Efficient Time-Domain Tuning of Microwave Filters Using Concepts From the Unscented Kalman Filter

• Published in: IEEE transactions on microwave theory and techniques Vol. 72; no. 4; pp. 2391 - 2413
• Main Authors: Kenney, Russell H., Jarvis, Rachel E., Sigmarsson, Hjalti H., McDaniel, Jay W.
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.04.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Analogue filters; Design specifications; Estimation; Filter design (mathematics); filter tuning; Filtering algorithms; Filtering theory; Kalman filters; Microwave filters; Resonator filters; Resonators; Time domain analysis; time-domain tuning; Tuning; unscented transform (UT)
• ISSN: 0018-9480; 1557-9670
• DOI: 10.1109/TMTT.2023.3314115

The procedure for designing analog filters is well-established, and the initial design of microwave filters can be performed quickly. However, due to the ease with which unmodeled effects can be introduced to designs at high frequencies, the initial design of a particular filter will usually not meet the design specification, often by a large margin. In these cases, the filter design must be tuned. This work presents a technique for tuning microwave filters using an estimation algorithm called the unscented Kalman filter (UKF). The technique produces a set of design variations to simulate in parallel, producing results from which a tuned design can be extracted. This approach is more efficient than traditional time-domain tuning because it reduces the number of tuning iterations and relaxes the requirement on the designer's intuition in the tuning procedure. An overview of time-domain tuning and the Kalman filter is provided. The filter tuning algorithm is presented, and design results are provided for two different interdigital filters. Finally, the proposed technique is applied to a third-order microstrip hairpin filter, which is simulated, fabricated, and measured. These design cases demonstrate that the technique can correct errors for even severely detuned filters in a small number of iterations.