A Novel Self-Adaptive Rectifier with High Efficiency and Wide Input Power Range

Abstract

A novel 2.45 GHz self-adaptive rectifier with high efficiency and a wide input power range is proposed in this paper. It consists of a high-power sub-rectifier branch, a low-power sub-rectifier branch, an impedance transform and isolation network (ITIN), and a feedback network. Impedance matching is realized by ITIN for both branches. The proposed design is able to switch between these two branches by the feedback network according to its output voltage level. The rectifier has been simulated, fabricated, and tested. The measured power conversion efficiency (PCE) exceeds 50% over the input power range from 5 to 29 dBm, with a total dynamic range of 24 dB. The input range when PCE exceeds 60% is from 10 dBm to 28 dBm. The maximum efficiency is 75.2% at 26 dBm input power.

1. Introduction

The microwave power transfer (MPT) system is designed to transmit and collect energy wirelessly through free space and converts received RF signal to DC currents for power supply [1,2,3]. MPT eliminates the need for cables and waveguides. Thus, it serves as an additional source of energy [4] and could be applied to scenarios that are beyond the reach of conventional methods. It has drawn increasing attention due to its potential contribution to green energy applications [5,6,7,8,9,10,11,12] and has been applied to applications such as space solar power satellites (SSPS) and power supply for high-altitude aircraft and drones. In recent years, the number of wireless charging devices is expected to increase after the concept of the Internet of Things (IoT) [13,14], especially when MPT is integrated with Mobile Edge Cloud (MEC) [15].

Despite its unique advantages, a significant advance in transmission efficiency is still required when it comes to practical applications [8]. One of the paramount problems at the receiver end is to improve rectification efficiency. The RF-rectifier comprises a rectifying diode, a DC-pass filter, and a load match by an impedance matching network [16,17]. In previous literature, rectifiers of different topologies, such as band-stop structure [18], bridge diode [19], Branch-line coupler [20,21], and rectifiers adopting GaN Schottky barrier diode [22,23], GaAs Schottky diodes [24], and reconfigurable hybrid structure [25] have been proposed to achieve high efficiency.

Due to the nonlinear characteristics of the Schottky diode, however, rectifiers could only achieve ideal efficiency at a given input power range. Otherwise, conversion efficiency will deteriorate rapidly [26]. In practical applications, the distance and incident angle from the RF source to the rectennas or harvester is very diverse, which greatly affects received RF power. The input power of the rectifier is not likely to always stay in the designed range, which will lead to very low efficiency in some circumstances. Thus, it is desirable to maintain high conversion efficiency over a wide input power range.

For rectifiers with a wide power input range, previous research has introduced techniques such as transmission-line-based resistance compression networks [27,28], branch-line couplers [21,29], and cooperative structure [30,31], which has presented a dual-channel rectifier with a Wilkinson power divider. The power dynamic range for efficiency of 50% is expanded by 2 dBm compared with a single HSMS2860 rectifier branch. In [32], He proposed a compact rectifier with a wide input power range, leveraging the nonlinear characteristics of the diodes. It achieves an RF-DC conversion efficiency exceeding 50% over an input range of 23.3 dB (from 6.5 to 29.8 dBm), with a maximum efficiency of 74.5% at 27 dBm input power. In [33], Lu et al. introduced a self-tuning matching structure to extend its input power range. The proposed rectenna reaches more than 50% conversion efficiency over a broadened input power range from −6–21 dB with a peak efficiency of 78.2%. [34] presents an auto-adaptive impedance matching for rectifiers. The strategy improves the efficiency of the system by up to 20%. [35] presents a reconfigurable harvesting device that consists of a passive RF detector, a three-level logic comparator, and three antenna switches. However, the detection and calculation units will increase the cost and size of the rectifier, and it will take an extra power supply to support the system. A novel dual-band rectifier with an extended power range and an optimal incident RF power strategy in settings where the available RF energy fluctuates considerably has been presented in [36]. A peak conversion efficiency of 60% was maintained from 5 to 15 dBm.

In this paper, we propose a novel rectifier with high efficiency and a wide input power range. The proposed rectifier consists of two sub-rectifier branches, an ITIN and a feedback network. The two sub-rectifier branches are self-biased and controlled by the feedback network according to the output voltage level. The ITIN network is able to improve the matching performance at high and low power levels, respectively. The rectifier has been fabricated and tested, achieving high efficiency over a wide input power range. Compared with [30,35], the load impedances at different power levels are constant. It does not involve logic comparators and MCUs like [35] does. Due to good matching performance, the input range for efficiency at higher efficiency is wider than that in [32]. The proposed rectifier has advantages in terms of high efficiency, wide input power range, reconfigurability, and easiness of integration.

2. Circuit Realization

The 2.45 GHz self-adaptive rectifier is shown in Figure 1. The overall design mainly consists of two sub-rectifiers branches operating at low and high-power levels, controlled by the feedback network. Between the two branches, there is a network for impedance matching and isolation of the two branches. A capacitor was used at the input port to block the reverse current. A DC-pass filter was deployed between the sub-rectifier branches and the load to suppress harmonics generated by the diode during rectification.

2.1. Design of Rectifier Branches and Feedback Network

In [18], Liu proposed a rectifier in series with a short-ended eighth-wavelength microstrip transmission line which works as a band-stop structure, as given in Figure 2. The proposed structure aims to compensate the diode capacitive impedance and turns to an open circuit to block the second harmonic for power recycling. In order to realize high efficiency over a wide input power range, two sub-rectifier branches I and II are designed based on [18] for low-power and high-power rectification, respectively, which are presented in Figure 3a,b. The adopted Schottky diodes D₁, D₂, and D₃ are HSMS282 with low junction capacitance (C_j0 = 0.7 pF), low series resistance (Rs = 6 Ω), and low forward voltage (V_F = 0.25 V).

The low-power sub-rectifier I is given in Figure 3a. It consists of an HSMS282 diode, a series quarter-length transmission line, and a series short-ended three-eighths-length transmission line. It is designed to implement rectification with about 16-dBm input power at ISM frequency 2.45 GHz. The high-power sub-rectifier is given in Figure 3b. It consists of two HSMS282 diodes, a series eighth-length transmission line, and a series short-ended eighth-length transmission line. It is designed to implement rectification with about 26-dBm input power at ISM frequency 2.45 GHz.

The input impedance Z_in of the eighth-wavelength short-ended transmission line is

$$Z_{in}=j{Z}_{SE}\tan \beta l=j{Z}_{SE}\tan \left(\frac{2\pi }{\lambda }·\frac{\lambda }{8}\right)=j{Z}_{SE}$$

(1)

where Z_SE is the characteristic impedance of the eighth-wavelength short-ended transmission line.

In order to compensate for the diode capacitive impedance, Z_SE should follow the below relation,

$$Z_{SE}=-Im\left\{Z_D\right\}$$

(2)

where Z_D is the diode impedance.

Therefore, the characteristic impedances of the microstrip lines TL₂, TL₃, TL₆, and TL₇ are obtained from Equation (2) and their respective line widths are determined.

The feedback network, as shown in Figure 1, consists of a bias resistor R₁, two capacitors C₃ and C₄, and two p-i-n diodes BAR64-02V from Infineon. The two p-i-n diodes are driven by output DC voltage. Bias resistor R₁ is deployed between the load and the diodes to control the applied voltage to the p-i-n diodes. Since the resistance is large, the feedback network only occupies a very small proportion of output power, which could be neglected. Capacitor C₁ and C₂ not only block DC current but also short-circuit RF signal when the diode is in forward bias. DC current passes the diodes and is short-circuited by the lines connected to the ground (TL₃ and TL₇), which eliminates the need for extra blocking capacitors.

For the low-power branch (Sub-rectifier I), without feedback voltage, the diode is in reverse bias. The feedback network is disconnected from Sub-rectifier I. Thus, the length of the transmission line connected to diode D₁ is 5λ/8, which is equivalent to λ/8 according to transmission line theory. The circuit is equivalent to that in Figure 2 and rectification is turned on. With feedback voltage, the input impedance of the feedback network is close to zero. Thus, TL₃ is short-circuited. Sub-rectifier I is equivalent to a diode in series with a short-ended λ/4 TL₂. The input impedance of TL₂ under such circumstances is a very large value. RF signal is blocked at the designed frequency. Thus, rectification is turned off.

For the high-power branch (Sub-rectifier II), without feedback voltage, the diode is in reverse bias. The feedback network is disconnected from Sub-rectifier II. Thus, the length of the transmission line connected to diodes D₂ and D₃ is λ/4. The input impedance of TL₆ and TL₇ in series is a very large value. RF signal is blocked at the designed frequency. Thus, rectification is turned off. With feedback voltage, the input impedance of the feedback network is close to zero. Thus, TL₇ is short-circuited and the transmission line connected to diodes D₂ and D₃ is λ/8 in length. The circuit is equivalent to that in Figure 2 and rectification is turned on.

By applying output voltage to the feedback network, we can either turn on sub-rectifier I and turn off sub-rectifier II when the input power level is low or turn off sub-rectifier I and turn on sub-rectifier II when the input power level is high, thereby enabling the automatic switch from the two sub-rectifier branches according to input power level. Additional bias network is eliminated, and the bias power is supplied by the rectifier circuit itself.

2.2. Design of the Impedance Transform and Isolation Network (ITIN)

The impedance of the Schottky diodes could be obtained by the following equation [37]

$$Z_D=\frac{\pi R_S}{cos{\theta }_{on}\left(\frac{{\theta }_{on}}{cos{\theta }_{on}} - sin{\theta }_{on}\right)+j\omega R_S{C}_j\left(\frac{\pi -{\theta }_{on}}{cos{\theta }_{on}}+sin{\theta }_{on}\right)}$$

(3)

where θ_on is the turn-on angle of the diode, and the junction capacitance of the diode is

$$C_j=C_{j0}\sqrt{\frac{V_{bi}}{V_0+V_{bi}}}$$

(4)

where V₀ is the output DC voltage and V_bi is the diode’s built-in voltage in the forward bias region (for HSMS282, V_bi = 0.25 V). From the above equations, we can obtain the impedances of the two branches Z_D1 and Z_D2, respectively.

The proposed ITIN consists of two quarter-wavelength transmission lines TL₄ and TL₅. The characteristic impedances of the two lines are Z_TL4 = Re{Z_D1} and Z_TL5 = (Re{Z_D1}·Re{Z_D2})^1/2. It is designed to isolate the two branches at the low-power level and transform the input impedance of sub-rectifier II to the impedance of sub-rectifier I at high-power level.

The equivalent circuits at low and high-power levels are given in Figure 4a,b.

At the low-power level, as illustrated in Figure 4a, sub-rectifier I is turned on and II is turned off. Z_in2 tends to be ∞. The electrical length of TL₄ and TL₅ combined is λ/2. Thus, Z₂ = Z_in2 = ∞ and the two branches are isolated. Without TL₄, sub-rectifier I will be short-ended. Meanwhile, Z_in1 = Re{Z_D1}. Z₁ = Re{Z_D1} and is transformed to input impedance Z_in = 50 Ω by TL₁.

At the high-power level, as illustrated in Figure 4b, sub-rectifier II is turned on and I is turned off. Z_in2 = Re{Z_D2}. It is transformed to Z₃ = Re{Z_D1} after TL₅. Z₂ = Z₃ = Re{Z_D1}. Meanwhile, Z_in1 tends to be ∞. Z₁ = Re{Z_D1} and is transformed to input impedance Z_in = 50 Ω by TL₁.

From the above analysis, the proposed design is able to isolate the two branches at the low-power level and transform the input impedance of sub-rectifier II to the impedance of sub-rectifier I at the high-power level. Z₁ = Re{Z_D1} at both high and low power levels and is transformed to input impedance Z_in = 50 Ω by TL₁.

3. Simulation and Experiment Results

The proposed rectifier has been optimized, fabricated, and tested for validation. The substrate is Rogers 5880 (ε_r = 2.2 and tanδ = 0.0009), with a thickness of 0.787 mm. The circuit layout is given in Figure 5a. The simulation is executed by Advanced Design System (ADS). Input power is provided by a signal generator. The output voltage is measured by a voltage meter. Output DC load is a standard resistor box.

The microwave-to-DC conversion efficiency η of the rectifier is defined as

$$\eta =\frac{P_{DC}}{P_{in}}\times 100\%=\frac{V_0^2}{P_{in}·Z_L}\times 100\%$$

(5)

where P_DC is the DC output power at load Z_L, P_in is the input power of the microwave source, V₀ is the output DC voltage, and Z_L is the DC load.

The simulation and measurement results are shown in Figure 6, which are in accordance with simulation. The conversion efficiency exceeds 50% over the input power range from 5 to 29 dBm, with a total dynamic range of 24 dB. The conversion efficiency of over 60% is from 10 dBm to 28 dBm. Maximum efficiency is 75.2% at 26 dBm. The proposed design achieves a wide input power range and high efficiency at a relatively high-power input range. The comparison of the proposed design with some recently published wide-input-power-range rectifiers is displayed in

Table 1. Comparison of rectifiers with wide input power range.

Ref.	Freq. (GHz)	Pin (dBm) Range for Eff. > 50%	Max. Eff. (%)	Methods	Extra Power Supply	Constant Load Impedance
[31]	2.45	8~27	75.5	Wilkinson power divider	No	No
[29]	2.45	2.9~20.2	80.8	Coupler	No	No
[30]	2.4	−3.5~26	72.8	Cooperative	No	No
[32]	2.45	6.5~29.8	74.5	Non-linear power division	No	Yes
[33]	2.45	−6~21	78.2	Self-biased network	No	Yes
[34]	2.45	−3~22	78	SPDT	Yes	No
This work	2.45	5~29	75.2%	Self-adaptive Branches	No	Yes

.

Compared with existing works in [29,30,33,34], the proposed design achieves a higher overall input power level. Note that our design is not aimed at achieving the best conversion efficiency or input power range. We have taken whether load impedance is constant, whether the circuit needs extra power supply, and impedance matching performance into consideration. For example, the designs in [30,35] achieve high efficiencies and wide input ranges. However, their load impedances at different power levels are not constant. In practical applications, the rectenna arrays are usually connected to the same load. Under such circumstances, the designs in [30,35] may not be applicable. Besides, the design in [35] consists of logic comparators and MCUs, which require an extra power supply. In [32], a non-linear power division strategy is adopted. Input power is divided according to the impedance variation of the two branches. It achieves a wide input range at a compact size. However, the circuit is mismatched. By contrast, in our design, due to the ITIN network, the circuit is matched when input power is 16 dBm and 26 dBm, respectively, achieving a better input range for efficiency >60% and efficiency >70%. Besides, the output impedance is constant, and no extra power supply is required.

The proposed design is not only suitable for traditional high-power MPT applications, such as SSPS and power supply for high-altitude aircraft and drones, but could also be applied to Wireless Power Mobile Edge Cloud (WPMEC). The WPMEC system adopts a time division multiplexing strategy. During the energy transfer period, the battery is charged. During the communication period, the battery supplies the communication circuit [15]. Higher power input is able to support devices with higher computational power. Meanwhile, it makes fast charging possible, which leaves a longer time period for communication. This investigation provides a good reference for future energy harvesting system design.

4. Conclusions

We have proposed a novel self-adaptive rectifier with high efficiency and a wide input power range which consists of two sub-rectifier branches for low and high power, an ITIN, and a feedback network. The rectifier is able to switch between these two branches according to output voltage, thus expanding its input power range. The ITIN network can improve the matching performance at high and low power levels, respectively. The circuit has been simulated, fabricated, and tested. The conversion efficiency exceeds 50% over the input power range from 5 to 29 dBm, with a total dynamic range of 24 dB. The conversion efficiency of over 60% is from 10 dBm to 28 dBm. The maximum efficiency is 75.2% at 26 dBm. The proposed design demonstrates great potential in future WPT applications.