寬頻微波接取技術---子計畫二：多層印刷電路板立體式寬頻微波濾波器與耦合器之開發與製作

Abstract

為配合總計劃寬頻微波通訊系統之要求，本子計畫將研製各項被動元件以滿足系統之需求，例如濾波器、平衡與非平衡轉換器、耦合器等。而重點將放在耦合器與濾波器之研究上。因為總計劃所擬的系統有寬頻、及多頻帶之特性，現今通訊系統之被動元件多半不適用。例如3至5甚至3至7甚或3至10GHz之濾波器便是其中之ㄧ例，而有時也需要可以變頻之濾波器或是超寬頻的180o魔術T岔路環耦合器，這種種都有待此子計劃來研究。 本研究將分三年來進行，第一年將製作多節式超寬頻的180o魔術T岔路環耦合器，第二年將利用多層印刷電路板技術製作四分之ㄧ波長的步階阻抗諧振腔(SIR，Step Impedance Resonator)濾波器，第三年則將研究對第二年提出的濾波器架構利用交錯耦合產生傳輸零點，藉以提高濾波器的選擇性，同時也要針對可調濾波器(Tunable filter)的可行性加以研究。而所有耦合器與濾波器之設計都將用最常見也最適合量產之印刷電路板製程來完成，為了達到特殊耦合器與濾波器之規格要求，將使用垂直安裝平面基板方式(VIP，Vertically Installed Planar circuit)，其特點是能夠製作極高阻抗的傳輸線或極緊耦合的耦合線，同時也是用印刷電路板製程。 第一年將發展一種多節式180o魔術T耦合器(multi-section 180o magic-T coupler)，如所周知多節式分支線90o耦合器(multi-section branch line 90o coupler)可以增加頻寬，但鮮少有人可以將180o魔術T耦合器以多節方式連接以增進頻寬，直到2002年Ang等人才找出了將傳統180o魔術T二個串接的方法，但他們只針對中心頻做解析，對不在中心頻率的響應會如何完全沒有探討，而且使用傳統岔路環式180o魔術T串接，頻寬並未獲得很大改善。基於此，第一年我們將提出一種以我們在1999年提出的理想化的180o魔術T岔路環結構以Ang的方式加以串接，初步模擬可以大大增加頻寬。預期將有二個困難點待解決，第一、因為Ang的串接方式必須將二微帶線交叉，如何將二微帶線交叉但不影響其射頻特性，關係到是否可以實際製作出所擬的耦合器，第二、如何推導出一套設計方程式可以預估串接後在整個頻段(而非只有中心頻)之頻率響應，以利設計者可以很容易的設計出想要的耦合量與響應函數。我們將在第一年的時間內解決這二個問題。 第二年將使用雙面印刷電路板製作四分之ㄧ波長的步階阻抗諧振腔(SIR，Step Impedance Resonator)濾波器，其耦合可以藉助交疊方式或使用一個非諧振點(NRN，Non-Resonant Node)介於二諧振腔之間的方式來達到寬頻之目的。同時也將進一步利用多層印刷電路板技術來將濾波器縮小，使用雙層板則濾波器中之諧振腔是以平行方式做二度空間之排列。使用多層板則有可能將各個諧振腔做三度空間垂直方向之堆疊與排列，如此則濾波器的尺寸將可以大幅縮小，目前多層板之層數已可達到八層，對我們的濾波器設計提供極大的彈性，預計也可以利用非諧振點的方式來控制諧振腔之間的耦合量，達到寬窄皆宜的需求。 第三年則將研究對第二年提出的濾波器架構探討交錯耦合或其他的方式產生有限頻率的傳輸零點的可能三度空間結構，藉以提高濾波器的選擇性，同時也要針對可調濾波器(Tunable filter)的可行性加以研究，因為可調濾波器有強大的性能，在軟體無線電中是一個重要元件，但相對的也非常難設計，尤其是濾波器的Q值將大幅下降是一個待克服的問題。

This subproject provides various passive microwave components such as filters, couplers, and baluns etc. to the main project so that the whole RF transceiver can meet the requirement of the proposed broadband microwave communication system. Among those broadband passive components, the retrace magic-T couplers and filters will be the most complicated and we will concentrate our research on them. Because the proposed broadband microwave communication system has broadband and multi-band characteristics, most of present commercial available passive components can』t fulfill the requirement. The filter with passband bandwidth of 3 to 5GHz or 3 to 7GHz or even 3 to 10GHz, which might be used in the proposed system, is an example. Sometimes a filter with tunable center frequency or an ultra-broadband rat race magic-T coupler might be required. All of these couplers and filters with special specifications will be studied in this project. This research will be done in three years. In the first year, a multi-section ultra-broadband rat race ring magic-T coupler will be studied. In the second year, quarter wave SIR (step impedance resonator) filters will be designed based on multi-layer PCB process. In the third year, reorganize the filters proposed in the second year to generate cross coupling in order to increase the selectivity. The feasibility study of tunable filter will also be done in the third year. The proposed couplers and filters will be built with the most popular and mass-production-suitable PCB (printed circuit board) process. In order to meet the special specifications a VIP(vertically installed planar) circuit will be applied because the VIP circuit can implement extremely tight coupled lines or extremely high impedance lines. Besides, the VIP circuit is also a PCB circuit. In the first year of the project, a multi-section rat race ring 180o magic-T coupler will be studied. As we all know that the bandwidth of a 90o branch line coupler can be increased by a multi-section configuration. However, 180o magic-T coupler can hardly be connected as a multi-section 180o magic-T coupler in the past, until Ang et. al. proposed a configuration in 2002. In Ang』s paper, two conventional rat race ring coupler can be connected in a special manner to form a two section 180o magic-T coupler. Unfortunately, the bandwidth of the Ang』s two section rat race ring coupler wasn』t largely improved because only conventional coupler was used. Besides, the mathematical analysis in Ang』s paper was only based on the center frequency the coupler responses other than center frequency had not been discussed. We propose a multi-section 180o magic-T coupler based on the ideal single section 180o magic-T coupler proposed by us in 1999, where the ideal single section 180o magic-T coupler will be connected using Ang』s configuration. Preliminary study shows that the bandwidth of the proposed multi-section 180o magic-T coupler can be drastically increased. During the research, two problems should be solved. First, there at least exists one microstrip crossover as we use Ang』s configuration. The crossover should not have influence on the performance of the ideal single section 180o magic-T coupler. If fail to develop this crossover, the multi-section 180o magic-T coupler can hardly be implemented. Second, to develop a mathematical analysis algorithm which can predict the responses over the whole passband so that a multi-section 180o magic-T coupler can be easily designed with various frequency responses. In the second year of the project, the quarter-wave SIR (step impedance resonator) filter will be studied based on double-side PCB process. The coupling structures between resonators are preliminarily proposed to be an overlapping structure or coupling through a NRN (non-resonant node) since a very strong coupling is required to implement a filter with enough passband bandwidth. Besides, the research will also be concentrated on the filter size shrinkage by multi-layer PCB structure. In the double-layer filter, the resonators are arranged in two dimensional parallel manner. Using multi-layer PCB structure the resonators can be stacked in three dimensional vertical way, so that the filter dimension can be largely reduced. Recently, the PCB can have as many as eight layers. Therefore, the design flexibility of the proposed three dimensional filter is greatly increased. The coupling through NRN might also be used to control the coupling strength, so that a filter with various of bandwidth can be easily realized. In the third year of the project, the three dimensional cross-coupling structures will be studied to make a filter with various kinds of finite frequency transmission zeros. By properly select the position of transmission zeros the filter selectivity can be improved without increasing the number of the resonator. In the mean time, we will study the feasibility of tunable filter with VIP structure. The tunable filter has the powerful characteristics to make software defined radio possible. It is an important device but the design is much more complicated than that of a fixed filter. Besides, the degradation of the resonator』s Q-value is also an important issue to study.