AN-GT101A-Microwave Power Amplifier Fundamentals 08-10-27 电路图.pdf

A Giga-tronics Technical White Paper AN-GT101A 4650 Norris Canyon Road San Ramon, CA 94583 925-328-4650 Ph Microwave Power Amplifier Fundamentals Written by: Carlos Fuentes Member Technical Staff Giga-tronics Incorporated Published: October 2008 Revision: A RadioFans.CN 收音机爱 好者资料库 A Giga-tronics Technical White Paper AN-GT101A 2008 Introduction The Need to Amplify Signals An amplifier is one of the most common electrical elements in any system. The requirements for amplification are as varied as the systems where they are used. Amplifiers are available in a large number of form factors ranging from miniscule ICs to the largest high-power transmitter amplifiers. In the following discussion the focus will be on solid state power amplifiers used at microwave frequencies, particularly in test and measurement applications. Microwave power amplifiers may be used for applications ranging from testing passive elements, such as antennas, to active devices such as limiter diodes or MMIC based power amplifiers. Furthermore, other applications include testing requirements where a relatively large amount of RF power is necessary for overcoming system losses to a radiating element, such as may be found at a compact range, or where there is a system requirement to radiate a device-under-test (DUT) with an intense electromagnetic field, as may be found in EMI/EMC applications. As varied as the system requirements may be, the specific requirements of a given amplifier can also vary considerably. Nevertheless, there are common requirements for nearly all amplifiers, including frequency range, gain/gain flatness, power output, linearity, noise figure/noise power, matching, and stability. Often there are design trade-offs required to optimize any one parameter over another, and performance compromises are usually necessary for an amplifier that may be used in a general purpose testing application. The following discourse includes a description of amplifier topologies introducing the basics of spatially combined distributed amplifiers, a discussion of typical amplifier specifications and a review of performance verification measurements. Broadband Microwave Power Amplifiers There are numerous techniques for designing microwave power amplifiers. These may be broadly split between tube and solid state technologies. For high power requirements ( 100 Watts), typically these are satisfied with tube based designs. Tube amplifiers, such as Traveling Wave Tube Amplifiers (TWTAs), require a high voltage power supply, typically require warm-up time, and have significant aging related issues. For solid state amplifiers to achieve similar performance often requires switching between narrow-band amplifiers, with deleterious effects to the overall linearity and gain/power flatness. The switches themselves embody performance compromises. Mechanical switches, while quite linear and relatively low loss, have switching speed limitations, and are subject to failure after repeated switching cycles. Solid state switches may overcome the speed issue, but are not nearly so linear or low loss. Both the signal fidelity and loss issues limit the usefulness of solid state switches for high power microwave amplifiers. Furthermore, switching between narrowband amplifiers requires external stimulus with the software control complication that entails. 2 2 AN-GT101A Microwave Power Amplifier Fundamental Giga-tronics Incorporated. All Rights Reserved. RadioFans.CN 收音机爱 好者资料库 A Giga-tronics Technical White Paper AN-GT101A 2008 A topology often favored for generating modest amounts of microwave power output is to combine the outputs of several relatively low output power amplifiers. The individual amplifiers usually have a “distributed” or “traveling wave” topology1. The distributed amplifier topology achieves a large frequency range by arraying individual transistors; each representing shunt capacitances between series inductances, to create a semi-lumped representation of a transmission line (see Figure 1). This amplifier topology is often fabricated using MMIC techniques, and has been optimized to the point where single amplifiers can provide up to nearly 1 Watt of saturated power output. Nevertheless, it is no trivial task, using conventional planar circuit techniques, to combine the power output of even a small number of these distributed amplifiers over a full decade frequency range, without incurring unacceptable losses or poor flatness characteristics. Figure 1: Four Cell Distributed Amplifier New techniques, employing spatial combining, enable high power and flat gain over a broad bandwidth. Spatial combining can be used to sum a much larger number of amplifiers over a decade-wide frequency range, than would be practical using conventional planar circuit techniques. For example, an amplifier cell can be created using tapered-gap antipodal finline baluns2 to transition microstrip to a balanced finline structure suitable for launching into a coaxial waveguide. Antipodal finline is a balanced planar transmission line structure where two conducting strips of metal, separated by a dielectric substrate, are offset by a typically small gap (see Figure 2). Antipodal finline can provide a wide range of characteristic impedances. Closed form design equations for this transmission line media are not available; however, if the dielectric media is thin, and the gap is large, the design equations for unilateral finline can be used as an approximation. 3 3 AN-GT101A Microwave Power Amplifier Fundamental Giga-tronics Incorporated. All Rights Reserved. RadioFans.CN 收音机爱 好者资料库 A Giga-tronics Technical White Paper AN-GT101A 2008 a h w b 0 r Figure 2: Antipodal Finline Both microstrip and antipodal finline transmission lines support quasi-TEM propagation; i.e., the electric and magnetic fields are mutually perpendicular, and perpendicular to the direction of propagation. TEM mode propagation is naturally a very low dispersion propagation mode. By tapering the transition from microstrip to antipodal finline, the electro-magnetic fields are gracefully transitioned from an unbalanced structure (microstrip), to a balanced structure (antipodal finline), without introducing excessive dispersion. Innately, this balun is a very broadband, low dispersion transducer. A cylindrical array of amplifier cells, each with the previously described transitions, can launch into (or from) a coaxial waveguide. With the input coaxial waveguide being an n-way power splitter, and the output being an n- way power combiner, the antipodal finline excites or receives a TEM wave in each coaxial waveguide. Unlike common rectangular waveguide, the principle propagation mode in a coaxial waveguide is TEM, just like an ordinary coaxial cable, thus is low dispersion, and yet low loss. Using a tapered coaxial waveguide allows many relatively high impedance sources to be efficiently combined (or split), whereas this would be very difficult and inefficient to reproduce using purely conventional planar circuit techniques over a decade of frequency range. Thus the entire spatially combined propagation structure from beginning to end is quasi-TEM with low dispersion. Furthermore, the microstrip-antipodal finline baluns create a very broadband impedance transformation, allowing efficient combining/splitting of a large number of parallel amplifiers. 4 4 AN-GT101A Microwave Power Amplifier Fundamental Giga-tronics Incorporated. All Rights Reserved. A Giga-tronics Technical White Paper AN-GT101A 2008 Specification Parameters There exist a very large number of potential electrical specifications that can be applied to a microwave power amplifier. Nonetheless, there are a number of specifications that would be nearly universal, and deserve further discussion. Gain/Gain Flatness Gain usually is specified within the context of power output. Often, if no context for power output is given, then this is assumed to be small signal gain. Conditions for small signals at the input and output are usually easy to reproduce and verify, whereas gain and gain flatness can vary significantly when an amplifier approaches compression. Gain flatness for an amplifier with a significant frequency range is often specified over subsets of the entire frequency range, as well as over the entire frequency range. Gain and Gain Flatness typically include an implicit assumption that the reverse gain from the output to the input is negligible; i.e., the amplifier is unilateral. Since voltages and currents can be difficult to directly measure at microwave frequencies, network scattering parameters, or s-parameters, were defined in terms of incident and reflected waves. The formal definition for scattering parameters for a two port network can be derived as3: 0 1 2 Z 101 IZV a + = Incident Wave - Port 1 (1) 0 2 2 Z 202 IZV a + = Incident Wave - Port 2 (2) 0 1 2 Z 101 IZV b = Reflected Wave - Port 1 (3) 0 2 2 Z 202 IZV b = Reflected Wave - Port 2 (4) Voltage V1 is the voltage drop across Port 1 of the two port network, and I1 is the current into Port 1. The quantities V2 and I2 are the corresponding values at Port 2. Impedance Z0 is the characteristic impedance of the network, and is usually 50 . The factor of applies where the voltage and current are peak values. The factor is dropped for rms voltages and currents. The voltages and currents are vector (real and imaginary) quantities. The s-parameters can then be defined as: 0 1 11 2= a a S 1 b Port 1 Reflection Coefficient (5) 5 5 AN-GT101A Microwave Power Amplifier Fundamental Giga-tronics Incorporated. All Rights Reserved. A Giga-tronics Technical White Paper AN-GT101A 2008 0 1= a 2 1 12 a b S Port 2 to 1 Transmission Coefficient (6) 0 2 21 1 = a a b S 2 Port 1 to 2 Transmission Coefficient (7) 0 1= a 2 2 22 a b S Port 2 Reflection Coefficient (8) The defined scattering parameters are also vector quantities. Nominally, it is assumed that port one is the input port, and port two is the output port. Using this assumption, S11 is the input reflection coefficient, and S22 is the output reflection coefficient of the two port network. The transducer power gain is defined as the power delivered to the load divided by the power available from the source. Using network analysis flow graphs, the expression for the transducer gain is: () () 22 22 2 1/ 1 gg L src del T b b P P G = (9) Where: g = Generator Reflection Coefficient L = Load Reflection Coefficient bg = Wave incident from Generator This expression can be re-written after some algebraic manipulation into terms of s-parameters as: ()() ()() 2 2 2 2 21 11 11 Lg T SSSS S G = 12212211gLLg (10) The wave incident from the Generator can be seen empirically as the factors of the difference of the wave reflected between the generator and port 1 (1-g), times the factor of the difference of the wave reflected between the load and port 2 (1-L), minus the term of the forward and reverse gains multiplied by the generator and load reflection coefficients. Again, all of these variables have vector (complex) values. 6 6 AN-GT101A Microwave Power Amplifier Fundamental Giga-tronics Incorporated. All Rights Reserved. A Giga-tronics Technical White Paper AN-GT101A 2008 The point of this analysis is to demonstrate that while conceptually simple, gain at microwave frequencies can be somewhat complicated to determine precisely. Nevertheless, with a few assumptions, the expression for the transducer gain can be greatly simplified. Assume that the generator and load are perfectly matched to the system characteristic impedance, Z0. This implies: 0= Lg , therefore equation (10) simplifies to: 2 SG = 21T (11) This can be expressed in decibels as: () 21T ) log20SdBG= (12) Implicit in all of these expressions is the assumption that the system is linear, or put another way, that signals are small signals. Later, in the section on measurement techniques, more will be discussed about the above simplifying assumptions. MMIC amplifiers with a distributed topology, have a very broadband, flat gain response. Since the transitions from antipodal finline to coaxial waveguide have low dispersion, the inherent gain flatness of the MMIC devices is undisturbed in the spatially combined amplifier topology. Typically, this could only be achieved over narrow bandwidths with classic reactive matching techniques, such as those used for internally matched devices. Attempts to broaden the gain-bandwidth of a high power microwave amplifier required trade-offs with resistive matching, or feedback techniques that rob power output. The spatially combined topology overcomes these limitations. Power Output: P-1dB and Psat Among the key specifications for microwave amplifiers are their power output specifications. Depending on the application, power output is typically specified at one decibel of compression and/or at saturation. Unlike the gain specification, implicitly it is assumed that the specification is at an operating point where the amplifier is exhibiting some degree of non-linear behavior. Depending on the degree of non-linear behavior, and the type of results desired, deriving a model for power output as a function of input power may require very sophisticated CAE tools. Nevertheless, a useful formula for estimating power output as a function of power input that only requires relatively simple calculations that can be implemented in a spreadsheet can be expressed as follows4: Assume: (a) Pin-Pout curve is linear with small signals, and (b) The curve compresses quickly to approach Psat asymptotically. ( sat inT inT out P PG PG P + = 1 (13) 7 7 AN-GT101A Microwave Power Amplifier Fundamental Giga-tronics Incorporated. All Rights Reserved. A Giga-tronics Technical White Paper AN-GT101A 2008 where: = sat PG P exp inT (14) = satdB P Pk P Pk 211 ln dBsat1 2 (15) 258925412. 1=k (16) 1 2056717653. 0=k (17) When using these formulas for calculating Pout there are some caveats to take into account. First, the P-1dB and Psat levels need to be realistic. Generally, to achieve good convergence if these equations are used for numerically calculating the gain compression of a chain of components, the P-1dB and Psat numbers should be within 3 dB of each other. The input power cannot be greatly above (or below) the input P-1dB power level, as many numerical solvers will overflow under those conditions. For those interested, more discussion about the constants, k1 and k2, can be found in Appendix A. With an inherently broadband amplifier, power output as a function of power input does not vary discontinuously as a function of frequency. Typically, a wideband microwave power amplifier that could deliver in excess of several Watts required a solution where numerous narrowband