircles the upper line. As power flow takes place in a direction perpendicular to the electric and magnetic fields, the power flow is mostly between the signal line and the ground

AN INTRODUCTION TO ACTIVE ANTENNAS

Figure 11. A simple transistor oscillator implemented in CPW tech­nology.

AN INTRODUCTION TO ACTIVE ANTENNAS

Figure 12. A depiction of (a) a patch antenna in a microstrip line and (b) a slot antenna in a CPW line.

line in the dielectric. On a low-frequency wire (a line whose transverse dimensions are small compared to a wavelength), the voltage and current waveforms reinforce each other. The coupling of the electric and magnetic fields in the microstrip is analogous to the coupling of voltage and current on the Hertz antenna wire, except that the microstrip line can be electrically long in the sense that the distance from the signal line to the ground plane is kept constant so that the imped­ance can be kept constant, as with the earlier-discussed coax­ial cable. Lines that carry along their ground planes are gen­erally referred to as transmission lines. Components (i. e. capacitors and inductors) can be built into the line by chang­ing the width, cutting gaps into the upper line, or putting slits in the ground plane. In this sense, we can still describe transmission line circuits by conventional circuit theory if we use a special circuit model for the line itself. The CPW line is quite similar to the microstrip line except that there the ground planes are on top of the dielectric slab. Either of these line types is reasonably easy to fabricate, as one needs only to buy a metal-coated dielectric plate and then pattern the needed shapes by photographically defining the patterns us­ing a technique known as photolithography, a process com­mon to all present-day circuit fabrication. These planar struc­tures are quite compatible with transistor technology, as is indicated by the simple transistor oscillator circuit depicted in Fig. 11. The gap in the line on the drain side is there in order to provide the proper feedback for oscillation. In this case, the total oscillator linear dimension can be less than a wavelength.

In order to have an active antenna, one needs to have a radiating element—that is, a passive antenna element in the active antenna. There are antenna technologies which are compatible with microstrip and CPW technologies, and the resulting antenna types are illustrated in Fig. 12. The idea behind either of these antenna types is that the patch (slit) is designed to have a transverse length that matches the op­erating wavelength (as we discussed in conjunction with Hertz dipole antennas). In the case of the patch, the electric field points primarily from the patch to the ground plane, as is illustrated in Fig. 13. The edges of the transverse (to the input line) dimension will then have a field pattern as sketched in Fig. 13(a), and the longitudinal edges will have a field pattern as sketched in Fig. 13(b), with a composite sketch given in Fig. 13(c). The important part of the sketches, however, is really the so-called fringing fields in Fig. 13(a)—

Ct 1111 0

(b)

AN INTRODUCTION TO ACTIVE ANTENNAS

Figure 13. Illustration of the electric field directions along (a) the nonradiating edge and (b) the radiating edge, and (c) a schematic depiction of the edge fields around the patch.

that is, the fields that point neither up nor down but rather across. Beyond the longitudinal edges of the patch are fields, in phase for the two edges, that are normal to the surface. It is these fields (when combined with transverse magnetic fringe fields in the same strips) that give rise to the upward radiation. Similar arguments describe the operation of the slit antenna if one exchanges the electric and magnetic fields in the argument.

We have now introduced all of the pieces necessary to de­scribe the new resurgence in active antenna research. A pos­sible active antenna design could appear as in Fig. 14 (7), where the transistor is actually mounted right into the patch antenna element, and therefore the design can be quite com­pact. That is, the source plus oscillator plus antenna can all be fitted into less than a wavelength. The design of Fig. 14, which comes from R. Compton’s group at Cornell (31,32), will be discussed further in the next section.

AN INTRODUCTION TO ACTIVE ANTENNAS

Figure 14. Depiction of the upper surface metallization of a micro­strip active patch antenna discussed in Ref. 7. The short on the gate together with the slit between gate and drain provides the proper feedback delay to cause oscillation.

There are a number of advantages to the use of active an­tennas. One is that an active antenna can be made compact. Compactness in itself is advantageous, as throughout the his­tory of microelectronics, miniaturization has led to lowered costs. There are two more advantages, though, which relate to compactness. One is that the power-handling capabilities of a device go down with increasing frequency. We would therefore like to find ways to combine the power from several devices. One can try to add together outputs from various os­cillators in the circuit before feeding them to the elements, but this goes back to the conventional solution. A more advan­tageous design is to make an array of antennas, with proper spacing relative to the wavelength and antenna sizes, and add the power of the locked oscillators in the array quasi — optically in free space. (In other words, optical radiation tends to radiate into free space, whereas radio frequency in micro­wave radiation needs to be kept in guiding waveguides until encroachment on radiating elements. Quasi-optics uses the principle of the optical interferometer to combine multiple co­herent microwave fields in free space.) The locking requires that the oscillators talk to each other so that the phases of all of the array elements stay in a given relation. As will be dis­cussed in more detail in the next section, however, an impor­tant problem at present in the active antenna field relates to keeping elements locked yet still being able to modulate the output as well as steer the beam in order to be able to elec­tronically determine on output direction. These issues will be discussed in the next section and taken up in more detail in the last section.

where E is the electric field vector, B is the magnetic induc­tion vector, H is the magnetic field vector, D is the electric displacement vector, J is the current density vector, and p is the volume density of charge. An additional important quan­tity is S, the Poynting vector, defined by

S = E x H

If one takes the divergence of S, one finds V S = V(E x H)

If one assumes a free-space region,

D = e0E B = ix0H

which is therefore lossless,

J = 0

and charge-free,

p = 0

(where e0 is the permittivity of free space and p0 is the perme­ability of free space), one can use vector identities and Max­well’s equations to obtain

V-S = -^ — — (H H)

2 31 ; 2 31 ;

AN INTRODUCTION TO ACTIVE ANTENNAS

Updated: 13.02.2014 — 23:08