Tunable Band-Pass Filters with a Wide Tuning Range and Minimum Number of Tuning Elements

Abstract

Tunable filters play a vital role in modern communication systems by offering flexibility and adaptability that fixed-frequency filters cannot provide. These filters can dynamically adjust their center frequency and bandwidth in real time, making them ideal for use in wireless networks, satellites, defense systems, and RF testing. In wireless communication, they help reduce hardware complexity and cost by replacing multiple fixed-frequency filters and switches. In satellite payloads, tunability enables flexible, multimode operations and reduces the overall weight of the system. In military applications, tunable filters are critical for secure and adaptive communication, allowing for rapid frequency changes to avoid jamming or interception. Their versatility also extends to RF interference cancellation and use in other microwave components like tunable phase shifters and matching networks. In recent years, research has focused on developing tunable filters that use fewer tuning elements to simplify the tuning process, reduce cost, and minimize physical size. A key challenge in this effort is achieving a wide tuning range without compromising RF performance—particularly bandwidth, return loss, and the location of transmission zeros. This challenge becomes even more critical in compact and high-frequency applications, where space and integration constraints are more severe. Traditionally, an all-pole filter of order N with no transmission zeros requires 2N +1 independent tuning elements to achieve constant absolute bandwidth. This includes N elements for tuning the resonators and N + 1 elements for tuning the input/output and inter-resonator couplings. To reduce complexity, one major step forward was limiting tuning to the resonators only, which cut the total number of tuning elements by more than half. After that, further simplification was achieved by using a single tuning element to control the entire filter. When tuning is applied only to the resonators, the achievable tuning ratio (= fmax/fmin) is typically less than 2.3. In designs that use only a single tuning element for the entire filter, the tuning ratio is often further limited to below 1.4. This thesis presents several novel configurations of bandpass filters that employ the minimum number of tuning elements without compromising RF performance, as well as a novel technique for collecting data points to expand the validity range of a trained Neural Network Model. A helical resonator band pass filter is optimized for an almost stable bandwidth across a wide tuning range. The effects of tuning element placement and adjacent resonator orientation are examined, and operating in a higher-order mode is shown to widen the tuning span. Using BST varactors, a fabricated three-pole prototype demonstrates an almost stable bandwidth and acceptable insertion loss over the tuning range. A tunable filter using mode switching is presented, employing a single tuning element to switch between λ/4 and λ/2 resonant modes within a single cavity. By rotating an internal fin, the filter tunes continuously from 1.9 to 3.8 GHz, achieving a tuning ratio of 2 while maintaining a high unloaded Q-factor. Optimized coupling structures ensure stable bandwidth and group delay across the entire tuning range. The design demonstrates smooth and continuous tuning with either manual or motor-driven control. A dual-fin tunable filter is introduced, utilizing two λ/4 resonators of different lengths within a single cavity to cover two adjacent sub-bands using one tuning element. The housing geometry ensures that only one fin is active at a time, enabling a continuous tuning range from 1.95 to 3.5 GHz. Carefully designed irises and probes provide stable coupling and performance across both sub-bands. This approach achieves wide tunability with minimal mechanical complexity. A dual-fin tunable dual-band filter is proposed, enabling simultaneous tuning of two adjacent frequency bands using a shared resonator structure. The resonator supports two modes, each controlled by capacitive loading through cavity shape adjustments. Different tuning patterns are achieved by modifying the cavity slope, offering flexibility in how each band shifts. This configuration maintains sharp band separation and stable bandwidth using a single tuning element. A dual-band tunable filter with independently controlled bands is presented, employing two λ/4 resonators mounted on separate rods within the same cavity. Each resonator is tuned by rotating its fin independently, allowing one band to remain fixed while the other is adjusted. Inter-resonator and input/output coupling structures are carefully optimized to ensure consistent performance and minimal interaction between bands. The design supports automated tuning and enables flexible, independent control over both frequency bands. A novel technique for extending the range of neural network models of microwave structures is presented. This method uses a double-mapping approach to expand the NNM’s accuracy beyond its original training range without needing many new EM simulations. It is applied to a 4-pole dielectric resonator filter and shows accurate results over a wider range.