Ambient Electromagnetic Wireless Energy Harvesting using Multiband Planar Antenna

Ambient Electromagnetic Wireless Energy Harvesting using Multiband Planar Antenna Antwi Nimo *, Dario Grgić and Leonhard M. Reindl University of Freiburg IMTEK, Department of Microsystems Engineering, Laboratory for Electrical Instrumentation, Georges-Koehler-Allee 16, D-7911 Freiburg, Germany *E-mail: antwi.nimo@imtek.de Abstract Ambient Electromagnetic (EM) energy harvester using multiband planar antenna and broadband radio frequency (RF) to direct current (DC) circuit is presented. The wireless EM harvester consists of a Printed Circuit Board (PCB) planar antenna and RF to DC circuit using schottky diodes and passive components for impedance matching. Measured and simulated S11 and gain for various resonant frequencies of the multiband antenna are presented. Average open circuit voltage and efficiency of the broadband RF to DC circuit from 8 MHz to 2 GHz is.45 V and 8 % respectively at - 1 dbm input power. It is shown in this work that using a single multiband antenna and broadband RF to DC circuit, the performance of an EM wireless harvester can be enhanced by harvesting energy from different ambient EM frequencies in its operating band at the same time. Keywords; wireless RF energy harvesting, multiband antenna, broadband RF to DC circuit, impedance matching. I. INTRODUCTION Micro energy harvesting has the potential to replace remote sensing elements requiring microwatt power levels for its operation. Ambient EM power is among the most common sources of ambient energy which is not tapped for powering remote wireless sensors. Since ambient energy density from EM waves is diverse in frequency and the lowest among ambient renewable energy sources [1], there is the need to have both multiband/broadband capabilities of the realized wireless EM harvester and relatively high harvesting efficiency for possible applications in powering remote sensors. There is increasing availability of ambient EM radiation at several frequencies as reported by [2], hence harvesting these ambient energy needs harvesters which can operate at multiple frequencies at the same time. The accumulation of EM energy from various frequencies will increase the energy density of an EM wireless harvester and make it a viable alternative to other existing techniques discussed in [1]. Ungan [3] have presented high quality (Q) EM wireless harvester with high sensitivity at 24 MHz. The harvester s functionality is drastically degraded when the operating frequency is changed from the optimal resonance frequency. Le [4] and Umeda [5] have presented EM wireless harvesting systems based on CMOS processes which provide better integration and has advantages for batch production of the harvester. The drawback of most CMOS based RF to DC circuits is the need for some form of gate biasing at low input power levels which makes such systems not completely passive. Wu [6] has presented serially connected wireless EM harvesters to improve the sensitivity of the RF to DC circuit at 9 MHz. By serially connecting the output voltages of independent harvesters, the sensitivity of the overall system was improved. By using up to six different harvesters the overall size of the reported harvester was large and may not be convenient for applications where harvester size is a constraint. Hagerty [7] presented rectenna arrays for broadband ambient EM harvesting and characterized the harvesters from 2 GHz to 18 GHz. Rectennas eliminates the need for separate impedance matching network but its realization is simulation intensive and can be less predictive when realized. The array of rectennas increases the overall size of the EM harvester as well. In this paper, the scavenging of EM power with wireless EM harvester capable of harvesting EM waves from multiple frequencies will be presented. The harvester can harvest EM radiation in the frequency range of 6 MHz to 2 GHz. This is made possible by using a single PCB planar multiband antenna and broadband RF to DC circuit using broadband impedance matching of schottky diodes. II. A. Multiband antenna design MULTIBAND ANTENNA The multiband antenna is shown in Fig. 1; the design is based on the planar PCB antenna presented in [8]. The antenna is realized on a Duriod 588, 1.57 mm substrate with dielectric constant 2.2 and a dissipation factor of around.4 (between 8 MHz and 2 GHz). 52 5 19 6.6 1.3 6.7 12.9 7 1 Fig. 1. Multiband antenna, antenna top view and dimensions in mm (left).

The antenna measures 52 mm by 5 mm. The multiple resonance frequencies are realized with a design approach where the area of the meander forming the antenna is changed intermittently. By changing the surface area of the meander at intermittent sections, it creates sections with different capacitance and inductance. The succession of different connected capacitance and inductance increases the poles of the antenna (resistor/inductor/capacitor) and results in multiple resonance frequencies. The dimensions of the intermittent meander can be changed to tune the resonance frequencies to desired values. This approach of antenna design is similar to building multiple resonant networks with passive components as in [9] and [1]. FEM analysis using high frequency structural simulator (HFSS) software was used simulate the behavior of the antenna. B. Multiband Antenna S 11 and impedance The measured and simulated reflection coefficient (S 11 ) of the proposed multiband antenna is shown in Fig. 2. The antenna has several resonant frequencies between 7 MHz and 8 GHz. Fig. 3 shows measured resistive and reactive impedance of the multiband antenna. Resonance occurs when the antennas impedance is purely resistive. The impedance at the various resonance frequencies varies from 3 Ω to 5 Ω. C. Multiband Antenna gain Fig. 4 shows HFSS simulated far field radiation pattern of the multiband antenna at 1 GHz. The maximum simulated gain is almost 6 dbi at about 7 in the plane of the antenna. For the angle in the plane of the antenna, the simulated gain is around -2 dbi. The gain of the antenna is different at different angles in the plane of the antenna and for different resonant frequencies. -9-6 -3 electric 4. 2.. -2. 3 magnetic 6 9 C db(gain Setup1 : Sw e Freq='1.3 db(gain Setup1 : Sw e Freq='1.3 1 DB( S[1,1] ) Measured DB( S[1,1] ) HFSS Simulation -12 12 SII (db) -1-2 -3-4 Fig. 2. Measured (Δ) and HFSS simulated ( ) reflection coefficient of the proposed multiband antenna. Impedance (Ohm).2 2.2 4.2 6.2 8 6 3-3 -6 Im(Z[1,1]) Multiband Antenna Re(Z[1,1]) Multiband Antenna.7 2.7 4.7 6.7 8-15 -18 Fig. 4. Far field simulated electric and magnetic radiation planes at a frequency of 1 GHz. Maximum antenna gain is about 6 dbi around 7 in the antenna plane. At the gain is about -2 dbi. The gain of the proposed multiband antenna was measured based on the Friis transmission equation. The equation is given in equation 1. P r = G P r G r ( λ t 4πR )2 (1) where P r is the received power, P t is the transmitted power. G t and G r are the gain of transmitting and receiving antenna respectively, R is the distance between the transmitting and receiving antenna and λ is the wavelength of the EM wave. Equation 1 is valid, if the receiving antenna is at far-field from the transmitting antenna. For the gain measurements, the farfield is taken as R λ. For a transmit/ receive arrangement using the same transmitting and receiving antenna in the same orientation, the Friis equation can be written in logarithmic form as equation 2, since the transmitting and receiving antenna gains can be equated. 15 Fig. 3. Measured resistive (Δ) and reactive ( ) input impedance of the multiband planar antenna.

Received power (dbm) G = 1 [2log (4πR ) + 1log 2 λ (P r )] (2) P t where G is the gain of the transmitting or receiving antenna in dbi. The antennas were arranged vertically to each other (Fig. 4; - -18 plane) for each measured gain. The receiving antenna was positioned at a distance not less than 1 m from the transmitting antenna. Since the angle (orientation) of the antennas were fixed in a plane of the antenna and R from transmitter increased, the measured gain were not necessarily the maximum gain but the gain in the measured - -18 plane (see Fig. 4) at the various frequencies. Fig. 5 shows a plot of received power from the same transmitting/receiving multiband antenna, transmitting 14.5 dbm of power for various received distances at a frequency of 1 GHz. This is ideal for wireless EM harvesters since their input power densities are generally low. A 5 Ω resistive source and -3 dbm input reference (from antenna) was chosen for RF to DC broadband impedance matching. Since the antenna has different induced power based on different antenna distances or orientation, and different impedance at different resonant frequencies, the impedance matching is not perfect for all the operating frequencies and become increasingly poor when the input power is not -3 dbm from a 5 Ω resistive source. These imperfections of the matching circuit at different power levels and frequencies are accepted without changes to the matching circuit. The input impedance of the HSMS-285C diode is first measured at -3 dbm for a 5 Ω source so it can be impedance matched. The measuring board is as shown in Fig. 6. -2 Distance (m).8 1.3 1.8 2.3 2.8 3.3 3.8-3dBm 5Ω HSMS- 285C 1pF 1MΩ SMA HSMS-285C 1MΩ load -25-3 -35-4 Fig. 5. Received antenna power versus distance for the same transmitting (-14 dbm) and receiving multiband antenna at 1 GHz. The antennas were arranged vertically (Fig. 4; - -18 plane) to each other. 1cm 1pF capacitor PCB Fig. 6. Equivalent circuit (left) and PCB (right) to measure the input impedance of the HSMS-285C diode. The SubMiniature version A (SMA) is soldered directly on the input pin of the HSMS-285C diode to eliminate impedance effect introduced by additional lead or copper route. The board is then connected to a network analyzer and the reflection coefficient (S 11 ) measured at -3 dbm for frequencies of up to 3.7 GHz. Fig. 7 shows measured impedance of the HSMS- 285C from.7 GHz to 3.7 GHz at -3 dbm. TABLE I. The received power was then fit to a second order polynomial with respect to the distance R and the corresponding antenna received power and distance read for the gain calculation using equation 2. The measured gain is -1 dbi in the vertical plane. This compares well to the HFSS simulated gain of about -2 dbi in this plane. Table 1 shows the measured gain for various resonant frequencies in the - -18 plane. III. MEASURED GAIN IN VERTICAL PLANE 1. 1.6 1.8 2.3 2.7 6.4 Measured Gain (dbi) -1..5-2.5 -.8 -.2-3. BROADBAND RF TO DC CIRCUIT A. RF to DC circuit design The broadband RF to DC circuit is realized with a back-toback L-circuit in-between the antenna input port and a schottky diode HSMS-285C from Avago [11]. The HSMS- 285C diode is a zero biased series connected diode pair with low turn on voltage optimized for small signal applications. Impedance (Ohm) 5-1 -2-3 Re(Z[1,1]) HSM285C Im(Z[1,1]) HSM285C.7 1.7 2.7 3.7 Fig. 7. Measured impedance (Δ resistive, capacitive) of the HSMS-285C diode at -3 dbm input from a 5 Ω source. The measured impedance of the HSMS-285C at -3 dbm input at 2 GHz is 2.5 Ω j 9 Ω. This is equivalent to a 2.5 Ω resistor in series with a.88 pf capacitor.

L-circuit matching converts source series impedance to the equivalent load parallel impedance or vice-versa and tunes out any stray reactance with its counter impedance. Series impedance is converted to its parallel equivalent impedance with equations 3, 4 and 5, 6. R p = (Q 2 + 1)R s (3) For Q greater than 1 [12], R p Q 2 R s and X p X s (4) where R p is the parallel resistance and R s is the series resistance. The Q for series and parallel impedance are given by equation 5 and 6 respectively. Q = Xs Rs, (5) Q = Rp Xp, (6) where X s and X p are the equivalent series and parallel impedance respectively. The loaded Q of an L-circuit matching network can be predicted from equation 3. By using back-to-back L-circuit matching, maximum circuit band (minimum Q) can be achieved. A minimum Q is achieved when a virtual resistance is imagined between successive L- matching sections of the multiple L-matching networks. This virtual resistance provides virtual load to the successive L- sections for matching using equations 3, 4, 5 and 6. The virtual resistance is given in equation 7 [12], R = R s R pp (7) where R is a virtual resistance between successive L-sections of the matching network and R pp is the load resistance. For the purpose of this work, L-matching is made with inductors connected only in series and capacitors shunt. Resistors are not used for tuning. This is preferred so that decent sensitivity and efficiency can be achieved at the RF to DC circuit output without power seeping into any shunt inductors or resistors used for tuning. The drawback for constraining the tuning process to only series inductor and shunt capacitor is that, for loads with large shunt capacitance (diodes for example), it is sometimes impossible to tune out the shunt capacitance provided by the diode by using inductors only in series and capacitors in shunt for certain source impedances. This drawback notwithstanding, series inductors and shunt capacitors still provide better sensitivity and efficiency for diode based RF to DC circuit conversion than no impedance matching or using inductors and resistors as shunt. Bouchouicha [2] presented a high Q RF to DC circuit using shunt tuning inductors. The return loss at the input port was -3 dbm but resulted in a low circuit efficiency of.6 % at -42 dbm. The equivalent parallel impedance of the HSMS-285C diode at 2 GHz and -3 dbm is 324 Ω resistive and 9 Ω (.88 pf at 2 GHz) capacitive. Any broadband matching network ought to equate this resistive and shunt capacitive impedance provided by the diode with the counter impedance while at the same time lowering the system loaded Q. Fig. 8 shows 5 Ω source back-to-back L-circuit (broadband) matching to a 324 Ω (HSMS-285C resistive impedance at -3 dbm and 2 GHz) load at -3 dbm. Without the load s (diode) shunt capacitance, broadband matching with a loaded Q of 2.6 around 2 GHz is achieved. The same is true in the case where the tuning capacitance C2 is greater than the diodes shunt capacitance so that a resultant C2* is obtained by subtracting the diodes shunt capacitance from the calculated original shunt capacitance C2. But since the tuning capacitor C2 in the second stage L-matching cannot tune out (less than) the diode capacitance, the resulting broadband matching network produces a non-perfect matching around 2 GHz at -3 dbm to the HSMS-285C diode. 2GHz@ -3dBm 5Ω Back to back L-circuit L1=1nH C1=.5pF L2= 83nH C2=.6pF HSMS-285C @ 2 GHz, -3 dbm 9 Ω,.88 pf diode shunt capacitance 324 Ω Vout Fig. 8. Broadband RF to DC circuit around 2 GHz with backto-back L-circuit matching of HSMS-285C diode at -3 dbm. This difficulty of matching diodes impedance becomes even greater when diodes are connected in parallel as in multipliers. The shunt capacitances are added up and become difficult to match when one restricts itself to using inductors only for series tuning. The broadband RF to DC circuit was adjusted from this quasi-perfect match shown in Fig. 8 by lowering the reference source impedance and going through the tuning process again to see if acceptable broadband impedance matching around 2 GHz will be achieved. PCB side view Circuit diagram Impedance matching 1 mm Vout_2 Vout_1 2 nh 15 nh.5 pf via HSMS-285C Vout_2 Vout_1 1 pf 1 pf Fig. 9. PCB layout of the broadband RF to DC matched circuit. Top left is the circuit diagram, bottom left indicate the side view of the double-sided PCB. via via

Additionally the fixed diodes shunt capacitance is taken as second shunt capacitance (replace C2 in Fig. 8) in the back-toback L-matching and its resistive impedance as the load, the inductors L1 and L2 are then modified for an acceptable matching to the newly lowered reference source impedance at -3 dbm. This fine tuning was carried out in Advanced Design System (ADS) from Agilent [13]. Fig. 9 shows the PCB layout and values used for the modified matching circuit. The shunt capacitance of the second L-section is the HSMS- 285C diode shunt capacitance. The copper route and the surface mount inductors provide the inductances to match the shunt capacitances. The measured circuit S 11 of the broadband RF to DC circuit is as shown in Fig. 1. There is acceptable return loss at the input port from 5 MHz to 2.5 GHz. 2 where η is the efficiency. Fig. 11 and 12 show the measured frequency sweep versus open circuit voltage and efficiency for the broadband matched circuit respectively. For -3 dbm input power, measured average open circuit voltage for the broadband matched circuit is 11 mv and.5 % efficient at 17 kω load between 8 MHz and 2 GHz. At -1 dbm input power the broadband RF to DC circuit has an average open circuit voltage of.45 V and 8 % efficient at 17 kω load between 8 MHz and 2 GHz. The measured average open circuit voltage for a non matched circuit (directly feeding source impedance to diode impedance) was 4 mv and.22 V for -3 dbm and -1 dbm respectively. These results show that the broadband circuit achieves good efficiency and sensitivity for a broad range of operating frequencies. The performance of the broadband RF to DC circuit is however lower than reported high Q matched circuit at its operating frequency [3] [8]. Open Circuit Voltage (V) S11 (db) -2-4 -6-8 -1.3 1.3 2.3 3.3 4 Efficiency (%) 3 25 2 15 1 Fig. 1. Measured reflection coefficient (S 11 ) of broadband circuit at -3 dbm input versus frequency sweep from 3 MHz to 4 GHz. B. Broadband RF to DC circuit results Open circuit voltage and efficiency for 17 kω load were measured at -3 dbm and -1 dbm input power for a frequency sweep of 8 MHz to 2 GHz. The efficiency of the RF to DC circuit is given in equation 8. η = DC output power circuit input RF power Fig. 11. Measured broadband circuit open circuit voltage at -1 dbm input versus frequency from 8 MHz to 2 GHz. Average open circuit voltage is.45 V. (8) 1.9.8.7.6.5.4.3.2.1 8 1 12 14 16 18 2 Frequency (MHz) Fig. 12. Measured broadband circuit efficiency at -1 dbm input versus frequency from 8 MHz to 2 GHz. Average efficiency is 8 % at 17 kω load. IV. WIRELESS RF HARVESTER MEASUREMENTS The EM harvester was setup in a laboratory with ambient noise from nearby objects. This is to test real world functionality of the EM harvester. Multiband antenna 14.5 dbm, 1 GHz 7cm 14.5 dbm, 1.8 GHz 5 8 1 12 14 16 18 2 Multiband antenna Distance from transmitters Frequency (MHz) EM wireless harvester Multiband antenna Broadband RF to DC circuit Fig. 13. Setup for wireless EM harvesting from different transmitting stations.

Harvestered Open Circuit Voltage (mv) Two transmitters using the proposed multiband antenna were used to send 1 GHz and 1.8 GHz signals at 14.5 dbm. The broadband harvester; consisting of the same proposed multiband antenna and the realized broadband RF to DC circuit was positioned at various distances from the transmitters. The positions of the transmitting and receiving antennas were vertical (Fig. 4; - -18 plane). The setup is as shown in Fig. 13. Fig. 14 shows the harvested open circuit voltage as a function of distance from one transmitter and the two transmitters. It can be seen that the harvester s open circuit voltage increases from receiving a single signal at 1.8 GHz to receiving both 1.8 GHz and 1 GHz signals at the same time. The proposed wireless EM harvester does harvest EM waves at different frequencies at the same time. This improves the overall performance of the harvester. The more the signals in its operating range, the more efficient the harvester functioned. The quasi sinusoidal nature of the harvested voltage is due to the travelling wave s nodes and antinodes which provide different intensity (amplitude) of the propagation radiation along its wavelength. The same nodes and antinodes phenomenon is observed for the gain measurements shown in Fig. 5. It can be deduced that the harvested voltage/power generally degrades as a function of the distance squared from transmitter (equation 1), but modulated by the nodes and antinodes intensity of the waves on its harmonic mean. 16 14 12 1 8 6 4 2 5 7 9 11 13 15 Distance from transmitters (cm) Fig. 14. Harvested open circuit voltage versus distance for 1.8 GHz, 14.5 dbm transmitter and 1.8 GHz/1 GHz, 14.5 dbm transmitters. CONCLUSION 1.8 GHz transmitter @ 14.5 dbm 1.8 GHz and 1 GHz transmitting @ 14.5 dbm A broadband wireless EM harvester is presented for wireless electromagnetic energy harvesting. The harvester is completely passive and can harvest multiple ambient EM waves from 7 MHz to 2 GHz at the same time. Measurements showed the concept of using multiband/broadband harvesting can improve the efficiency of a wireless EM harvester. Future work includes improving the sensitivity and efficiency of the broadband RF to DC circuit. ACKNOWLEDGMENT This work is part of the graduate program GRK 1322 Micro Energy Harvesting at IMTEK, University of Freiburg, funded by the German Research Foundation DFG. Special thanks to Jean-Michel Boccard and Uwe Burzlaff for Antenna measurements. REFERENCES [1] Vullers, R.J.M.; van Schaijk, R.; Doms.; I.; van Hoof, C.; Mertens, R.;, "Micropower energy harvesting, " Solid-State Electronics 53, pp. 684 693. April 29. [2] Bouchouicha, D.; Dupont, F.; Latrach, M,; Ventura, L,;, "Ambient RF Energy Harvesting," International Conference on Renewable Energies and Power Quality, pp. 1-4, March 21. [3] Ungan, T.; Le Polozec, X.; Walker, W.; Reindl, L.;, "RF energy harvesting design using high Q resonators,". IEEE MTT-S International Microwave Workshop on Wireless Sensing, Local Positioning, and RFID, 29. IMWS 29, pp.1-4, 24-25 Sept. 29, doi: 1.119/IMWS2.29.537869. [4] Le, T.; Mayaram, K.; Fiez, T.;, "Efficient Far-Field Radio Frequency Energy Harvesting for Passively Powered Sensor Networks," IEEE Journal of Solid-State Circuits, vol.43, no.5, pp.1287-132, May 28. doi: 1.119/JSSC.28.92318 [5] Umeda, T.; Yoshida, H.; Sekine, S.; Fujita, Y.; Suzuki, T.; Otaka, S.;, "A 95-MHz rectifier circuit for sensor network tags with 1-m distance," IEEE Journal of Solid-State Circuits, vol.41, no.1, pp. 35-41, Jan. 26, doi: 1.119/JSSC.25.85862. [6] Zhang Jun Wu; Wang Lin Biao; See Kye Yak; Tan Cher Ming; Boon Chirn Chye; Yeo Kiat Seng; Do Manh Anh;, "Wireless energy harvesting using serially connected voltage doublers," Microwave Conference Proceedings (APMC), 21 Asia-Pacific, vol., no., pp.41-44, 7-1, Dec. 21. [7] Hagerty, J.A.; Helmbrecht, F.B.; McCalpin, W.H.; Zane, R.; Popovic, Z.B.;, "Recycling ambient microwave energy with broad-band rectenna arrays," IEEE Transactions on Microwave Theory and Techniques, vol.52, no.3, pp. 114-124, March 24, doi: 1.119/TMTT.24.823585. [8] Antwi Nimo, Dario Grgić and Leonhard M. Reindl, Electricall small planar Antenna for compact electromagnetic wireless energy harvesting, Procedings PowerMEMS 211 Korea. pp.31-313, November 211. [9] de Queiroz, A.C.M.;, "A simple design technique for multiple resonance networks," The 8th IEEE International Conference on Electronics, Circuits and Systems, 21. ICECS 21., vol. no 1, pp.169-172 vol.1, 21, doi: 1.119/ICECS.21.95777. [1] de Queiroz, A.C.M.;, "A generalized approach to the design of multiple resonance networks," IEEE Transactions on Circuits and Systems I: Regular Papers,, vol.53, no.4, pp. 918-927, April 26, doi: 1.119/TCSI.25.859619. [11] Avago Technologies, Data sheet HSMS-285x. [12] Chris Bowick, RF circuit design (Second edition), Newness press (28), pp. 28 and 72. [13] Agilent ADS 28 Agilent Technologies. www.agilent.com.