Portal de Periódicos da CAPES

Fully dielectric interference‐based SPDT with liquid crystal phase shifters

2017; Institution of Engineering and Technology; Volume: 12; Issue: 6 Linguagem: Inglês

10.1049/iet-map.2017.0695

1751-8733

Matthias Jost, Roland Reese, Matthias Nickel, Holger Maune, Rolf Jakoby,

Photonic and Optical Devices

IET Microwaves, Antennas & PropagationVolume 12, Issue 6 p. 850-857 Research ArticleFree Access Fully dielectric interference-based SPDT with liquid crystal phase shifters Matthias Jost, Corresponding Author Matthias Jost jost@imp.tu-darmstadt.de Institute of Microwave Engineering and Photonics, Technische Universitaet Darmstadt, Darmstadt, 64283 GermanySearch for more papers by this authorRoland Reese, Roland Reese Institute of Microwave Engineering and Photonics, Technische Universitaet Darmstadt, Darmstadt, 64283 GermanySearch for more papers by this authorMatthias Nickel, Matthias Nickel Institute of Microwave Engineering and Photonics, Technische Universitaet Darmstadt, Darmstadt, 64283 GermanySearch for more papers by this authorHolger Maune, Holger Maune Institute of Microwave Engineering and Photonics, Technische Universitaet Darmstadt, Darmstadt, 64283 GermanySearch for more papers by this authorRolf Jakoby, Rolf Jakoby Institute of Microwave Engineering and Photonics, Technische Universitaet Darmstadt, Darmstadt, 64283 GermanySearch for more papers by this author Matthias Jost, Corresponding Author Matthias Jost jost@imp.tu-darmstadt.de Institute of Microwave Engineering and Photonics, Technische Universitaet Darmstadt, Darmstadt, 64283 GermanySearch for more papers by this authorRoland Reese, Roland Reese Institute of Microwave Engineering and Photonics, Technische Universitaet Darmstadt, Darmstadt, 64283 GermanySearch for more papers by this authorMatthias Nickel, Matthias Nickel Institute of Microwave Engineering and Photonics, Technische Universitaet Darmstadt, Darmstadt, 64283 GermanySearch for more papers by this authorHolger Maune, Holger Maune Institute of Microwave Engineering and Photonics, Technische Universitaet Darmstadt, Darmstadt, 64283 GermanySearch for more papers by this authorRolf Jakoby, Rolf Jakoby Institute of Microwave Engineering and Photonics, Technische Universitaet Darmstadt, Darmstadt, 64283 GermanySearch for more papers by this author First published: 28 March 2018 https://doi.org/10.1049/iet-map.2017.0695Citations: 4AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract This study presents the first continuously tuneable, interference-based W-band single-pole double-throw (SPDT), fully implemented in subwavelength (sWL) dielectric fibre topology. Key components are the phase shifters in the two branches of the SPDT, which are realised as line sections filled with liquid crystal. They allow not only to adjust the signal ratios at the two output ports but also to tune the operating frequency between 93 and 110 GHz with isolation >25 dB. The first proof-of-concept SPDT demonstrator is realised and characterised at W-band. However, concept and technology are scalable to higher frequencies. The complete SPDT with the WR10-to-sWL fibre transitions exhibits an insertion loss (IL) at the thru-port between 5 and 7 dB in the frequency range from 85 to 110 GHz, where back-to-back measurements of the transitions itself already indicate an IL of up to 1.6 dB. The measurements at the isolated port of the SPDT exhibit transmission coefficients of less than −25 dB between 93 and 110 GHz with a minimum of −40 dB. 1 Introduction Reconfigurable microwave components play a crucial role in modern communication systems in order to meet the ever-increasing demands in terms of functionality and flexibility of the systems. While mainly parallel architectures with individual components optimised for the individual frequency bands and standards have commonly been used, such approaches are not applicable or can only be used to a very limited extent in future transceiver architectures. Not only tuneable phase shifters and filters for feeding networks provide a high flexibility in application [[1]], but also tuneable single-pole double-throws (SPDTs). Up to now, the most common technologies for state-of-the-art SPDTs are semiconductor [[2]] and micro-electro-mechanical system (MEMS) [[3]] technology, where combinations of λ/4-transformer and switchable ground-connections provide a low loss switching with a high isolation to the unused ports. Nevertheless, with these technologies only a switching from on- to off-state and vice versa is possible. Further drawbacks are the power consumption of semiconductor as well as mechanically moving parts for MEMS devices. With the help of continuously tuneable phase shifters, a new type of SPDT is introduced, being based on the superposition of already split-up and purposely delayed signals. With such tuneable interference-based SPDTs it is not only possible to tune the centre frequency, making it operate in a wide frequency range, but also to adjust every power ratio at the two outputs between the lowest insertion loss (IL) (on-state) and highest isolation (off-state). Therefore, the proposed SPDT can be used for many different applications beside the standard switching application. For example, if one output is terminated by a matched load, this circuit can be used as a variable attenuator. Further, it can be used in a radiometer, where both outputs are connected to detectors, one having a high power limit but a low sensitivity and the other having a low power limit but a high sensitivity. While the first is suitable for continuous high power detection, the second is able to detect already small or single signals. To be sure not to harm the sensitive detector, a continuous tuning is beneficial. The key components of the interference-based SPDT are the continuously tuneable phase shifters, which have been realised in liquid crystal (LC) technology to make them electrically tuneable. LC is a very promising material for tuneable components, since its dielectric losses are decreasing with increasing frequency in the lower GHz range, and from a certain frequency on, they stay nearly constant up to the THz regime [[4]]. Furthermore, they can be electrically oriented, thus, no mechanically moving parts are needed, reducing the possibility of wearout failures to a minimum [[5]]. However, with increasing frequency, the implementation of biasing systems is getting very challenging, especially for hollow waveguides [[6], [7]]. To overcome these challenges, the well-known dielectric waveguide technology can be adapted for tuneable millimetre wave devices [[8]]. No metallic walls are limiting the area of usage, keeping the implementation of a biasing system simple. In [[9]], the first comparison between two static, non-tuneable SPDTs, one in the hollow waveguide and one in dielectric fibre technology has already been presented by our group. For this, well-known components of the hollow waveguide technology were adapted to the dielectric fibre technology, such as the Y-shaped power divider and the coupled line combiner. Together with simple WR10-to-dielectric fibre transitions and a fixed air-tapered phase shifter, this static SPDT demonstrator could not sufficiently prove the concept of an interference-based SPDT in terms of an adequate discrimination between the on- and off-state. Therefore, for the very first time, a dynamically, fully dielectric SPDT based on LC technology has been investigated and presented in this paper. It is realised in a subwavelength (sWL) dielectric fibre topology to provide a low loss propagation. For a proposed application in a radiometer, the first proof-of-concept of the complete SPDT has been carried out in the W-band [[10]]. However, concept and technology are directly adaptable to higher frequencies, where – despite fabrication tolerances – this tuneable SPDT is feasible. To assess the feasibility of the continuously tuneable SPDT, tuneable dielectric line phase shifters based on LC technology have already been pre-investigated and published in [[11]]. However, its performance was significantly below our expectations. Moreover, different power divider topologies have been pre-investigated in [[12]]. As a result, the multimode power divider, being based on the self-imaging principle, has been pointed out as the best alternative for power division with dielectric fibres. In Section 2, the basic concept of the interference-based SPDT is presented, before a short introduction of LC technology will be given in Section 3. The design and measurement results of the new WR10-to-sWL fibre transitions, as well as the improved phase shifter, are given in Section 4. Finally, Section 5 presents the design and measurement results of the proof-of-concept interference-based SPDT demonstrator. 2 Basic concept of interference-based SPDTs A schematic of the basic principle of an SPDT is shown in Fig. 1. The incoming signal is split into two equal parts, being in phase to each other (1) Fig. 1Open in figure viewerPowerPoint Basic principle of an SPDT based on tuneable phase shifters. By varying the differential phase shift between the two branches, the interference of both split-up signals at the output ports can be continuously selected. The lengths of the power divider branches, the separated paths including the phase shifters as well as the combiner are given by , and , respectively. At the bottom, the on- and off-states are depicted Each of the upcoming paths includes a phase shifter for differential phase shift. Taking this into account as well as the total length of the two branches , the two signals can be described up to the combiner section by the following equations: (2) The signals are superimposed afterwards in the combiner section and recombine to following output signals: (3) Therefore, the difference in phase of the two split-up signals is introduced just by the transformation within the combiner as well as the differential phase shift applied by the phase shifters. If the phase shifters are tuned equally, e.g. , the SPDT will ideally provide two identical output signals with a 3 dB power split. If one of the phase shifters is tuned to differential phase shift compared to the other one, the interference conditions in the combiner section will lead to a constructive interference of both split-up signals at one and a destructive interference at the other output. Assuming , , , , and , as shown in Fig. 1, the interference scenarios at the outputs result in (4) and (5) This is one of two typically desired configurations for the on- and off-state. The other is given vice versa by , , leading to and (see Fig. 1). The advantage of using LC-based phase shifters in this configuration is that the phase shift, and therefore, the possible signal ratios at the outputs can be adjusted continuously. 3 Fundamentals of LC The name ‘liquid crystal’ is a combination of the main properties of this material, since it is in a mesophase between a solid crystal and an isotropic liquid. In this mesophase, the nematic mixtures show an orientational order but lack translational periodicity, resulting in an anisotropic behaviour. The main direction of the molecules inside a unit volume element can be described with the help of the director , a macroscopic unit vector. Depending on the orientation of the directors compared to the applied RF E-field, the dielectric properties of the materials can be continuously varied between the two extreme states perpendicular ( and ) and parallel ( and ). The LC directors can be aligned by means of quasi electro-/magneto-static fields, by which the directors will be aligned parallel to the biasing field. The magnetic biasing is only used for the proof-of-concept of first lab demonstrators, whereas final demonstrators will only be electrically biased. This has the advantage that no mechanically moving parts or high currents for generating a variable magnetic field are needed for the final application. In this work, the high performance LC mixture GT5-26001 from Merck KGaA, Darmstadt, Germany, was used. The dielectric properties of this mixture are given in Table 1. Table 1. Dielectric properties of the LC mixture GT5-26001 measured at room temperature and 19 GHz (source: Merck KGaA) Mixture GT5-26001 3.27 2.2 × 10−3 2.39 7.0 × 10−3 4 Dielectric waveguide components To keep the dielectric loss at a low level, an sWL topology has been chosen. In comparison to the commonly known classical dielectric waveguide with core and cladding material, the sWL fibre only consists of one dielectric material with a diameter of around , where the evanescent fields are propagating in the air outside the fibre (see Fig. 2, bottom). Due to this maximum difference in permittivity, the field is confined better and the evanescent fields are nearly facing no dielectric loss. Compared to [[9], [11]], the fibre has a cross-section of . The fundamental mode is a hybrid, rotationally symmetrical mode (see Fig. 2 bottom). Fig. 2Open in figure viewerPowerPoint Top: field distributions for (left) an abrupt transition and (right) a transition including a horn-like structure. By the horn, the evanescent field components can be led smoothly inside and outside the metallic waveguide. Bottom: cross-section of the sWL fibre including (left) the field distribution of the fundamental mode and (right) the corresponding intensity plot As a dielectric fibre material, the hard plastic material Rexolite 1422 from C-Lec Plastics, Inc., Philadelphia, USA, with a permittivity of 2.53 and a loss tangent of at 10 GHz was chosen [[13]]. According to the literature [[14]], these values stay nearly constant up to the W-band, which is why these values are used for the simulations in CST Studio Suite. Although a low loss dielectric was found, one must be aware of further loss mechanisms, e.g. radiation caused by discontinuities. Moreover, sharp bends can cause multiple reflections/bouncing of the wave inside the waveguide, resulting in additional radiation, due to loose field confinement. All simulations within this work have been conducted with CST Studio Suite, while the measurements have been carried out with a Keysight PNA, combined with two ANRITSU 3740A-EW transmission-reflection modules. 4.1 WR10-to-sWL fibre transition The first investigations have been focused on the WR10-to-sWL fibre transition. In [[9], [11]], the transitions ended abruptly on the fibre side, causing increased radiation, due to the bending of the E-field around the corners (see Fig. 2 top). The focus in this work had been on horn-like transitions. The optimised design of the transition includes a horn-like extension on the fibre side and the tapering is done continuously for the metal as well as the dielectric. A detailed schematic of the taper sections is given in Fig. 3. Fig. 3Open in figure viewerPowerPoint Top: schematic of the dielectric and metallic tapering inside the WR10-to-sWL fibre transition. The different dielectric sections are marked (A) Triangular tapering to a width of 1.8 mm is shown. While in, (B) Dielectric dimensions stay constant, (C) Height is increasing to 1.8 mm. Further, from (A) to (B) the metallic waveguide is tapered to a width of 1.8 mm, while the height is kept constant. The most important dimensions are , , , , , and . Bottom: Photograph of the realised WR10-to-sWL fibre transitions In section (A), the dielectric shows a triangular taper up to a width of 1.8 mm. The metallic waveguide is starting with the WR10 dimensions and is decreasing in width down to the size of the dielectric fibre from sections (A) to (B), to suppress higher order modes. This results in a completely dielectric filled metallic waveguide with a size of in section (B). In section (C), the height of the dielectric material is increasing, ending up in the final dimensions of the sWL fibre of . By the use of the horn-like structure, the simulated loss caused by radiation could be decreased from 30 to 1%. A photograph of this new transition is presented in Fig. 3 bottom. To our surprise, the simulations cannot determine the radiation losses reliably, as shown in Fig. 4. There, back-to-back measurement results of the WR10-to-sWL fibre transitions are shown for different lengths of the Rexolite fibre. First, only the two taper sections are connected to each other, leading to a distance of 1.5 mm between the horns. Due to the vicinity of the horns radiation is not an issue, leading to a good match of measurement and simulation. Fig. 4Open in figure viewerPowerPoint Back-to-back measurement and simulation results of the WR10-to-sWL fibre transition for different fibre lengths. First, the two taper sections have been connected to each other (taper), where the horns are separated 1.5 mm from each other. In addition, Rexolite segments of 10 and 36 mm length have been connected between the taper The same setup was measured again after including a Rexolite segment of 10 mm length between the taper. This setup is later on referred to as the reference line of the phase shifter introduced in Section 4.2, since it is the same setup as the phase shifter, just that the tuneable segment has been left out. There, the measurement results are 0.5 dB worse than the simulation results. The same trend can be observed for the measurement results with an additional Rexolite segment of 36 mm length, though, having the same length as the phase shifter, and therefore, being 26 mm longer than the reference line. These deviations of 0.5 dB can be attributed to the insufficient representation of the simulated radiation losses of the WR10-to-sWL fibre transitions, due to fabrication tolerances as well as possible air inclusions at the connection points. They are represented in the results as a systematic error, not depending on the fibre length as long as a critical length has been overcome. This is visible in the big difference between the measurement of the taper and the reference line. The distance between the reference line and the 26 mm longer version shows less difference in , although the increase in dielectric length is 2.6 times larger than between the first two measurements. However, it can also be observed, that the IL for the measurements of the reference line as well as the full Rexolite fibre are degrading steeper with increasing frequency as in the simulation. This can be caused by slightly increased dielectric losses as well as not perfectly sealed WR10-to-sWL fibre transition split-blocks, both having a higher impact with increasing frequency. However, since it is not observed for the measurement of only the two tapers, it is assumed that these increased losses at higher frequencies are also originating from radiation. 4.2 LC-based tuneable phase shifter The key component of the continuously tuneable SPDT is the tuneable phase shifter based on LC. As described in Section 3, the effective permittivity of the dielectric waveguide can be continuously changed. For the application in an interference-based SPDT, a differential phase shift of is mandatory. In this work, the phase shifter from [[11]] has been modified according to the points mentioned before. This means the taper section has been adapted to the new metallic transitions and the triangular dielectric tapering. The cross-section has been chosen to be quadratic as well as edge lengths of 1.8 mm. It is still built up in three pieces (excluding taper), with a drilled LC cavity of 26 mm length and 0.6 mm in diameter. The three phase shifter parts are glued together with the help of a special Rexolite glue. The LC filling is done by the use of a syringe. After the phase shifter is filled, it can be sealed with Rexolite pins. For the measurements in the lab, the LC has been oriented either magnetically or electrically (see Fig. 5). For magnetic biasing, a pair of permanent rare-earth magnets () is placed once horizontally and once vertically with respect to the phase shifter. Fig. 5Open in figure viewerPowerPoint Cross-section of the sWL fibre phase shifter including the (top) magnetic and (bottom) electric biasing system consisting of two rare-earth permanent magnets and two pairs of electrodes, respectively. The magnetic field and electric equipotential lines are given for both orientations (left) parallel and (right) perpendicular. The distance between the electrodes is not to scale since the dimensions are , and Two pairs of parallel plates surrounding the fibre are building the electrical biasing system. Maximum biasing voltages of were used. Such high voltages are needed since the biasing electrodes need to be separated sufficiently far from each other to not disturb the evanescent field components. Therefore, the electrodes are mounted on a block of Rohacell (), which is not disturbing the RF signal. The demonstrator with its measurement setup including half of the electrical biasing system as well as the simulation and measurement results of the phase shifter are shown in Figs. 6 and 7, respectively. Fig. 6Open in figure viewerPowerPoint Demonstrator and measurement setup of the electrically-biased sWL fibre phase shifter. For the sake of clarity, only the left and right electrodes for the horizontal orientation are shown. The copper electrodes are mounted on a block of Rohacell to ensure a sufficient distance of the metallic electrodes to the evanescent field Fig. 7Open in figure viewerPowerPoint Top: S-parameter, differential phase shift and FoM simulation and measurement results of the tuneable phase shifter. The measurements are given in solid, the simulations in dashed lines. Bottom: Differential phase shift of the phase shifter, depending on the biasing condition of the electrical biasing system at 100 GHz. On the left (negative voltages), the perpendicular orientation is given, while the parallel orientation is given on the right (positive voltages). Note that the values given on the x-axis are always the absolute difference between the voltages applied to the electrodes The reflection coefficient is matching the simulation very well with a broadband matching better than −10 dB and even −20 dB from 95 to 110 GHz. The transmission coefficients are for both alignment mechanisms around 1 dB lower than simulated with values between −1.5 and −2.5 dB. Using magnetic biasing, is slightly worse compared to electric biasing (see inset in Fig. 7) since magnetic biasing provides a better alignment of the LC than electrical. The differential phase shift for the magnetic biasing matches the simulation well and is even a bit increased compared to it. This is due to a small possible offset in the drilling process, by which the LC cavity length is slightly increased. The differential phase shift for the electrical biasing is less than for the magnetic biasing, being due to a not completely oriented LC. The resulting figure-of-merit (FoM), being defined as the maximum differential phase shift divided by the maximum IL (6) is decreased, due to the increased IL compared to the simulation. The FoM reaches values between for magnetic and for electrical biasing above 100 GHz. Therefore, the performance has been increased by 50% compared to the preliminary design in [[11]], which is mainly due to the new WR10-to-sWL fibre transition design, which is still included in these measurement results. Measurements of a reference line, which is the same waveguide just without the LC section, resulted in transmission coefficients around −1.2 dB at 100 GHz (see Fig. 4). Further, a measurement of a full Rexolite sWL fibre with the same length as the phase shifter, therefore, being 26 mm longer than the reference line, showed an IL of about 1.7 dB at the same frequency. Hence, the influence of the measured transitions is higher as the impact of the dielectric loss of LC. Fig. 7 bottom shows the differential phase shift provided by the phase shifter, depending on the biasing condition of the electrical biasing system at 100 GHz. There, the unbiased state (U = 0 V) is chosen to be the state of of differential phase shift, since this is also a well reproducible state. It can be seen that the main differential phase shift can be achieved even with lower voltages. 10–90% of differential phase shift can already be achieved with in the perpendicular case and for the parallel orientation. In the measurement, a slight change in the linear trend is observed around 0 V, because a certain threshold voltage is needed to produce an electric force, which exceeds the force of the surface anchoring at the LC container. 5 Continuously tuneable SPDT For the realisation of the SPDT the already characterised components, the multimode power divider introduced in [[12]] as well as a multimode interference coupler, being based on the same self-imaging principle are used. The output-to-output isolation of the SPDT demonstrator is mainly depending on the coupler design. Since it is based on a dielectric waveguide topology, the incident wave would rather be radiated than reflected if fabrication tolerances are present. Further, there is no significant impedance mismatch within the design, which could cause reflections. For the provision of a good isolation in metal-based waveguide topologies, one would choose well-known components with very good isolation, e.g. the hybrid coupler for microstrip line topology or the Riblet type coupled line combiner for rectangular waveguide topology. However, a limiting parameter of the multimode interference coupler in terms of transmission is the length of its main body (see Fig. 8). In comparison to the power divider [[12]], the coupler is asymmetrically fed, which is why the main body needs to exceed a certain length until the multimode interference pattern has become well-defined. Therefore, if the main body is designed too short, the preferred interference conditions cannot be met anymore and the waves will radiate. Fig. 8Open in figure viewerPowerPoint Top: Simulation model including the field distribution for one extreme configuration and (middle) photograph of the tuneable SPDT. It has a total length of and width of . Also, the points i, ii and iii from Section 2 are given. Bottom: Biasing system of the proof-of-concept SPDT demonstrator. While the left phase shifter is biased electrically, the right one is kept in a constant orientation by means of permanent magnets In the final demonstrator, the connection points are sticked together with Rexolite register pins. For the measurements, one of the phase shifters was aligned parallel with the help of permanent magnets, while the other one is biased electrically. This is due to the fact that only one suitable voltage source was available in our lab. Thus, an independent tuning of both phase shifters was not possible at the same time. However, this is not limiting the performance of this dielectric proof-of-concept SPDT. The field distribution along the fibre, the final demonstrator as well as the measurement setup is shown in Fig. 8. The two branches needed to be separated 20 mm to ensure the possibility of a magnetic biasing. In future applications, this distance can be reduced for a fully electrically biased system, which would not only decrease the amount of radiation but also the total size of the demonstrator. The SPDT can further be embedded into a matrix made of Rohacell to provide a higher mechanical stability without affecting the RF field. In Fig. 9 left, the results of the best performing biasing configuration of the proof-of-concept SPDT are shown. This means: for the optimum frequency of 98 GHz, the best performing configuration in terms of lowest IL of the thru-port and highest isolation of the isolated port will be achieved for the designed final demonstrator when both phase shifters provide the maximum differential phase shift. This is the case when the LC in one phase shifter is aligned parallel by using magnetic biasing, while it is aligned perpendicularly in the other phase shifter by using a biasing voltage of or vice versa. These extreme cases are needed, since the multimode interference coupler showed a slightly decreased phase difference between the outputs, compared to the from theory, which needed to be compensated by the phase shifters. Fig. 9Open in figure viewerPowerPoint Left: S-parameter simulation and measurement results of (top) the matching and (bottom) the best biasing configuration (signal at port 2, isolation at port 3) of the continuously tuneable dielectric SPDT. The measurements are given in solid, the simulations in dashed lines. Right: Measurement results of (top) the matching and (bottom) the transmission coefficients and for a complete sweep of one phase shifter of the tuneable SPDT. The second phase shifter was constantly kept in parallel orientation by means of permanent magnets The reflection coefficient matches the simulations well with values of −15 dB nearly over the whole frequency range and even below −20 dB above 100 GHz. For the best biasing configuration, the IL of the thru-port is always better than 7 dB, with a minimum of 5 dB between 88 and 95 GHz. This is at lower frequencies in a good agreement with the simulation, while there is a deviation of about 1.0–1.5 dB at higher frequencies. The isolated port in this configuration is better than simulated, being directly linked to the slightly increased loss of the thru-port. It shows a transmission coefficient below −15 dB above 86 GHz, even down to −30 dB around 100 GHz. This results in an isolation between and better than 15 dB