Eight-element liquid crystal based 32 GHz phased array antenna with improved time response

Phased array radar gives precise, time-based position information [1]. These radars are critical in a variety of applications requiring precise object tracking. One limitation to the use of these radars is the physical size of the arrays [2]. Their large size prevents their use in collision avoidance systems in automobiles and other platforms that could benefit from such a system [3].

The size and spacing of the antenna elements are dictated by the frequency of the radiated energy [4]. Thus, one method to reduce the size of the devices is to increase the operational frequency. As the frequency used in the systems has increased, the components used to control the individual element phase have reached physical limits. MEMs devices become size limited and experience sticking and other failures as the device size is reduced [5–7]. Magnetic phase control components become larger and heavier as frequency increases [8]. Ferroelectric based phase shifters exhibit high losses above 40 GHz [9, 10]. These physical shortcomings limit the useable frequencies, creating a barrier to how small the antenna elements can be.

In certain applications, one solution is the use of liquid crystal (LC) media to control phase in the individual elements [11]. LC is considered transparent to the signal in frequencies ranging from MHz to THz [12, 13]. The permittivity of the LC is easily controlled with a low frequency bias signal and the entire phase controlling LC cell can be produced easily and cheaply with well-established techniques developed for display technology.

We constructed and tested a linear, 8 element phased array antenna of a unique modular design. The LC cell was filled with a proprietary dual frequency liquid crystal (DFLC), 1909C, designed and fabricated at the Military University of Technology (MUT), Poland. We chose to test with a DFLC to improve the time response of the phase shifter assembly. The phase shifter time response is critical in a phased array radar system as the tuning speed and therefore the beam scan rate is directly dependent on this parameter.

A literature search produces some instances of LC based phased array antennas [14–17]. These studied devices typically use a standard nematic liquid crystal as the tuning medium. It is well known that standard nematic liquid crystals have a long relaxation time, typically measured in seconds [18]. The long relaxation time associated with these types of liquid crystals precludes their use in any phased array system requiring a rapid scan rate. Other researchers have investigated other methods to reduce the tuning speed of liquid crystal-based devices. Of particular interest is an optical retarder developed by Golovin et al using a polymer network with a LC media dispersed within the polymer [19]. This design produces very fast response times, the negative impact of this design is a very low phase shift per unit length, meaning a device constructed to operate in the microwave range would have an unreasonably long length resulting in increased metallization losses in the transmission path. In addition, optical frequency devices are thin devices. To create devices in the microwave range, a thicker liquid crystal layer is needed to provide the proper spacing for the microstrip. As it will be discussed in section 3, this increase in thickness results in a slower response time (tuning speed). This is demonstrated by Li and Chu who reported a tuning speed of 4–12 s for a device using 140 micron thick LC [20].

Overall, very few papers discuss the tuning speed or scan rate of the presented array. One paper by Strunck et al [3] describes a tuning speed on the order of seconds and Engstrom et al [21], discussing an electro-optical array based on ferroelectric liquid crystal, report a response time of 140–180 ms. In contrast, we demonstrate a response time of 42 ms for our DFLC phased array system. And our system does not suffer from the size limitations of a polymer network-based LC cell resulting in a compact device suitable for many applications including collision avoidance systems in automobiles, line-of-sight secure communications and industrial robotic guidance systems.

The design and construction of the phased array antenna is described in section 2. In section 3, we discuss phased array theory and document theoretical beam profiles. LC theory is also discussed in this section, specifically how it pertains to the response time of the LC phase control cell and why the use of nematic, dual frequency liquid crystal (DFLC) is preferred in these devices to minimize response time [18].

Section 4 covers the various test setups used to measure the performance of the array. The modular nature of the array allowed us to perform independent testing of the LC phase control cell with a standard vector network analyzer (VNA). We were able to measure the performance of each of the 8 phase control lines independently. With this testing, we verified each line exhibited the proper phase shift and transmission characteristics. The line transmission data was used to determine the optimal operating frequency for our linear array. The final assembly was then tested in an anechoic chamber to measure beam profiles, determine steerability and measure the response time of the array.

In section 5, we present the results from testing our linear array. This includes the individual phase shifter performance, the beam profile and the steerability of the completed array, and array response time and repeatability measurements. Our conclusions are given in section 6.

We designed and constructed a linear, 8 element array. This system was intended as a test platform to demonstrate the improved tuning speed of DFLC as compared to other LC media, demonstrate an acceptable phase shift/unit length in the device, and show a proof-of-concept demonstration of our unique modular approach to developing LC based phased array antenna systems.

We chose an eight-element array as this provides sufficient resolution to demonstrate beam steering. This design is readily adaptable to adding elements and stacking assemblies. 8 × 8 and 16 × 16 element arrays may be produced with this method and different design antenna elements may be used depending on the application.

A schematic of the system is shown in figure 1. The LC media is contained within the area in light green. The areas outside this boundary are pcb. The RF input is a standard RF end launch from Southwest Microwave. The input is split into eight signal paths using cascaded Wilkinson dividers. DC blocks (high pass filters) isolate the eight, discrete low-frequency control signals to the desired signal path. The AC bias used for these control signals is fed into the individual signal paths via low pass LC filters prior to the transition to the liquid crystal region. Tuning of the device is created in the LC region. The dielectric constant of the LC surrounding the individual transmission lines is varied from 2.8 to 3.4 by the applied bias signal producing the desired phase shift of the RF signal across each line.

Zoom In Zoom Out Reset image size

The output of the LC region is fed to the antenna elements. Each of the transmission lines in the LC region is a discrete phase shifter such that the individual antenna elements receive the RF signal with the appropriate phase shift to achieve beam steering. The antenna elements are a forward-firing Vivaldi design.

The glass provides the strength and rigidity necessary to maintain the internal cell spacing required by both the LC and the microstrip waveguide. Traditional photolithographic techniques can be used to define the copper transmission features on the glass. The glass also readily accepts traditional LC alignment methods used to properly control the initial orientation of the LC media, a necessity in maintaining proper phase and time response of the array [22].

The use of PCB solves several problems with the associated electronics typically encountered if one chose to design on-glass components. The microwave glass used in the array is 1.1 mm thick. Our experience is this glass readily cracks and breaks during drilling. This limitation reduces the working surface to one plane. With only one plane, the routing of the phase shifter lines and the associated AC bias lines is difficult requiring bond wires or other complications to successfully implement the design. We discovered bond wires and other such approaches approximated antenna elements for the RF energy necessitating additional design to suppress this stray coupling resulting in increased complexity and an increased number of required components.

With a modular design where the glass LC cell is manufactured separately from the PCB containing the bulk of the circuitry, these complexities were avoided. A multilayer PCB allowed us to work on several planes using industry accepted practices. The PCB utilized microstrip traces on the top surface, low frequency signal paths on the bottom, with a ground plane sandwiched between them. This design allowed us to route the AC bias lines on the bottom and the RF circuit paths on the top. Connections between the two were accomplished by vias passing through the ground plane effectively suppressing stray RF energy coupling to the bias line.

With PCB, we could easily produce transitions between coplanar waveguide (CPW) and microstrip transmission lines. We sourced existing, surface mount components for the required circuitry. The plugs and end launch connectors the design required were easily implemented without worrying about working on a one-dimensional surface, drilling through glass, or bonding to glass. And the board was designed with readily available, commercial software and manufactured by an outside company. The phased array system performance was simulated at various steps of design using both Ansys HFSS and CST Design Studios software [23, 24].

The design parameters listed in Table 1 were used for modeling the PCB section of the device. In this paper, microns refer to micrometers.

The design parameters shown in Table 2 were used for modeling the LC section of the device. The epsilon values were determined for 30 GHz using a method developed by Economou et al [26].

Figure 2 is an electrical schematic of the entire assembly. The overall circuit is designed for 50 Ω impedance at 30 GHz. In the PCB section, the microstrip trace width is 650 microns. In the LC section, the microstrip trace width is 60 microns. Coplanar waveguide (CPW) sections in the PCB section use a gap between center trace and ground planes of 70 microns. Transitions between microstrip and CPW use tapered transformer sections to match impedances.

Zoom In Zoom Out Reset image size

The Wilkinson dividers use ¼ wavelength legs and 2 Z_O lumped resistors. The legs are 325 microns wide giving a √2 Z_O impedance. Figure 3 is a picture of a Wilkinson divider on the actual device. The dividers are cascaded on the actual device in such a way as to maintain symmetry.

Zoom In Zoom Out Reset image size

The AC bias signal to each of the 8 lines is introduced with an LC lowpass filter. In this filter design, a bondwire is used to provide the connection to the RF trace. This bondwire (1.5 mm long) creates approximately 1.5 mH of inductance. An 8.2 nF lumped element capacitor is used to complete the LC filter circuit. This arrangement creates a 45 kHz cutoff frequency. The abrupt impedance change between the RF trace and bond wire and the low cutoff frequency both ensure the RF signal is not coupled back through the AC bias network. Figure 4 is a picture of the AC bias circuit for one RF trace showing the bondwire and capacitor arrangement.

Zoom In Zoom Out Reset image size

The Vivaldi antenna design was chosen to produce a forward firing RF pattern with good control of inter-element interference. The elements were spaced 1/2 wavelength apart to minimize this interference. Dummy elements are used to improve the steerability of the phased array. All element design and the overall design of the array was performed for an RF frequency of 30 GHz. Comparison of simulations and actual device performance will be presented in section 5.

Of special interest is the design of the transition from CPW to microstrip in the LC glass cell. This area created several challenges. The CPW to microstrip transition must be accounted for in this area. In addition, we had to design the transition to accommodate dielectric changes at the air-to-glue and glue-to-LC boundaries. The glue serves to bond the glass substrates together and seal the LC media within the active region of the phase shifter.

Figure 5 shows this region of the glass cell. A tapered transition from the CPW to microstrip regions is used as a transformer to match impedances. This tapered transition is modified by a notch at the interface with the glue. The transition also contains cutouts on the CPW ground planes and the microstrip groundplane (outlined in green) to match impedance across the LC boundary. This arrangement was based on similar designs reported in other publications [26, 27].

Zoom In Zoom Out Reset image size

The final assembly of the glass LC cell and the PCB board used a novel, modular 'flip-chip' design. We call this a flip-chip design because the glass LC phase shifter cell is mounted with the signal path interconnects facing down to mate with the interconnects on the PCB. The two major components of the antenna assembly are the PCB substrate that is constructed from Roger's RO4003^TM laminate and the phase shifter cell made with Schott Borofloat microwave glass substrates. Many of the advantages of this construction method have been discussed above. This method also allowed us to individually measure the LC cell RF transmission paths prior to assembly using a standard VNA and probe station with GSG probes. These measurements, as described in section 5, allowed us to measure the phase shift per unit length as a function of AC bias signal of each RF path independently, thus simplifying the tuning of the phased array assembly in the anechoic chamber.

The Roger's PCB substrate shown in figure 6 has two sections. The input section contains the mount for the microwave end launch, the input bias signal jack mount, Wilkinson power dividers and the circuitry necessary to couple the biasing signals to the microwave signal paths. The output section has the Vivaldi design antennas and the dummy antenna elements used for impedance matching. There is a cutout between the two sections to accept the LC cell. The microwave signal end launch connector mount is fabricated in coplanar waveguide and there is a transition to microstrip geometry for the majority of the signal line. The edges of the liquid crystal cell cutout require a transition back to CPW to provide a connection between the PCB and liquid crystal cell.

Zoom In Zoom Out Reset image size

Figure 7 shows an exploded view of the liquid crystal phase control cell. The liquid crystal cell is constructed of two 1.1 mm thick Schott Borofloat microwave glass substrates. The waveguide structures for the 8 control lines were photolithographically defined on the upper substrate and 2 micron thick copper was deposited with a magnetron sputtering system. A 10 nm layer of polyimide, covering the surface of the substrates and metallization, was uniaxially rubbed to create an alignment layer for the LC. The edges of the cell are CPW to match the connections on the PCB. The CPW also allows the use of a standard probe station to measure individual line performance. The 8 transmission lines transition to inverted microstrip in the active region of the cell containing the LC media [27, 28]. An inverted microstrip is utilized to maximize the effects of changing permittivity in the active region of the LC phase control cell.

Zoom In Zoom Out Reset image size

The lower substrate uses 2 μm thick copper deposited on the substrate glass as a ground plane for the microstrip waveguides. This ground plane is connected to the CPW grounds on the upper substrate by soldering. The two substrates are assembled with NOA 65 optical glue and spaced apart with 22 μm spherical glass spacers. The volume between the two substrates is filled with the LC media. The LC director orientation, and subsequently the permittivity, is controlled by an applied bias field. The voltage that generates this field is applied between the signal line and the ground plane. In this manner, with each signal line biased individually, the phase of each of the 8 lines can be varied individually.

Figure 8 is a model depicting how the glass LC cell and the PCB are joined. The upper glass substrate is extended on the sides to overlap the PCB. This provides additional support and alignment. The critical parameter is to ensure the CPW areas on the cutout of the PCB and the ends of the glass overlap. Properly aligned, the signal trace on the PCB will align with the signal trace of the LC cell.

Zoom In Zoom Out Reset image size

To achieve electric connections between the glass and the PCB, R&D Interconnect Solutions^® conducting elastomer Invisipin^® elements were used. These crushable elements made it possible to clamp the PCB and LC cell together in a 3D printed mount designed for the anechoic chamber. Clamping these elements together allowed us to reuse the PCB for several different cells during the design and testing process. The Invisipin^® elements and the other surface mount components were mounted to the PCB using Amtech NC-31 low temperature solder paste, see figure 9. Three pins were used for each signal path, two on the ground planes and one on the center signal trace. Southwest Microwave 2.4 mm end launch connectors were used to input the microwave signal.

Zoom In Zoom Out Reset image size

The fully assembled phased array antenna assembly is shown in figure 10. The assembly is mounted on a special 3D printed test stand designed to properly position the assembly in the anechoic chamber. The cross bar over the phase shifter cell acts as a clamp to provide contact between the signal paths on the PCB and the phase shifter cell. This arrangement allows us to change out phase shifter cells to test the performance of different types of LC material.

Zoom In Zoom Out Reset image size

In a phased array antenna system, the array is held stationary and the beam is formed by constructive and destructive interference occurring between the waveforms emitted by each antenna element. The beam is steered by varying the phase relation between the elements thereby changing the interference patterns resulting in beam steering. The beam forming and steering capabilities of a linear n element array are given by the following formula which is appropriate in the far field [1]:

$$\begin{eqnarray}&&{A}_{tot}=\displaystyle \sum _{n}^{{\rm{N}}}{A}_{n}{e}^{i\,\left(\phi \left(n\right)+nkdSin\theta \right)}\end{eqnarray} \tag{ 1 }$$

here A_n is the amplitude of the nth element, k is the wave vector, d is the spacing between the elements, φ(n) is the nth element output phase and θ is angle (relative to normal) where the measurement is taken. Our array design was modeled in Wolfram Mathematica [29] to simulate the expected beam profile and profiles at various beam steering angles. These modeled profiles are given in figure 11. Additional modeling was performed in HFSS. These models allowed us to test simulated beam profiles and determine the effects of the dummy elements. A model of the radiation profile of an individual antenna element is shown in figure 12. Two modeled beam profiles are presented in figure 13.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The system was designed to provide a ±45 degree steering range with an 8 degree steering resolution. The system was modeled to achieve an insertion loss of >−20 dB across a frequency range of 30–35 GHz (figure 14).

Zoom In Zoom Out Reset image size

We now examine how small variations in phase change the resulting directed beam. Using equation (1) with a phase distribution of ±1 deg across the elements resulted in a boresight power reduction <5%. When the number of elements is increased to 16, the boresight power reduction is <3%. This demonstrates that the LC phase shifter is a good candidate for a phased array antenna as these small shifts in output phase produce minimal degradation in the output beam. To explore this further, figure 15 shows the theoretically calculated intensity pattern, with and without variations in both the amplitude and phase of the 8 antennas. The patterns are nearly identical indicating that the small transmission variations reported in later sections will not have a significant effect on the performance of this system. This is consistent with the findings of Rebeiz et al who concluded that the failure of 3%-4% of the MEMS phased shifters in a large phased array system does not significantly impact the array performance [6].

Zoom In Zoom Out Reset image size

Of special interest is the response time of the LC as this dictates the scan rate of the phased array as discussed in the introduction. The LC response time (τ) of a standard nematic LC is given by the following set of equations [30].

$$\begin{eqnarray}&&\tau \sim \displaystyle \frac{{\tau }_{relax}}{{\left(\displaystyle \frac{V}{{V}_{th}}\right)}^{2}-1}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{\tau }_{relax}=\displaystyle \frac{{\gamma }_{1}{D}^{2}}{{\pi }^{2}K}\end{eqnarray} \tag{ 3 }$$

where V is the applied bias voltage, V_th is known as the threshold voltage (the minimum voltage required to change the anisotropy of the LC cell) which is typically in the range of 2 to 3 volts, γ₁ is the viscosity of the liquid crystal, K is the elastic constant of the liquid crystal and D is the thickness of the liquid crystal controlled by the spacing between the two substrates.

In a standard nematic LC, the LC orientation is changed from the initial orientation to the desired orientation by increasing the biasing voltage applied to the cell, the time it takes for this orientation change is known as time on (equation (2)). The LC is returned to the original state by removing the bias voltage and allowing the liquid crystal to relax back to the original state. The time associated with this relaxation process is known as time off and is given by equation (3) which shows the relaxation time is proportional to the thickness of the cell squared. In microwave applications, the thickness of the cell is relatively large to satisfy the requirements of the microwave waveguide structure resulting in a long time off [31] as discussed in the introduction.

In a DFLC cell, the change from the original orientation occurs as described above for the standard nematic LC (equation (2)). In contrast to the nematic LC cell, to return the DFLC to its original state, voltage is held constant and the frequency of the bias signal is increased. In this condition, both time on and time off are given by equation (2). There are minor variations in the value of threshold voltage between the two times. However, using a large bias field results in a large denominator (V/V_th) drastically reducing the time off [18]. We chose to use DFLC because of this rapid response time. In figure 16, we present the results of time response measurements of a phase shifter filled with Merck KGaA MLC-2048 DFLC [32] operated in a voltage biasing scheme and in a frequency biasing scheme. In the voltage biasing scheme, the DFLC reacts as a standard nematic LC exhibiting a long relaxation time. In the frequency biasing scheme, the DFLC is maintained in the driven state resulting in rapid time on and time off. These results are consistent with the equations as we clearly see the time on values for both schemes are identical and the time off value for the frequency biasing scheme is reduced by orders of magnitude from the voltage biasing scheme.

Zoom In Zoom Out Reset image size

A second important consideration in the design of a phased array antenna system using LC as the tunable dielectric is the phase change per unit length created by the LC. LC for these and similar applications are designed to maximize the phase change per unit length, maintain transparency to the transmitted signal and minimize response times. The phase change in unit length is maximized when the change in anisotropy, and hence the change in the dielectric constant is maximized. Through maximizing the phase change per unit length, the actual length of the LC cell is reduced, reducing the length of the transmission line in the cell. This is important because, even though the LC is transparent to the RF signal, the relatively thin (2 micron) copper transmission path has high metallization losses created by its physical size, surface roughness, impurities, and other effects.

The use of DFLC in microwave device both with and without alignment layers has recently been discussed [33]. The use of an alignment layer in a standard nematic LC is critical as the alignment layer governs the response of the LC and provides an orientation for the LC to relax to when the voltage is removed. This layer maximizes the phase change per unit length of the cell while minimizing the LC response time. In the referenced paper, it was demonstrated a DFLC can be operated without an alignment layer with only minor degradation in response time and phase change per unit length. It is also shown that, for microstrip transmission with Cu features, a thicker alignment layer is necessary to achieve optimal phase change per unit length when compared to optical applications.

Two test setups were used for testing the phased array and its components. To measure individual transmission line performance, a VNA and standard probe station were used. A block diagram of the setup is shown in figure 17. Waveguide sections were used as high pass filters to isolate the VNA from the AC Bias signal. The AC bias signal was generated with a function generator and amplified with a broad band amplifier. The bias tee serves to combine the AC bias signal and the microwave test signal. LabView was used for instrument control and data collection over a General Purpose Interface Bus [GPIB] control bus.

Zoom In Zoom Out Reset image size

To determine the overall array performance, we used an anechoic chamber with data taken by a VNA. Figure 18 is a cartoon of the anechoic chamber. The receiving microwave horn is mounted on a swing arm that is rotated with a pivot point on the axis of the antennas as shown by the dotted line. The horn is mounted at a distance of 34 inches from the antennas and the center of the horn is on the antenna plane. The chamber is lined with microwave absorbing material to minimize reflections and external microwave signals. The swing arm can be positioned with tenth of degree accuracy in a range of ±50 degrees from the boresight (beam projection normal to the array) position.

Zoom In Zoom Out Reset image size

A block diagram of the arrangement used for testing in the anechoic chamber is shown in figure 19. For this setup, a VNA generates the microwave signal and measures the return signal. A 30 dB low noise amplifier (LNA) amplifies the output of the VNA. This amplified output is applied to the input of the phased array device. A pyramidal microwave horn with a 10 dB gain captures the beam formed by the device. The signal from the horn is amplified by a 20 dB LNA and then applied to the receiving port of the VNA. Figure 20 is a photograph of the actual test stand showing the test equipment and the phased array device mounted in the anechoic chamber.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The bias signals are created by a National Instruments card with 8 analog output channels controlled by LabView. Each channel is independently amplified to a value of 100 V_peak-peak. This arrangement allows the voltage and frequency of each channel to be independently set to achieve proper control of the DFLC for each particular antenna element. The bias signal is then fed to the phased array device through a plug on the PCB board. Figure 21 is a picture of the device mounted for testing. The plug on the left is for the AC bias signal. The microwave signal is fed through the end launch connector. The device is partially inserted into the anechoic chamber in such way that the antennas extend past the inner layer of microwave absorbing foam.

Zoom In Zoom Out Reset image size

The Liquid Crystalline material used in the phased array phase shifter cell was selected from the Dual Frequency Nematic Liquid Crystal (DFLC) family, where the sign and magnitude of the electric permittivity's anisotropy can be modulated by the frequency of the driving electric field. Origins of this class of materials began in the mid 70's [34]. Unfortunately, the first generations of DFN materials, composed of polar aromatic triring diesters, suffered from poor stability and high ionic contents. This generated strong parasitic currents making driving problematic. The material selected for this study is a state-of-the-art mixture 1909C composed of stable, fluorinated LC compounds, where the molecular dipole moment is mainly generated by fluoroaromatic systems with a small portion of isothiocyanato terminated units [35, 36]. The proper formulation of the mixture was performed by optimizing the mixing of two submixtures. The first, a dielectrically negative base submixture, produces a stable, negative dielectric anisotropy and ensures the optical anisotropy and appropriate viscosity for the second submixture. The second submixture, composed of long, dielectrically positive compounds with a low molecular relaxation frequency of 8.57 kHz at 20 °C and an activation energy of 60.2 kJ mol⁻¹, produces a low cross-over frequency (the frequency where the sign of dielectric anisotropy changes). Multicomponent liquid crystalline mixture 1909C has a crossover frequency 10.3 kHz at 20 °C and the values of the dielectric anisotropy at low- and high-frequency driving are Δε_L = +3.4 and Δε_H = −2.6 respectively. The parallel component of the electric permittivity varies from 10.5 to 4.7 for low and high frequencies respectively.

In this section, we present the results of the operational testing performed on the DFLC based phased array antenna assembly. First, we present the results we obtained while measuring the individual phase shift lines in the DFLC phase control cell. The individual phase shift RF transmission lines were measured with the setup described in figure 17. We then discuss the results obtained from anechoic chamber measurements of the antenna assembly.

Figure 22 is a graph of the relative phase versus bias voltage for individual RF transmission lines (L1, L2, etc). This is an important test for a phased array antenna system as each element must exhibit the designed phase shift to ensure proper steerability. Although there are minor differences in phase shift from one line to another, these will cause small changes in the beam profile as discussed in section 3. Additionally, these can be readily corrected for during the antenna tuning. The test also describes one of the figures of merit for this LC phase control cell. The measured phase shift per unit length has mean value of 55.5 deg/cm and a maximum difference of 1.6 deg/cm between any two of the 8 transmission lines. The calculated phase shift per unit length for the 5 cm long active length of the LC cell is 55.8 deg/cm.

Zoom In Zoom Out Reset image size

In addition, we see in this graph a second example of why LC phase control is desirable; the graph demonstrates the output phase of the microwave signal is continuously tunable unlike some MEMs applications which only produce discrete phase shifts that must be accounted for in building the beam steering profiles. With a continuously tunable dielectric, one can set up the conditions for various beam steers easily once the relationship between bias conditions and output RF signal phase is known for each line. Also, minor variations between transmission lines can be readily accounted for in the beam steer profile.

The data shown in figure 23 was used to choose the operating point of the phased array. In this test, the RF transmission (S21) of each individual transmission element (L1, L2, etc) in the LC cell was measured. The initial design was for 30 GHz, 32 GHz was chosen based on the convergence of S21 for each line to less than 2.3 dB maximum deviation between lines. This point was also chosen as it was closest to the design point of 30 GHz to minimize any issues with the other components in the input section and antenna section. The data is uncalibrated. The many components used to obtain the data interfered with obtaining an accurate calibration with the VNA.

Zoom In Zoom Out Reset image size

Following the individual line characterization, the phased array antenna was assembled and tested in the anechoic chamber. Figure 24 provides the time response of the overall array. To generate this test, the array was initially set for a boresight condition with a constant 50 V_peak bias voltage. With the receiver horn directly across from the array, a received power of −27.5 dB was measured. Then the beam was steered, by varying the bias frequency of individual lines, to 10 degrees relative to boresight. With a 10 degree steer condition set, the power along the boresight was then recorded as approximately −32.7 dB. The steer condition was then switched between 10 degrees and boresight with the horn maintained at boresight and transmitted power was measured as a function of time. With this setup, the time to switch from −32.7 dB to −27.5 dB is the response time of the array. For the proprietary DFLC under test, the steering time is 42 ms. However, based on the time response for MLC-2048 given in figure 16, we expect the steering time could be reduced to under 10 ms. This would also require the use of an appropriate driving scheme that we were unable to obtain with our current biasing system [19].

Zoom In Zoom Out Reset image size

To achieve the response times shown in figure 24, a driving scheme known as 'overdriving' was used. In this driving scheme, both the AC bias frequency and voltage are changed to drive the LC molecule. With the system stable, a high voltage is applied to create a rapid rotation of the LC molecules. The molecules are then 'caught' in the desired state by reducing the voltage and setting the desired frequency. This results in a fast rotation of the LC molecule and minimal stabilization times. Table 3 provides the voltage and frequency relationships used to obtain this data.

Initially, the system is in the boresight state. The first row is the 'overdrive' to the 30 deg beam steer that begins the LC molecule rotation. The next row 'catches' the LC molecules at the desired orientation to produce the beam steer. The reduced voltage and frequencies minimize the molecular stabilization times at the new rotational orientation. The voltage values are alternated between each line to control the output phase and produce a 30-degree steer. The 100 ms time for this step provided enough time to ensure stabilization was reached during the test. The frequency was then driven to 50 kHz forcing the LC molecules into the boresight configuration with additional time added to ensure the system is stable. The final step sets up the sequence for the next test by reducing the force holding the system at boresight. Such a driving scheme can be readily programmed into the AC bias control system.

In figure 25, we give the results of beam profile testing performed at 32 GHz. For this test, beam steering was set to boresight, 30 deg and 40 deg. The bias signal was set to 100 V_p-p and the frequency of the bias signal was varied between each element to achieve the desired RF output phase relation. The horn was rotated around the array to determine the beam profile at each steering angle. One can see from the data, the measured beam profile closely agrees with the modeled profiles given in figures 11 and 13. We primarily attribute the loss in power at increased scan angles to coupling between the antenna elements [37].

Zoom In Zoom Out Reset image size

This coupling between elements can be reduced. One method to achieve better coupling characteristics is to switch to a patch type antenna design. In this configuration, known designs provide better compensation of inter-element effects [38, 39]. Additionally, in such a system, one can easily arrange the beam steering antenna elements in a grid on a single plane while maintaining a modular approach to the overall array design.

To make a functioning phased array, repeatability of the tuning must be demonstrated. To test this, a single line was tested, and the results are given in figure 26. The bias signal was stepped through a sequence of frequencies to set a desired phase shift and the phase was recorded at each value. This test was repeated 100 times. The standard deviation of the phase at each frequency step was calculated to use as a figure of merit for repeatability. The greatest standard deviation in phase recorded was <0.5 deg.

Zoom In Zoom Out Reset image size

To determine the power consumption of the AC biasing signal, we tested a 2 cm long phase shift device filled with the 1909C LC. In this test, we measured bias signal current at different bias signal voltages and frequencies. The results are given in figure 27. These results highlight another highly desirable characteristic of LC based phase shifters; the biasing signal power requirements are minimal in these devices.

Zoom In Zoom Out Reset image size

The design, construction, and testing of our array illustrates the feasibility of using DFLC for a small footprint, modular based, phased array radar system. We demonstrated fast response times, good beam formation, large phase shift per cm, and good repeatability. Critical results are summarized in table 4. Table 5 presents a comparison of tuning speeds found in recent literature with this work.

We designed and constructed a 32 GHz, linear, eight-element phased array antenna using liquid crystal as a phase shift medium. A novel 'flip-chip' method was used for assembly construction combining the best features of glass and pcb. The dual frequency liquid crystal phase shifters were characterized for total phase shift, transmission losses, and repeatability. The antenna assembly was tested for response time and steerability. The phased array antenna assembly demonstrated the viability of using dual frequency liquid crystal media for faster beam steering as compared to standard nematic liquid crystal. The array beam steering time is 42 ms as opposed to the best previously published results that gives a range of 140–180 ms. We also show that this technology allows for antenna design for higher frequencies over conventional phase shift technology. This opens the way to create small, compact systems that can be utilized in new areas including self-driving vehicles, hand-held secure communication devices, and robot guidance systems.

The data that support the findings of this study are available upon reasonable request from the authors.