Geniş bandlı empedans uydurucu devre tasarımı için genişletilmiş frekans bandı yaklaşımı ile çözüm

Abstract

Birçok mikrodalga devresinde olduğu gibi geniş bandlı kuvvetlendirici tasarımında da, empedans uydurma sorunu en önemli konuların başında gelmektedir. Literatürde, bu sorunun çözümü için önerilen pek çok yöntem yer almaktadır. Bu çalışmada, geniş bandlı empedans uyumlaştırması ve tek katlı mikrodalga kuvvetlendiricisi tasarımı yapan, reel frekans tekniğinin optimizasyonuna dayalı olarak geliştirilen bir yöntem ile, buna bağlı olarak yazılan bilgisayar programı açıklanmıştır. Çalışmada, reel frekans tekniğinin düzenlenmesi ile gerçekleştirilen "geniş bandlı empedans uydurucu ve kuvvetlendirici tasarımı için genişletilmiş frekans bandı yaklaşımı" yöntemine ilişkin teorik bilgi verilmiş ve geliştirilen bilgisayar programına ilişkin algoritmayla, programın kullanımına ilişkin detaylar anlatılmıştır. Son bölümde, programın kullanılmasını açıklayan pratik örnekler yer almıştır. Çalışmalarda kullanılan bazı kavram ve tanımlamalar İse "Ekler bölümünde bulunmaktadır.

The impedance matching requirements related with the microwave electronic devices become a major problem in the design of stable wide-band amplifiers. In this thesis, a new approach to the problem is introduced and a CAD optimization programme that is written in GW-BASIC and Fortran Languages is given. In general, impedance matching problem occurs when transfering power from one circuit to another. For instance, power transfering between a high frequency transmitter and an antenna will be possible by using a network that matches the output impedance of the transmitter to the input impedance of the antenna Similarly, in one or multi-stage microwave amplifier design, an impedance matching network is necessary in transfering power from a signal source to an active device and from the active device to a load. The load and source impedances may be real or complex. If both of them are real, maximum power transfer is achieved by using filters, (see Fig. 1) AA/V Lossl ess two-port Rl Fig. 1: Filter problem. This situation is mentioned in references [1 > 4], -v- If the source impedance is real but the load impedance te complex (see Fig. 2), this case is called as "Single Matching" and analytical or computer aided solutions are possible which are given in references [5-II]. © Lossl ess t wo -port Zl Fig. 2: Single Matching problem If both of the" source and load impedances are complex, the problem is called as "Double Matching". Again, analytical or computer aided solutions are possible and given in references [12 - 19]» -VI Fig. 3: Double matching problem. In this thesis, 'Single Matching" case in amplifier design is taken into consideration. Besides impedance matching requirements, stability and fiat gain requirements have also the key importance. Especially in broad-band design applications, stability and flat gaftı conditions together are rather difficult to be satisfied. Several approaches to ensuring amplifier stability and flat gain conditions have been reported in the literature. First, a resistor can be connected between the output terminal of the device and ground, so that the device and the resistor together become unconditionally stable {5]. However, since the resistor dissipates signal power, gain reduction results. Second, unstable regions of source and load reflection coefficients can be marked out on a smith chart The elements of input and output matching networks are then selected graphically in order to transform the mostly 50 Û source and load resistances to impedances in the stable regions to obtain power gain. This is a time - consuming trial-error process [2]. Third, the balanced amplifier technique [5] provides a larger dynamic range and a high stability factor, its disadvantages are that the size is larger and the power dissipation is doubled because of requiring two quadrature couplers plus two transistors. Another way of ensuring stability is to select load and source admittances, Y^ and Ye respectively, based on Stern's Formula [6]. This method is limited to narrow - band applications, because the admittance values, Y^Qop and Yşûo»/) for instance, calculated from transistor y parameters at various sample frequencies, Q: in the passband do not always represent a physically realizable admittance function. Besides, the magnitude of the chief gain-determining parameter, y^i. decreases as the frequency increases, so even if Y^(j), the real parts of Y^Qo) and YgQu) respectively, are obtained simultaneously [9J. In addition, for alignability considerations, the associated Stern K factors at all frequencies are greater than a prescribed minimum [6], and the transducer power gain approaches the specified shape by optimization. input matching circuit Rs ; output watching circuit Ys Yi Yo Yl Fig. 4: An amplifier with input and output port matching network. -vxix - The method described here is based on the real frequency technique, but takes gain flatness, amplifier stability and matching circuit realizability into account simultaneously. In addition, only one set of resistive excursions is necessary for synthesizing the input and output matching circuits. They are optimized concurrently, not sequentially; this simpfifies the design procedure and removes the possible gain limitation imposed by the resistor initially terminated at the transistor output port so that the input matching desigg can be started. In this methot, input port is assumed to be matched which means Yc = Y; in Rg.4 and can be defined by means of the set of resistive excursions belonged to me output port This also simplifies the iterations in optimization. Regardless of the stability and realizability constraints, the method has been formulated as an unconstrained problem. Computer programs which are developed as a design tool of microwave circuits have also the key importance in industry. To meet the needs of industry, the analytic gain-bandwidth theory had been employed by several authors to make it applicable to daily problems vis-a vis the explicit design formulas. In real world problems, when the analytic theory is utilized, first the measured data obtained from the physical devices to be matched have to be modelled. Especially for the wide-band problems, in order to end up with a good fit, the models become very complicated with several reactive elements. In the second step of the analytic approach, a proper transfer function which inherently includes the circuit topologies of the models, must be chosen with unknown parameters, in the literature, monoton Butterworth or Chebyshev response functions are commonly employed. These functions, however, are only appropriate to handle simple matching problems where the generator and the load networks include a few reactive elements as LC ladders. In this case, if the models are more complicated, the theory is inaccessible. In the third step, the unknown parameters of the transfer function is determined to satisfy the gain- bandwidth restrictions. It is wed known that each reactive element, present in the generator or load network drastically penalize the gain response of the matched structure. Thus, the analytic theory of broadband matching leads to suboptimal, unnecessarily complicated, and sometimes practically unrealizable equalizer structures. Based on the above discussion, use of CAD prosedures is inevitable from the practical point of view. Over the last decade, computer-aided matching techniques have been developed to handle single matching as well as double matching problems. Computer packages which cover a wide range of microwave engineering problems, are quite useful for engineers with sophisticated analysis and optimization capabilities, but they do not actually design or synthesize microwave networks. Rather, the circuit topology is supplied to the program with a good initial guess on the element values. Such kinds of packages employ brute force methods and mey suffer from certain deficiencies. Modern CAD techniques, the so-called real frequency techniques, which simply bypass the analytic theory, were qualitatively shown to offer superior design performance, with simpler equalizer structures over the other available techniques. In the following sections, the programme developed due to the new technique is listed with its algorithm and some outputs are given. Also, several amplifier design examples are presented.