A design method for negative refractive index (NRI) metasurface lenses whose thickness is thinner than that of conventional NRI metamaterial ones is presented. A one-dimensional distributed transmission-line model is introduced to the design and the design formulas for determining the line length and the characteristic impedance are shown. As an example, a NRI metasurface lens operating at 1.85 GHz is designed based on the proposed method. A circuit simulation is carried out and complex voltage distributions are calculated to confirm the focusing operation of the designed lens whose thickness is sufficiently thinner than the wavelength.
This paper designs a transmissive metalens unit with independent control of amplitude and phase. The proposed metalens can be used as an antenna-in-package (AiP) solution to shape the radiation patterns for the millimeter-wave (mm-Wave) transceiver chip modules. The proposed metalens unit cell has three metal layers. The top and bottom layers are metal grids. The middle layer is a T-shaped metal structure used to control the amplitude and phase. The unit cell works in the frequency band of 50-60GHz, the amplitude control range is 0-0.935, and the phase shift range is over 360 degrees, which can realize the independent control of amplitude and phase. Based on the unit cell, a wideband high-gain, low sidelobe levels metalens antenna is designed, with excellent performance (SLL better than -25 dB, beam deflection of ±30 degrees).
A 60 GHz direct up-conversion mixer in 22 nm FD-SOI is presented. Based on an output buffer with negative feedback, a 1-dB output compression point (P o1dB ) of 0.4 dBm and a conversion gain of 4.5 dB are achieved. The required local oscillator (LO) power is only −11 dBm due to an LO buffer. Including the buffers, the total DC power is 18.6 mW. The circuit core occupies 0.12 mm 2 . Compared to state-of-the-art CMOS mixers at 60 GHz, the circuit shows the highest P o1dB and the lowest LO power at high conversion gain. Moreover, for the first time, the switching between active and standby mode via back-gate biasing is investigated for such a mixer. The back-gate switching avoids the breakdown problems of front-gate switching.
This paper presents a wideband 5-bit vector-sum phase shifter implemented in a low-cost 90-nm CMOS process. The proposed vector-sum phase shifter consists of a wideband quadrature generator with low amplitude/phase imbalance and a wideband, low-phase variation variable-gain amplifier integrated with 0/π phase shifter to satisfy most of the operation frequency bands of the 5G millimeter-wave communications. From 25 to 43 GHz, the proposed 5-bit phase shifter demonstrates the RMS amplitude/phase errors of better than 1.4 dB/4.8°, and the insertion losses of all states are lower than 6.4 dB from 25 to 43 GHz.
An I-V global parameter extraction technique for the industry standard FinFET compact model BSIM-CMG using deep learning (DL) is presented in this paper. The training data of 750k is generated by Monte Carlo simulation of key BSIM-CMG Parameters and gate length (L G ) for multiple devices. The created deep learning parameter extractor is trained to use I-V and L G data to predict the BSIM-CMG parameters. The DL parameter extractor is verified using measured device data, with L G ranging from 50n to 970nm. The created global model was able to create an accurate fitting for the input characteristics of multiple devices while capturing the trends in key electrical parameters. The results show the tremendous potential of using DL to create accurate global models instantly where measurement and manufacturing errors are present.
An ultralow-power Ka-band low-noise amplifier (LNA) for next-generation radio astronomical receivers, fabricated in 90-nm CMOS technology, is presented in this paper. To reduce noise factor and bias current simultaneously, the gatesource transformer feedback (TF. Feedback) technique is selected to make a compromise between noise figure and input matching. From the measurement results, the proposed Ka-band LNA achieves a 19.1-dB small-signal gain with a 3.6-GHz 3-dB bandwidth (34.2-37.8 GHz) and a noise figure of 4.2 dB with only 1.4 mW dc power consumption (Pdc). To the best of the authors’ knowledge, this LNA shows the highest figure of merit (FoM), which is 3950 (1/W), among published Ka-band ultralow power LNAs.
This paper presents the development of Wilkinson topology-based compact unequal power divider implemented for indoor navigation system. The proposed unequal power divider which is designed to have a power ratio of 1:2.2 aims to operate at the frequency of 2.4 GHz satisfying the IEEE 802.11b/g protocol standard of wireless communication. A 1.6 mm thick flame-retardant 4 (FR4) epoxy dielectric substrate with the size of 41.2 mm by 32.8 mm is used for the design and realization. A surface mount technology (SMT) resistor type with the value of 110 Ω is employed to fulfill the design. From the characterization results, the measured reflection coefficients at the input port and two output ports, as well as the transmission coefficients, show good achievements which are comparable to the simulated ones. Furthermore, at the frequency of 2.4 GHz the realized unequal power divider achieves a –20.49 dB isolation coefficient.
This paper presents an octave tuning-range LC VCO which can be used in the phase-locked loop to synthesis all frequencies up to the highest oscillation frequency. The proposed VCO employs the E-M mixed-coupling technique which compromises four LC tanks coupled either through magnetic or electric coupling, thereby generating four distinct resonances. Switched tail resistors are added to suppress FoM variation over the tuning range. To further enhance the phase noise performance, customized transistor designs are utilized instead of the PDK provided alternatives for a more compact layout to minimize wiring length from tank to active core. The proposed VCO, fabricated in 65nm CMOS process, demonstrates a 75.7% tuning range from 5 GHz to 11.1 GHz with an optimal -145.8 dBc/Hz phase noise at 10-MHz offset, corresponding to a figure-of-merit-tuning (FoM T ) of 205 dBc/Hz. Notably, the FoM T variation remains considerably small, at less than 2.5dB, over the entire tuning range.
This paper presents a D-band wideband double-balanced up-conversion mixer using a 40-nm CMOS process. In conventional Gilbert cell mixers, the input signal is input to the gate of a MOSFET. However, the MOSFET gate has high input impedance, making it difficult to achieve wideband matching. In the proposed circuit, the gate width of the MOSFETs is appropriately adjusted and the signal is input directly to the source to achieve wideband characteristics. In addition, by reducing the number of MOSFETs in series, low-voltage operation is possible and high IP3 is achieved. The proposed circuit achieved a 3-dB bandwidth of 50.7 GHz and a maximum conversion gain of –11.5 dB. The supply voltage was 0.9 V and the power consumption was 8.1 mW. The LO input signal was –0.2 dBm.
This paper presents a 10GHz voltage-controlled oscillator (VCO) with low power consumption, implemented in 22nm CMOS technology. The VCO utilizes a transformer to achieve negative gm-boosting, which improves the phase noise and start-up conditions. By optimizing the impulse sensitivity function (ISF), the VCO achieves an after-layout simulated phase noise of -113.5 dBc/Hz at 1MHz offset, with a power consumption of only 1.08 mW at 0.6 V supply voltage. The figure of merit (FOM) is as low as -193.1 dBc/Hz.
This paper presents a miniaturized high-power handling CMOS T/R switch. The chip size can be miniaturized by adapting the stacked inductor to compensate for the non-ideal switching characteristics of the transistors in both transmit and receive paths for the asymmetric T/R switch. The minimum insertion loss is 1.8 dB, and 1-dB bandwidth is 68% from 19.5 GHz to 39.5 GHz in Tx mode. For the Rx mode, the minimum insertion loss is 2 dB, and the 1-dB bandwidth is 51% from 25 GHz to 42 GHz. The measured Tx mode IP1dB is 30 dBm at 28 GHz and more than 25 dBm within the operating bandwidth.
Tunability of a varactor diode connected in series with a non-Foster negative capacitor, is investigated. Significant enhancement of tunability is shown possible for a practical negative capacitor implemented as a Linvill’s negative impedance converter (NIC) on bipolar transistors. To illustrate the advantage for tunable microwave devices, a series LC-tank with tunable resonant frequency was designed using a commercially available varactor diode and a Linvill’s NIC. Almost two-fold extension of tuning range of the resonant frequency is demonstrated while the Q-factor is comparable or even higher than it is without using the negative capacitor. The circuit remains stable within the entire tuning range.
In this paper, characteristic evaluation of a reconfigurable capacitor-based loaded-line phase shifter is presented by utilizing an equivalent circuit approach. For this purpose, a straightforward equivalent circuit of the phase shifter composed of three microstrip lines and four loading capacitors is developed. The values of lumped elements for the equivalent circuit are determined from the physical dimensions of the phase shifter structure. By using the developed equivalent circuit, phase shifter characteristics, including reflection and transmission coefficients, are evaluated for a varied 0.2 pF to 1 pF capacitance range. The characteristics obtained from the proposed approach agree with the measured results.
Sub-THz transceivers are anticipated to emerge soon for supporting communication systems beyond 5G. Precise printed circuit board (PCB) modeling has therefore become essential for designing these modules. Accordingly, the experimental model extraction for a low transmission loss PCB will be explained in this report. The modeling work is aiming at defining the model parameters of both the dielectric and metal layers while considering the influence of the surface roughness. For extracting the model experimentally, planar transmission lines (TL) had been fabricated on the concerned PCB, measured over a 0.1 to 110 GHz frequency range, and then multi-Thru Reflect Line (m-TRL) calibration was introduced. Ansys Electromagnetic simulator, HFSS, had then been used to simulate the measured TL while optimizing the PCB’s model to fit the associated measurement. Djordjevic-Sarkar formula was referred and defined in HFSS in terms of the sweep parameters to model the dielectric material enabling optimization with the full degree of freedom. Copper conductivity and the snowball radius of Huray roughness model, on the other hand, were optimized on the account of the conductor interconnect under the influence of the surface roughness. A Rat-Race BALUN had been referred as a case study to approve the extracted model. Based on the optimum extracted model, the simulation results are showing a good agreement with the lab measurements. Furthermore, the designed 50 Ohm TL had been optimized and the measurement of the new TL had noticeably shown an enhancement in its performance.
In this paper, we present a novel and simplified topology for a vector-sum phase shifter. The proposed topology leverages only two biphase modulators (based on 90° and 180° hybrid couplers), Wilkinson power dividers, and variable gain components. The biphase modulators generate the necessary quadrature vectors (±I, ±Q) for vector synthesis. To demonstrate the effectiveness of the proposed approach, a passive Kaband four-bit vector-sum phase shifter was designed using PCB technology. The EM simulation results exhibit a return loss of less than 10 dB across the bandwidth from 36 GHz to 44 GHz for all phase shifts, a phase shift range of 360°, and RMS phase and amplitude errors of 1.9° and 4.4 dB, respectively, at 39 GHz. The average insertion loss (IL) was simulated to be 19.5 dB. Despite its simplicity, the phase shifter structure achieves excellent phase performance, showcasing its potential in practical applications.
An ultra-wide tuning range LC voltage-controlled oscillator (VCO) with an excellent FoM T of 192.5dBc/Hz at 10 MHz offset in 130-nm SiGe BiCMOS is presented in this work. The VCO employs a pair of HBTs to build cross-coupling configurations and works in class-C mode. To obtain the tuning range as wide as possible, a 6-bit switched capacitor bank is applied to expand the tuning range with a pair of varactors. Simultaneously, a centrally symmetrical layout is used to reduce the parasitic effect of layout asymmetry of the HBT cross-coupling pair on the circuit under class-C working conditions. The proposed VCO possesses phase noise of -113dBc/Hz at 1MHz offset and -138.3dBc/Hz at 10MHz offset, and can continuously tune between 9.58-13.8GHz (36.1%). The FoM T of this VCO is competitive in this frequency band, which shows a great advantage in the application of 5G millimeter-wave or satellite communication.
A wideband divide-by-4/5 dual-modulus prescaler (DMP) with operating frequency from 4 GHz to 38 GHz has been designed and fabricated in 0.13-μm SiGe BiCMOS technology. An on-chip wideband marchand balun is used for the conversion of single-ended to differential signal at input port of the DMP and a pair of emitter followers serve as output buffers. In addition, the split-resistor load technique is also adopted to enhance the operating frequency. The proposed DMP can operate effectively at any supply voltage from 2.5 V to 3.3 V for a wide temperature range from –40 °C to 125 °C. It achieves a measured maximum operating frequency of 38 GHz with 50.2 mW power consumption at 3.3 V voltage supply. The chip area is 1.0×0.57 mm 2 with a core area of only 0.14×0.1 mm 2 .
This report presents a distributed model with a high scaling accuracy for GaN HEMTs up to 100 GHz. Using the distributed model, RF voltage V grf_i (i = 1, 2, …, 8) at each gate terminal of N intrinsic transistor models in the finger was simulated. The simulation revealed that the amplitude and phase differences of V grf_i in the finger were larger at 100 GHz than at 30 GHz, and the differences were larger when the gate inductance in the finger increased. The simulation results also showed that the gate inductance of the distributed model was more sensitive to gain than that of the non-distributed model and that gain at 100 GHz decreased with increasing gate inductance in the distributed model. The accuracy of the distributed model was verified, and the distributed model was confirmed to reproduce the measured S-parameters well. The MSG/MAG of the distributed model, considering the effect of gate inductance in the fingers, was in good agreement with the measured MSG/MAG up to 100 GHz over a wide gate width range of 30–100 μm. The distributed model was confirmed to be useful for improving the scaling accuracy of sub-THz GaN HEMTs.
In this paper, we propose a 20GHz-band low backlobe Vivaldi antenna array for digital beamforming (DBF) transmitter module. The proposed endfire antenna array can be directly integrated with DBF transmitter module. With the low backlobe characteristics, the influence on radiation pattern from transmitter module behind antenna can be eliminated. By using two corrugated slits on both edge sides, the front-to-back (F/B) ratio of antenna element can be improved beyond 20 dB in simulation. 4-element Vivaldi endfire antenna array is fabricated and measured. The measurement result of proposed antenna array has a good agreement with the simulation result.
This article reveals a 3D printed broadband metasurface that enhances the measured effective isotropic radiated power (EIRP) when integrated into a millimeter-wave (mm-Wave) transceiver with antenna-in-package (AiP) solution. Pancharatnam-Berry (P-B) phase shift approach and phase delay lines will be involved simultaneously to achieve beam-shaping performance. An antenna-circuit-antenna (ACA) configuration is applied to achieve a flexible phase transition between the transmitting and receiving antenna elements for dynamic beam-shaping and broadband radiation. The proposed metasurface operates at 25 GHz-30 GHz, with a peak transmission efficiency of 72%.
