Lluís Acosta

This work proposes a synthesis method for high-order acoustic wave (AW) multiplexers with any manifold configuration, focusing on enhancing numerical stability while minimizing design execution times. The synthesis method is crucial for efficiently analyzing multiplexer floor plan in a short time (up to 90 possible configurations for herringbone hexaplexers), enabling rapid exploration of design alternatives and optimizing performance for complex RF module applications. The proposed approach reformulates the synthesis process towork primarily with polynomial roots rather than coefficients, leveraging the Durand-Kerner (DK) algorithm for efficient root computation. Additionally, a strategy to map polynomial roots to a less compressed domain is introduced, reducing numerical noise. The method involves computing the polynomial representation of a multiport network with fully canonical channels and deriving a set of K distorted channel polynomials, from which each AW ladder filter is synthesized independently. This methodology ensures the desired multiplexer response without channel loading degradation, allowing independent adjustment of return loss (RL) and transmission zeros (TZs) for each channel. It facilitates the design of multiplexers that meet the technological requirements of AW technology, including electromechanical coupling coefficient (k² _eff) and area efficiency, among other parameters. This approach avoids intensive optimization, representing a useful and controlled technique for designing advanced multiplexer filter modules.

The synthesis of nonconventional topologies has become the key enabler of next-generation filters. However, the coupling matrix (CM) reconfiguration path of noncanonical forms may become complex and difficult to trace, especially in high-order filters. As a consequence, the synthesis of this kind of networks is usually tackled by means of optimization and numerical interval methods. Although these synthesis methodologies have proven to be effective, for some cases they may take large amounts of time to get to the target topology. This work presents a methodology to extract the reconfiguration path of any target topology once the general transformation matrix is obtained, which is reproducible independently of the transfer function. This enables the option of using the powerful and rapid tool of CM reconfiguration. Moreover, this letter discusses the nonunique solution of a target topology with their reconfiguration path, which is strongly related to the eigenvalues of the network. Two different topologies are used as a vehicle to validate the presented methodology.

Microwave-acoustic ladder filter devices are enabled by piezoelectricity, but achievable filter responses are harshly constrained by material properties and complex resonator geometries. While a response depicting an equiripple return loss (RL) is known to provide the highest out-of-band (OoB) rejection, such response might not be feasible after considering material-related constraints, due to footprint limitations or to an excess of helper reactive elements to be implemented in the laminate. This article explores how the ladder filter structure composed of acoustic wave (AW) resonators can trade the location of reflection zeros (RZs) of the transfer function in exchange for greater flexibility to accommodate technological constraints, such as the electromechanical coupling coefficient. This work proposes to move away from the classical generalized Chebyshev (GC) response toward a solution space defined by nonstandard transmission responses. For this purpose, two methods based on a rigorous and accurate filter synthesis methodology are proposed to deal with the arrangement of RZs. The first proposes detuning the RZs along imaginary axis producing a non-equiripple (NE) response, while the second proposes arranging the RZs in the complex plane. This arrangement on the complex plane is carried out in such a way the device shows an equiripple RL but depicting shallow RZs in comparison with the deep ones of the Chebyshev response. Two synthesis examples validate the proposal reaching filters where all resonators require a common electromechanical coupling coefficient, while the ideal equiripple response would require resonators made of different materials, and so, the use of external reactive elements. The solution that in which RZs are arranged into the complex plane depicts an OoB rejection close to that achievable with an ideal equiripple response.

This work assesses three different approaches to the design of dual-band filters with single-ended input and output based on acoustic wave (AW) technology together with the filter synthesis methodology behind each structure. First, the straightforward option, already on the market, is to connect two standalone ladder filters in parallel. This work provides a synthesis view to this option and also describes its strong limitations regarding achievable isolation. Nonetheless, departing from the classical ladder, an innovative topology is proposed in which series and shunt resonators of different sub-bands are concatenated forming a double-ladder topology. Although this latter structure is able to achieve better isolation levels, this work describes the particular constraints arising from this innovative solution in terms of response and technology requirements. Finally, the conventional ladder structure with a single signal path is proposed as a strong candidate to accommodate dual-band filtering responses for the first time. All three structures are discussed from a filter synthesis perspective, describing the particular constraints behind each topology. Also, each structure is reviewed and compared in technological terms to highlight its main advantages and disadvantages.

This work proposes the first analytical method for the direct synthesis of multiplexers based on acoustic wave (AW) ladder filters connected in a star junction topology. It describes the computation of the polynomial representation of a multiport ( ${K}$ $+$ 1 ports) network with fully canonical (FC) channels and derives a set of ${K}$ distorted channel polynomials from which each of the AW ladder filters is synthesized. The channel polynomials are distorted in a way such that when connected together at the star junction, the desired multiplexer response is obtained avoiding degradation due to channel loading. The method allows to freely set the fractional bandwidth (FBW), return loss (RL), and transmission zeros (TZs) of each channel enabling the search for multiplexers that fulfill the technological requirements of AW technology in terms of electromechanical coupling coefficient and area occupancy among others. Moreover, since the multiport network polynomials incorporate the complete effects of the loading between channels as, for example, additional TZs, the synthesized multiplexers depict an improvement of isolation when compared with a design based on only controlling the reflection phase of each channel filter. The proposed method is rigorously demonstrated by means of a synthesized multiplexer example providing the Butterworth–Van Dyke (BVD) elements of each channel filter. This approach avoids intensive optimization procedures and represents a useful technique for the design of carrier aggregation (CA) filter modules in a direct and controlled manner.

This letter provides a new mapping function which allows us to carry out the dual extracted-pole synthesis in inline fully canonical dual-band filters for the first time. To do this, it is a necessary condition that the reflection coefficient phase takes exactly the same value at the position of the two prescribed transmission zeros (TZs), one belonging to one band and the other to the second band, involved at simultaneous extraction. This is achieved by applying an antisymmetric or odd symmetry mapping function. Two different filters, one with asymmetric and the other with symmetric prescribed TZs at each of the passbands have been synthesized which validates the proposed approach.