A Compact Hybrid G-band Heterodyne Receiver Integrated with Millimeter Microwave Integrated Circuits and Schottky Diode-Based Circuits

Abstract

This paper presents a compact hybrid G-band (170–260 GHz) heterodyne receiver module incorporating both Millimeter Microwave Integrated Circuits (MMICs) and a Schottky diode-based circuit. An on-chip sextupler and a Low Noise Amplifier (LNA), along with a diode-based Sub-Harmonic Mixer (SHM), are integrated into the demonstrated singular module, which is carefully designed and arranged with the co-simulations in electromagnetic and thermal domain. Through this methodology, a terahertz receiver module is fabricated with a volume of only 27 × 20 × 20 mm³. The measured results indicate that the double-sideband conversion gain of the receiver is 10.5–17.5 dB from 195 GHz to 230 GHz, while the noise temperature is 1009–1158 K. As a result, this terahertz receiver provides recorded miniaturized hardware applicable for terahertz Integration of Sensing and Communication (ISAC) systems.

1. Introduction

In recent years, the exhaustion of spectrum resources in the microwave band has motivated the adoption of a higher and wider spectrum [1]. Following this trend of enhancing the carrier frequencies, the Terahertz (THz) (0.1–10 THz) band is envisioned as one of the key technologies for supporting the sixth generation (6G) of wireless communication systems [2]. The THz band’s extraordinarily wide bandwidth provides extremely high sensing precision, as well as extremely fast transmission speeds of up to hundreds of Gbps, and potentially Tbps [3,4]. THz waves have a short wavelength; hence tiny antennas are expected to be used to enable highly portable and wearable electronics [5]. Additionally, the non-ionization of THz radiation guarantees the safety of THz devices for the human body [6].

Over the past years, heterodyne transceivers have always been preferred in numerous terahertz applications, including security imaging, earth observation, and high data-rate communications [7,8,9,10,11]. Researchers in the field of Integration of Sensing and Communication (ISAC) are increasingly fascinated with the terahertz band because of its unique advantages, including small device size, ultra-high data rate up to Tbit/s, strong sensing ability, and high security at the physical layer [12]. Traditional terahertz transceivers, which still consist of a number of discrete modules in cascade and operate well, fall short when highly integrated and compact systems are required. To perform the up/down-convert function, traditional terahertz transceivers often require numerous independent functional modules in a cascade [13,14,15,16,17,18,19]. However, the latest generation of terahertz front ends, applied in spaceborne and airborne platforms, put forward the requirements of miniaturization. As a consequence, the development of highly integrated terahertz Radio Frequency (RF) front-end modules with competitive performance is quite worthwhile.

Using a single integrated multi-function receiver chip to develop compact receivers is indeed a highly integrated, low-cost mainstream approach. However, this paper focuses on a method that incorporates both packaged Schottky diodes and MMICs. This method has several advantages over the multifunctional integrated receiver chip in reference [20], as follows: (1) High design flexibility. Modules with varying frequency bands and functionalities can be rapidly developed to accommodate diverse system requirements. (2) Better performance. Owing to the inherent properties of the components, Schottky diodes exhibit certain advantages over chips in terms of key metrics for a mixer, such as conversion loss and noise figure. (3) Shorter research and iteration cycles. The design and development process for discrete components exhibits a shorter iteration cycle compared to that of integrated chip design. (4) Enhanced modularity. The system architecture permits the integration of additional components, such as filters and power dividers, to accommodate specialized system requirements. Therefore, this study presents a viable alternative for compact packaging of terahertz receivers with superior performance.

In this paper, a hybrid method of assembly is used to combine MMICs and a Schottky diode-based circuit in a single container. With this method, a single module performs the functions of a low-noise amplifier, harmonic mixer, and local oscillator drive. Due to their different degrees of heat tolerance, Schottky diodes and MMICs require solders with various melting points during the micro-assembly process. Before each component is baked and welded, the solders must be organized according to the melting point, beginning with the greatest melting point and moving down. Finally, by merging electromagnetic and thermal co-simulation design, a miniaturized full-function terahertz receiver is made possible. Its efficacy might be superior to that of a standard terahertz receiver with cascaded discrete functional modules.

The rest of this article is structured as follows. We describe each module in the hypothetical G-band receiver in Section 2. Then, in Section 3, the receiver’s overall assembly information is presented, and in Section 4, the receiver’s measurement results are provided. Finally, in Section 5, we reach our conclusions.

2. Configuration and Design

2.1. Overview

Figure 1 depicts the schematic layout of a terahertz receiver. The W-band (75–110 GHz) sextupler chip, the Schottky diode-based Sub-Harmonic Mixer (SHM), and the terahertz Low Noise Amplifier (LNA) chip make up the majority of the terahertz receiver. Through a coaxial connector, the Local Oscillator (LO) signal enters the module. The 17.5–18.4 GHz signal is then multiplied by a sextupler chip to produce the W-band, which is used for providing the LO signal to the SHM. Additionally, the antenna receives the G-band RF signal, which is then channeled through the WR-4.3 rectangle waveguide and into the module. The SHM then downconverts the RF signal amplified by the LNA chip to Intermediate Frequency (IF).

2.2. LO Sextupler Design

An active sextupler is used to deliver adequate LO power to the mixer in order to convert the input K-band signals to W-band. This component uses an active sextupler chip, which can multiply signals from the 17.5 GHz to 18.4 GHz band to the 105 GHz to 110 GHz band. The output power of the sextupler chip is approximately 6 dBm when pumped with 16 dBm of input power. The InP High Electron Mobility Transistor (HEMT) process is utilized for manufacturing the sextupler chip. For the adjacent harmonics (fifth and seventh harmonics) suppression, the chip is a sextupler chip with a typical value of 20 dBc. In normal circumstances, a common supply voltage of +4 V is used for all the drains, and −0.35 V of the gate voltage is provided to bias the chip depending on its operating condition. The drain current is 80 mA for the chip. The chip was 1.33 mm in width, 70 μm in thickness, and 2.8 mm in length (input and output directions).

Firstly, as seen in Figure 2a, it is decided that the E-plane probe would transmit the 75–110 GHz output signal into the WR-10 rectangle waveguide, which is fabricated with a 127 μm-thick quartz substrate. Then, the probe and the chip packaging cavity need to be evaluated as a whole. Figure 2b shows that the W-band E-plane probe’s simulated S-parameters result in a full-band return loss that is better than 20 dB.

2.3. RF Low-Noise Amplifier Design

The low-noise amplifier chip is also manufactured by the InP HEMT process. The datasheet of the chip shows that this chip has a linear gain of better than 24 dB and a typical noise figure of 5.8 dB on the chip tested in the 195 GHz and 230 GHz operating bands. The drain current is 37 mA on typical conditions, which requires a common supply voltage of +1.2 V for all of the drains and −0.3 V of the gate voltage to bias the chip depending on its operating conditions. The chip was 0.97 mm in width, 50 μm in thickness, and 2.14 mm in length (input and output directions).

Figure 3a illustrates the factors that must be taken into consideration during the 3D EM simulation of the G-band low-noise amplifier module. Microstrip probes are suspended in the G-band E-plane and attached to the chip’s input and output ports. The DC supply cavities on both sides are divided and connected using just an over-DC low-pass filter in order to prevent resonance of the chip’s RF cavity. The microscale impacts of the electromagnetic field are particularly sensitive to the G-band, where the low-noise amplifier operates. As a result, when designing the THz low-loss waveguide transition probe, it is essential to model and simulate the arc and arc duration of the bonding wire. According to Figure 3b, the simulated transmission line and both sides of the transition probes with 18 um bonded gold wire have a total transmission loss of 0.5 dB at the 210–230 GHz band. As a rule of thumb, during wire bonding, the gold wire should be as flattened as feasible because the loss of a gold wire will rise with its length and height.

In order to test various probes for rectangular waveguide transition structures, specialized fixtures were created, as seen in Figure 4a. The probes are positioned back-to-back in the fixture and connected using gold wire bonding. Additionally, it has been determined through comparison trials of several probe types that the suspended microstrip probe performs best in the G-band. Figure 4b displays the probe’s back-to-back S-parameter performance. The demonstrated back-to-back probe exhibits a typical transmission loss of 1 dB in the 200 GHz to 230 GHz frequency range, from which the loss of a single-ended E-probe could be derived by multiplying a factor of 0.5.

2.4. Sub-Harmonic Mixer Design

The G-band SHM’s construction is shown in Figure 5a. The G-band SHM’s RF signal is supplied into the mixer Schottky diode pair through the WR4.3 rectangle waveguide, and then out to the microstrip probe. On the opposite side of the Schottky diode pair, the WR-8 rectangular waveguide supplies the LO signal. To decrease the RF signal leakage to the LO port, and stop the RF signal and LO signal from interfering with each other, a low pass filter must be installed at the end of the LO signal, near the Schottky diode. This will ensure that the input of the LO signal is receiving pure LO signals. The IF low pass filter is added to the IF output near the Schottky diode, and it can filter out spurious signals (including the local oscillator signal and other spurious signals obtained by mixing) except the IF signal.

This SHM contains GaAs Anti-Parallel Schottky diodes as a component. There are two anti-parallel anodes in this diode. The major spice parameters are series resistance Rs = 12 Ohm, ideality factor n = 1.18, saturation current Is = 1.5 fA, and nonlinear junction capacitor at zero bias voltage Cj0 = 1.45 fF.

The following strategies are adopted in the design to minimize the transmission loss in the circuit. Firstly, the suspended microstrip is adopted to design the mixer circuit instead of the traditional microstrip. Secondly, the intrinsic resonances of the hammer head filter are introduced to enhance signal choke. Compared to the conventional high-low impedance filter, it conduces to reduce the length of the transmission line. Thirdly, to realize impedance matching in large bandwidth, the RF and LO input waveguide consist of several reduced height waveguides, respectively. Finally, to reduce the LO pumping power, the IF port is designed to have a high impendence of 100 Ohm. It is better for reducing the electron thermal noise and improving the performance of noise temperature in the mixer. As detailed in Figure 5b, the S11-parameters of the LO and RF ports of the mixer are below 12 dB in the 210–240 GHz frequency range, and the conversion loss of the diode-based SHM is 6.5–8 dB from 210 GHz to 240 GHz.

3. Module Assembly

The design of the modules needs to address numerous structural challenges. The signal in the terahertz band only flows through a waveguide with little transmission loss; hence the three main functional components provided above employ waveguide connectivity. However, the output waveguide of the sextupler is a WR10 waveguide (2.54 mm × 1.27 mm) that is gradually reduced in height to become a WR8 waveguide (2.032 mm × 1.016 mm) in order to achieve the optimum impedance match for the mixer input. The RF input waveguide of the mixer is a WR4.3 waveguide (1.092 mm × 0.546 mm), which is the same as the output waveguide of the LNA. The entire cavity is then simulated in order to determine whether the combination of cavities will result in a resonant response. In addition, a 1000 pF capacitor and two 100 pf capacitors and voltage distribution resistor are arranged directly in the resonant cavity of the sextupler, in the same cavity as the chip but without resonance. However, the same number of capacitors and resistors placed in the resonant cavity of the LNA chip will produce resonance since the low noise amplifier operates in the G-band. Therefore, the resonant and supply cavities of the low noise amplifier are independently isolated, and each cavity is transferred to the DC supply via low pass filters. Additionally, some absorbent material has been installed on the top of the chip cavity in the appropriate location to avoid excessive resonance. A microstrip line is used to connect the coaxial connections for both the LO input and the IF output, and it has good transmission capability at frequencies below 20 GHz. The RF input port uses the same WR4.3 standard rectangular waveguide flange as the antenna. Figure 6 depicts a picture of the G-band receiver module’s internal interconnection arrangement.

Due to their different levels of heat tolerance, Schottky diodes and MMICs require solders with differing melting points during the micro-assembly process. The recommended solder for each component is shown in Figure 7. In order to bake and solder each component individually, the solders used in the micro-assembly must be arranged according to their melting points, starting with the highest melting point and going downward. The Au88Ge12 solder, having a melting point of 370 °C, is first used for soldering the DC insulator to the cavity. The second stage involves using Au80Sn20 solder with 330 °C melting temperature to solder the two chips to their respective locations in the cavity. Additionally, using the In70Pb30 solder with a melting point of 175 °C, the Schottky diode is attached to the appropriate location on the mixer substrate. In the following phase, electrically conductive silver epoxy (H20E), with a melting point of only 120 °C, is used to bond the capacitor and the complete substrates in the designated position. After all parts have been soldered and baked, the input and output ports of active chips are linked to the transmission substrates using gold wire bonding. The insulators and capacitors are connected by 100 μm wide gold tape to provide DC power to the chip.

In addition, the module has a power-supply secondary-voltage-stabilizing circuit installed inside of it that powers the two active chips. When the circuit is fed with only +8 V DC, the circuit is able to provide both 4 V and 1.2 V positive and −0.3 V and −0.35 V negative power for the two active chips. During typical module operation, the total module current fluctuates between 110 mA and 116 mA, resulting in overall power consumption of the receiver between 0.88 W and 0.93 W. Finally, the size of the G-band receiver module (excluding the coaxial connectors) is only 27 × 20 × 20 mm³, as shown in Figure 8.

4. Measurement Results

4.1. Conversion Gain

The conversion-gain-measurement platform for the terahertz receiver is shown in Figure 9a. One signal source is used to supply the receiver’s LO signal, and the second signal source serves as the RF input signal for the G-band multiplication expansion module. To prevent an overwhelming RF signal, a 20 dB attenuator is connected in series. To monitor the strength of the spectrum signal, the IF port is linked to the spectrum analyzer. The receiver module has increasing conversion gain due to the integrated low-noise amplifier chip.

In Figure 9b, the conversion-gain curves for the three states are shown. The first two states are the frequency conversion gain curves for fixed IF frequencies at 1 GHz and 5 GHz, respectively. To guarantee that the IF frequency is fixed at 1 GHz and 5 GHz, the receiver’s LO frequency sweeps follow the transmitter’s frequency sweep. The third state is the frequency conversion gain curve brought about by sweeping the LO frequency at a fixed RF frequency of 220 GHz. Fixing the RF frequency to scan the LO frequency is another test mode. Over the frequency range of 195–230 GHz, the typical conversion gain is 10.5–17.5 dB.

4.2. DSB Noise Temperature

The Double Sideband (DSB) noise temperature of the receiver is measured using a modified Y-factor method. Figure 10a depicts the DSB noise temperature measurement platform for the terahertz receiver. As a high-temperature radiator, an infrared radiation source operates at 750 °C. Low-temperature radiation comes from an environment that is at room temperature. The IF output noise power measured at various temperatures can be used to compute the noise power difference Y factor. The DSB noise temperature can then be obtained from the Y factor. Figure 10b shows the LO power versus DSB noise temperature at the LO frequency of 220 GHz. The lowest DSB noise temperature in 220 GHz is between 1009 K and 1158 K when the LO power is between 15.9 dBm and 16.4 dBm. The LO frequency versus noise plot in Figure 10c is at the LO power of 15.9 dBm. The minimal DSB noise temperature at LO frequencies between 215 GHz and 225 GHz with an LO power of 15.9 dBm is 1009–1088 K.

4.3. S-Parameters

The S-parameters of the receiver module’s port should be tested using the vector network analyzer and a G-band frequency expansion module. Figure 11a demonstrates that in order to test the receiver’s RF interface, a Y-band frequency expansion module is necessary. According to the test findings shown in Figure 11b, the return loss S11 of the receiver’s RF port is better than 10 dB over a 20 GHz band from 210 GHz to 230 GHz, and even better than 20 dB close to the 220 GHz frequency point.

The IF port of the receiver can be examined directly using the coaxial line of the vector network, as shown in Figure 12a. Figure 12b’s curves show that the return loss S11 of the receiver’s IF port is better than 5 dB over a 20 GHz band from 0 GHz to 20 GHz, and even better than 20 dB close to the 12 GHz frequency point. This demonstrates that each receiver port’s impedance-matching and return-loss performance is satisfactory.

4.4. P1dB

The P1dB of the receiver is the point at which the conversion gain is 1 dB less than the conversion gain in the linear region of the receiver. A signal source for the local oscillation signal, another signal source and frequency expansion module cascaded to provide the RF signal, and an IF power meter to measure the IF output power composed up the P1dB test platform, as depicted in Figure 13a. Before the measurement process begins, the output power of the frequency expansion module must be calibrated as a benchmark for the receiver’s RF input power. Then, set the two sources’ setups so that the RF frequency operates at 220 GHz and the receiver LO frequency operates at 219 GHz. The receiver is then connected to the expansion module, and the RF power is gradually increased initially at a low power level while the power meter monitors and records the IF output power. Next, the gain 1 dB gap point is determined by comparing the curves of the ideal linear output power and the actual IF output power. As shown in Figure 13b, the receiver’s P1dB_out is −13.4 dBm.

4.5. IF Spectrum

The same measurement platform was used for measuring the conversion gain in the preceding section, as well as the IF spectrum of the receiver on this platform, as shown in Figure 14a. By establishing a signal generator, the LO frequency of the terahertz receiver operates at 220 GHz. Another signal generator and frequency expansion module provides the RF signal for the receiver, which is transmitted at 219 GHz and 215 GHz, respectively. The IF spectrum was measured with a spectrum analyzer at 500 Hz BW. Figure 14b depicts the IF spectrum at 1 GHz, with spurious suppression of more than 66 dBc at a 1 GHz span. Similarly, Figure 14c shows the IF spectrum at 5 GHz with spurious suppression over 72 dBc at a 3 GHz span.

The reported terahertz receivers in G-band are compared to this study in

Table 1. Performance comparison of terahertz receivers in G-band.

Reference	Frequency (GHz)	Conversion Gain (dB)	Noise Temperature (K)	Dimensions (mm × mm × mm)	Configuration
[10]	173–205	−20~−13	Not indicated	≈80 × 30 × 40	Discrete
[11]	209.4–219.6	−8.4~−6.2	725~1550	150 × 60 × 19.1	Discrete
[12]	185–250	−11~−7.5	750–1600	20 × 20 × 35	Integrated front-end
[13]	221–236	+9.3	3052.8	80 × 40 × 20	Discrete
This work	195–230	+10.5~+17.5	1009–1158	27 × 20 × 20	Integrated front-end + LNA

. In this study, a cavity contains the mixing structure, local oscillator driver chip, and RF low noise amplifier chip. The use of RF low noise amplifiers allows the receiver to provide a significant amount of conversion gain. Meanwhile, this architecture is more competitive in performance when compared to conventional discrete, cascaded circuits due to the shorter signal transmission path and greater impedance matching caused by connectivity. The proposed module has enormous potential for creating more compact terahertz-receiving front ends. The overall power consumption of the receiver in normal operation is between 0.88 W and 0.93 W.

5. Conclusions

This study describes the design and verification of an integrated G-band heterodyne receiver module with high conversion gain and a low noise factor. The receiver’s size is decreased to 27 × 20 × 20 mm³ by integrating MMICs and Schottky diode components into a single module. According to the measured results, the receiver’s double-side-band conversion gain in the 195–230 GHz band is 10.5–17.5 dB, and its noise temperature is 1009–1158 K. The overall power consumption of the receiver in normal operation is between 0.88 and 0.93 W. Additionally, to act as the front end of the terahertz RF transmission link, the direction of the input and output of the low noise amplifier chip in the receiver can just be flipped around during the micro-assembly step. The terahertz RF transmission link may produce an output capacity of more than 100 mW saturated power in the 205–225 GHz frequency region when used in conjunction with a compact terahertz solid-state power amplifier. The integrated G-band receiver will serve as the core component for future airborne and spaceborne terahertz system applications. It also has the potential to be utilized in future terahertz ISAC systems.