Link Efficiency-Led Design of Mid-Range Inductive Power Transfer Systems

Link Efficiency-Led Design of Mid-Range Inductive Power Transfer Systems Christopher H. Kwan, George Kkelis, Samer Aldhaher, James Lawson, David C. Yates, Patrick C.-K. Luk, and Paul D. Mitcheson Department of Electrical and Electronic Engineering, Imperial College London, United Kingdom Power Engineering Centre, Cranfield University, United Kingdom Email: christopher.kwan9@imperial.ac.uk Abstract For mid-range inductive power transfer (IPT) systems, improving link efficiency entails operating in the multi- MHz region in order to increase coil Q factors. However, designing end-to-end systems at such frequencies poses challenges associated with the efficiency of the power electronics. This paper presents a set of design principles with the aim of achieving maximal DC-to-load efficiency of such systems. These include the selection of semiconductor devices and power converter topologies that are suitable for high frequencies. Through these design methods, a 6.78 MHz ISM-band IPT system has been implemented, transferring W of power across cm with a DC-to-load efficiency of ~7 %. Index Terms DC-to-load efficiency, converters, mid-range wireless power I. INTRODUCTION In this paper, design principles for maximising the efficiency of a mid-range inductive power transfer (IPT) system are detailed. These principles address the challenges of operating at multi-mhz frequencies due to the potential for significant switching losses of the power electronics in the MHz region. The methods include choosing appropriate types of semiconductor devices and selecting suitable converter topologies for the power electronics. Fig. shows a block diagram of a typical IPT system. Section II explains the rationale behind operating in the MHz range in order to improve link efficiency with aircore coils. Section III describes the choice of semiconductors available for high frequency power electronics. Section IV explains the use of the Class-E inverter in semi-resonant operation mode, the Class-E inverter with a saturable reactor, and the Class-EF or Class-E/F inverter, in order to drive the transmitter coil. Section V presents Class-D and Class-E rectifiers as efficient means of rectifying the high frequency AC voltage of the receiver coil. Section VI introduces load emulation in order for the IPT system to operate at maximal efficiency regardless of the actual impedance of the connected load. Section VII outlines the guidelines and regulations on exposure of humans to electromagnetic fields, which can also influence the design and usage scenarios of IPT systems. Section VIII highlights results from an implemented 6.78 MHz mid-range IPT system which was designed with these principles in mind. 6.78 MHz was selected as the operating frequency because it is the first ISM band in DC Power Supply Inverter Rectifier Load Emulation Fig.. Block diagram of an inductive power transfer system Load the MHz region; an example of an IPT application at this frequency is the charging of personal devices (e.g. mobile phones) where portability and being lightweight are highly desirable characteristics of the wireless charging system. Section IX concludes by summarising the methods described in this paper for designing mid-range IPT systems with the aim of maximising link efficiency. II. MAGNETIC DESIGN Coils with ferrite cores can be heavy and thus not very portable; in [], a khz system transferring.5 kw over 7 mm uses an H-shaped ferrite core which weighs.9 kg. Also, their directed magnetic flux leads to a restricted freedom of movement for both the transmitter and receiver sides due to the need for accurate coil alignment. Therefore, there are many situations in which air-core coils, with their wide flux coverage, are more suitable for wireless power transfer applications. In order to achieve high-q factors with an air core, MHz frequencies are necessary to maximise the coil Q factors. With the coils acting as a weakly-coupled air-core transformer, efficiency can deteriorate rapidly with distance. To maximise link efficiency, receiver resonance should be used to cancel the secondary leakage inductance, and the optimal load should be connected to the receiver s resonant tank. For a parallel resonant secondary, the optimal load is given by (), where k is the coupling factor, Q TX is the Q factor of the transmitter coil, Q RX is the Q factor of the receiver coil, ω is the frequency of operation and C RX is the secondary tank capacitance chosen to resonate with the receiver coil []. The optimal load for a series resonant secondary is given by (),

where L RX is the inductance of the receiver coil []. ( ) R opt,par = Q RX ωc RX k Q TX Q RX R opt,ser = ωl RX ( k Q TX Q RX Q RX As a result of using either form of secondary tank resonance, the maximum link efficiency can be evaluated using () []. η link = ) () () k Q TX Q RX ( k Q TX Q RX ) () From (), it can be seen that the Q factors of both the transmitter and receiver coils influence the maximum link efficiency. Therefore, to achieve improvements in the link efficiency, the Q factor of the coils should be maximised. This can be done by increasing the frequency of operation of the IPT system, but only up to a certain point after which the coils far-field radiation begins to dominate (causing the coils Q factor to drop) []. However, it does not necessarily follow that the overall efficiency of the system will be increased as well, as switching losses of the power electronics in both the inverter and rectifier circuits rise with frequency. Therefore, efficient high-frequency soft-switching power electronics are desired. These circuits, which will be described in more detail in Sections IV and V, rely also on fast devices which will be described in the next section. The advantages of air-core coils can also be seen in longer range applications (up to m between TX and RX coils) where a network of sensors with mlliwatt-level consumption could be remotely powered [5]. In these situations, the superior tolerance to angular offsets and transverse displacements compared to coils with ferrite cores is essential to being able to supply power to multiple sensor nodes concurrently. Furthermore, the reduction in size and weight of the coils due to the absence of the ferrite core means that the wireless sensors can be kept small and lightweight. III. SEMICONDUCTORS Due to the high frequency, high voltage and high current requirements of the power electronics, the task of selecting appropriate semiconductor devices is not trivial. As midrange high power IPT systems operate near the limits of the capabilities of traditional Si devices, wide-bandgap semiconductors such as SiC and GaN ought to be considered due to their superior characteristics as power devices, e.g. faster switching rates and higher breakdown field strengths. Alternatively, specialist high-speed RF MOSFETs such as those from IXYS RF can be incorporated into designs. For example, the IXYS RF IXZDFN has a total rise and fall time of 7.5 ns [6], whilst the combined rise and fall times of high-voltage MOSFETs from International Rectifier are typically at least ns [7]. An IXYS RF power MOSFET combined with gate driver (IXZDFN) was used in the Class-E inverter of the mid-range IPT system that was designed and implemented (see Section IV). Cree SiC Schottky diodes (CD7 and CD6) were used in the Class-D and Class-E rectifiers (see Section V). The packaging of the devices used in the power electronics can influence the performance of the inverter and rectifier circuits. The Cree CD7 SiC Schottky diode in the rectifier comes in a TO-7- package [8], which has long, narrow leads, adding stray inductances to the circuit. Contrastingly, the IXYS RF IXZDFN module for the inverter has a surface-mount low-inductance package which means that parasitic effects, which could potentially reduce the switching speeds, can be minimised. It also has a low intrinsic gate resistance, leading to a decrease in rise and fall times, and enabling faster switching of the devices. IV. INVERTER Conventional hard-switching inverters are not suitable for IPT systems when operating in the MHz region. Since the switching time of the devices becomes comparable to the period of the driving signal, the result is that they can be inefficient at higher frequencies. Soft-switching inverters, such as Class-D and Class-E inverters, address this issue by employing zero-voltage switching to minimise power dissipation in the MOSFET during switching. This achieved by preventing concurrent high voltage across and current through the MOSFET. A disadvantage of Class-D inverters, which are popular with low-power systems adhering to Qi or AWP standards, is that they have lower output power compared to Class-E inverters for the same input voltage and output load. Another issue is that they require a floating gate drive due to the presence of a high-side switching device. However, in contrast to Class-E inverters, Class-D inverters are able to operate over a larger load range with zero-voltage switching if the switching frequency is below the resonant frequency of the output load network. Fig. depicts the Class-E inverter in semi-resonant operation mode [9] used to drive the transmitter coil. In this topology, the transmitter resonant tank is tuned to a slightly higher frequency than the secondary resonant tank to keep the primary tank impedance inductive, a requirement for Class- E operation. The parallel combination of the capacitor C res (in Fig. ) and the transmitter coil forms an impedance transformer, which causes the load impedance to appear larger, leading to an increase in driver efficiency. Fig. shows the simulated drain-source voltage of the MOSFET of the Class- E semi-resonant inverter. Class-E inverters may also include a saturable reactor [] to tune for optimum switching operation when a change in the load occurs (see Fig. ). A saturable reactor is essentially an AC-to-AC transformer that consists of a primary and a secondary winding, both wound on a single magnetic core. It operates by applying a low DC current in one winding, which causes the magnetic core s permeability to decrease,

Fig.. Semi-resonant Class-E inverter from [9] Fig. 5. Circuit diagram of the Class-EF or Class-EF inverter. Inductor L ANSACTIONS ON POWER ELECTRONICS represents inductance of transmitting coil. R L represents reflected load seen 6 by inverter in addition to coil ESR. [] 6 6 6 TABLE I COMPARISON BETWEEN DIFFERENT RESONANT INVERTER CLASSES VDS (V) IDS (A) 6 Class-E.5768 Class-EF.556............. Class-E/F...76 time (ms) time (ms) time (ms) (a) Fig. k=.5. Simulated optimumdrain-source operation voltage for(b) semi-resonant k=. coils Class-E further apart inverter (c) k=.75 coils closer Effect of changing against thetime distance in µs between [9] the primary and secondary coils on the performance of the inverter in [], [] that adding a series LC resonant network in V parallel with the MOSFET can reduce its voltage or current stresses and therefore improve the efficiency of the inverter. RFC The added LC network is tuned to either the second or r C C third harmonic of the switching frequency. Adding resonant networks in inverters is a common technique used in Class-F C DC r r r LP r Q LS C CP and Class-F - r CS inverters to shape the MOSFET drain voltage f s Mand current waveform. This hybrid configuation of Class-E switching with a resonant R L network has been referred to as V GS C L Sat C P C L P the Class-EF LS S inverter when the added resonant LC network is tuned to the second harmonic, or the Class-E/F inverter when the added resonant LC network is tuned to the third harmonic. Fig. 5 shows the circuit diagram of the Class-EF or -E/F inverter; inductor L and capacitor C form the The Class E inverter including a saturable reactor for tuning Fig.. Class-E inverter with a saturable reactor [] added resonant network and their values are set such that their resonant frequency is either twice or three times the switching ance factor Al of nh/turn. The control windings frequency TABLE []. IFig. 6 compares the waveforms of the Classh toroids consist and therefore of 5 turns effectively giving achanging total inductance the impedance of of VALUES the second AND RANGES EF OFand SEVERAL Class-E/F PARAMETERS inverter OF THE with CLASS theeclass-e inverter. The INVERTER AND THE INDUCTIVE LINK MEASURED AT 8 KHZ H and the primary winding. windings The tuning consist procedure of turns relies giving on varying a the switching Class-EF inverter results in lower voltage stresses whereas nductance of frequency 5. µhand for the eacheffective toroid. reactance The DC control of capacitor C in Fig. the Class-E/F inverter results in lower current stresses through Component/Parameter Value ESR Value t (I C ) applied viatothe thesaturable control windings reactor. ranges from ma the MOSFET. L P 5.76 µh r LP.7 Ω ximum inductance Although to 5Class-E ma forinverters minimumcan inductance. achieve zero voltage and L S Table I shows6.69 a comparison µh r LS of the.8 normalised Ω output power o the low inductance current switching and lowoperation, number of their turns voltage of the and current stresses C S (Polypropylene) of the Half-Bridge 5.9Class-D nf ZVS, r CS the.5 Class-D Ω ZCS, the Class- R y windings, can it can be large be assumed compared thattothe other current inverters. flowing It has been reported L Range E, the Class-EF -and kωthe Class - E/F -inverters. Frequency Range. - MHz - - primary windings will not cause the magnetic core to Mutual Inductance (M) Range.7 -.7 µh - - te. Figure shows the measured effective capacitance Coupling Coefficient (k) Range. -. - - when the saturable reactor is connected in parallel with it Distance Range.5-5 cm - - nction of the DC control current. Hysteresis is observed Resonant Frequency f o 8 khz - - the core of the saturable reactor is ferromagnetic. A Component/Parameter Value ESR Value raph of the complete WPT system is shown in Fig.. Input Voltage 5 V - - Inductive Link 6 Inverter Class Normalised Output ( ) PoR L Power Vi Half-Bridge Class-D ZVS.6 Class-D ZCS.98

v DS VIN v DS VIN v DS VIN π π π π π π i S IIN i S IIN i S IIN π π π π π π (a) Class-E (b) Class-EF (c) Class-E/F Fig. 6. Comparison of normalised voltage and current waveforms of different inverters [] v in i in v Lr L r v Cr i Cr C r D r i Dr C st i Cst R dc I dc V dc be absorbed into L r. In addition, the diode-capacitor parallel combinations means that the diode s junction capacitance can be absorbed into C r. When designing a Class-E rectifier, the primary objective is to ensure that the rectifier s input impedance, R in is equal to the optimal load R opt. Since the rectifier depicted in Fig. 7 is a voltage-driven Class-E rectifier for a parallel-tuned secondary, the optimal load is given by R opt,par in (). The input impedance R in relates to the output load R dc of the Class-E rectifier by (), where M is the AC-to-DC gain [5]. Fig. 7. A voltage-driven low dv/dt class-e rectifier [] V. RECTIFIER As with hard-switching inverters, rectifiers can suffer from significant diode reverse recovery losses in the MHz region if they are hard-switched. The use of soft-switching rectifiers avoids the requirement of a hard recovery and almost eliminating the associated losses with re-establishing reverse blocking function. Class-E rectifiers are soft-switching topologies; one such circuit topology with low dv/dt that is voltage-driven is shown in Fig. 7 []. This specific Class-E topology contains an inductor L r in series with a parallel connection of a capacitor C r and a diode D r. The inductor L r is in resonance with the capacitor C r at the operating frequency of the system. Therefore, the L r -C r -D r connection provides half-wave rectification. Output filtering is performed by the first-order low-pass filter consisting of stabilising capacitor C st and the output load R dc. Any leakage inductance from the secondary coil can R dc = M R in () Fig. 8 shows the waveforms of the Class-E rectifier. The top plot is the rectifier s diode voltages, whilst the bottom plot shows the current through the diode D r (dotted) and the capacitor C r (solid). Another type of Class-E rectifier is shown in Fig. 9 []. This half-wave low dv/dt Class-E rectifier is current-driven, so it is suitable for a series tuned receiver. This rectifier consists of a capacitor-diode network connected to a second-order output filter. This filter is made up of an inductor L f, a capacitor C f and the output load R dc, and provides load independent filtering. For this rectifier, C d acts as a snubber capacitor and therefore ensures zero voltage at turn-on and turn-off and zero rate of voltage change at turn off. The relationship between the input impedance R in of this current-driven rectifier and its output load R dc is given by (5), where K I is the AC-to-DC current gain. R dc = R in K I (5)

Class-E Rectifier, Class-D Inverter. Diode Voltage [V] Diode (solid) and Capacitor (dotted) Current [A] I. INTRODUCTION The dependence of maximal IPT link efficiency on the ac resistive load connected at the receiving end has been highlighted in several 5 publications [], []. However, in a complete IPT system the ac load is represented by the input impedance of the, rectifier []. Therefore, when a rectifier is integrated to the system it not only has to be efficient at the frequency of operation,5and comply with the output type of 9.75 9.8 9.85 9.9 9.95 the receiving resonant tank (voltage output when parallel tuned and current output when series tuned) but Time it must [s] also emulate 6 the required ac resistance for maximal link efficiency. This paper compares two types of current driven rectifiers for high power IPT applications at 6.78 MHz. The rectifiers The input impedance of the topology is functionally resistive, attached to its output. Therefore, in order to have maximum link i.e. efficiency, the fundamental a load emulation frequency circuit component should be of attached the square to are suitable for connection.5 to series tuned receiving coils and voltage the output across of the rectifier, currentsosource that theisrectifier in phase (andwith the receiver the input they must emulate an ac resistance given by () in order for current resonant and tank) hence, is always the rectifier presented haswith a resistive the optimal input load, impedance. given the link efficiency to be maximised []. Furthermore, by () for aitparallel is frequency tuned secondary independent and () if the for aparasitic series tuned capacitances secondary. of theotherwise, diodes have different negligible levels impedance of current draw compared in the to ( ) k Q T x Q Rx the load impedance would detune of the the resonant magnetic tank link, capacitor. resulting The in a dc dropload in is R ac,ser = ω L Rx () 9.75 9.8 9.85 Q Rx 9.9 9.95 the efficiency only component and received thatpower. affects the input ac resistance (R in ) Such a circuit may take the form of a DC-to-DC converter, Time [s] 6 which must equal to R ac,ser when the rectifier is integrated such as the Buck converter, which can be controlled to achieve where ω the frequency of operation, L Rx the inductance to an optimal IPT loading. system. This The is following done by measuring expression the voltage relates at the the two of the receiving Fig. 8. Simulated coil, voltage-driven k the coupling Class-E rectifier coefficient waveforms: between [Top] Voltage resistances output of[]. the rectifier and dividing by the desired load in order the coils forming across D r; the [Bottom] magnetic Current through link and D r (solid) Q T x and and C r (dotted) Q Rx [] the to obtain the current demand. R unloaded quality factors of the transmitting and receiving coils dc = π This R in current demand is then () compared with the actual measured current in order to give an respectively. L f therefore functions as the input current source applying power i Lf B. Class-E error signal Topology with which the duty cycle of the Buck converter can the be adjusted rectifiertounder emulate test the[7]. desired load. i Cd i D i I II. SELECTED CURRENT DRIVEN HALF WAVE dc Cf RECTIFIERS TheIncurrent addition, driven from () Class-E and (), it low candv/dt be seenrectifier that optimal (Fig. load ) is composed i in C d D v D C f R dc V varies with of athe capacitor-diode coupling factor, network which in connected turn changes to awith second A. Class-D Topology dc order distance filter. between The second the transmitter order filter and receiver includes coils. a filter Hence, inductor the The current driven Class-D half wave rectifier utilises two (L f load ), a filter emulation capacitor circuit(c should f ) and bethe controlled dc load C Rx Lin Rx (R such dc ). a Lway f ensures that diodes (D i and D ). With respect to the input current source the the flowoptimal of dcload current is always through beingremulated dc, and even C f bypasses distanceany Fig.. Class-E Low dv/dt Half Rectifier V in of Fig., D provides Fig. 9. Current a path Drivenfor Class-E the positive Low dv/dt Half partwave of the Rectifier input [] uncompressed changes to ensure ac current maximal fromlinkl efficiency. f. The capacitor (C d ) across current to flow to the dc load (R dc ) and D circulates the the diode In the acts case asof a snubber longer range capacitor. systems v in It(up Rectif provides to m ier under zero between voltage TX and RX coils) such as those for wireless sensor network test negative part of the current back to the source. The filter across the diode at turn ON and it suppresses the rate of change As Although the capacitor-diode conventional network hard-switching is shunting rectifiers the input cancurrent be applications, the coupling coefficient is extremely low, which capacitor (C source inefficient f ) is large enough to ensure the voltage across of voltage and the in the second multi-mhz orderregion, output Class-D filter, rectifiers the current withflowing SiC means that during the optimal turn OFF, load varies thus little minimising with distance the switching and is the load is through Schottky dc. Furthermore, the diodesand have it capacitor shown conducts is be the the usable ac component input inac thecurrent MHz region. of losses. superimposed An on example the output of suchdc a topology current. istherefore shown in Fig. the, softwhich switching is tank impedance. In these scenarios, receiver power is typically dominated by the complex conjugate of the receiver s resonant Fig.. Rectifier Test Rig a half-wave Class-D rectifier consisting of two diodes []. It is very small (in the tens of mw range), so an open-loop type property of the topology increases conduction losses as the The real power delivered from the inverter to the rest of the also current driven, making it appropriate for series resonant of load emulation circuit is preferred; this type of circuit diode receivers. current The exceeds input impedance the resonant of this tank current-driven current, rectifier when theeliminates circuit the (average power input consumption power) of can anybeactive calculated controlby cir- therequired rms of forthe a closed-loop input square system. voltage, Examples theof acsuitable current through multiplying tankiscurrent affected is only negative by its output (Fig. load, ). Furthermore, and relationship the peak is given diodecuitry voltage by (6). during reverse bias is larger than the output voltageultra-low the resonant power open-loop tank (input loadcurrent) emulators and include the phase the buckboost measured and the flyback by the converter Power Analysis operatingutility discontinuous of the oscilloscope. difference, or and consequently, device utilisation is poorer than in the Class- D circuit. Nevertheless, soft R dc switching = π R in allows the utilisation (6) of conduction Using mode the power [6]. delivered to the dc load (output power), large diodes, slower than the frequency of operation, as reverse thevii. efficiency ELECTROMAGNETIC of the receiving FIELD end LIMITS of theand IPT system can be recovery The effects advantages are largely of the Class-D eliminated. rectifier, which include a calculated. This efficiency REGULATIONS also includes losses in the Rx coil simpler design process, lower cost implementation, higher The component stress and the input impedance of the circuit and thus, part of the inductive link efficiency. Furthermore, tolerance to DC load variation and better semiconductor In addition to the challenge of operating efficiently in the depend on the duty cycle, which is affected by R dc, C d and utilisation, are more noticeable in higher voltage operation, MHzR region, in canthe be design determined of IPTfrom systems themust values alsoofconsider input power and thewhere frequency semiconductor of operation. parasitic Theeffects Class-E arerectifier minimised. presented haslimits input on electromagnetic current. When(EM) thefield inverter levelsvoltage that are and in place input current an input impedance of a series connection between a capacitor in order are in to phase protect the humans impedance from theatadverse Rxhealth endeffects is resistive and (C in ) and a resistor VI. RECEIVER (R in ) [], LOAD [5]. When EMULATION the rectifier is addedof exposure represented to EMby fields. the input Oneresistance of these isofthermal rectifier. effects, to anas IPTshown system in (), R in (5) must and be (6), evaluated the Class-D forand maximal Class-Elinkwhich are caused by tissue heating through energy absorption B. Experimental Results efficiency rectifiersand havecan in input must resistance be taken which intodepends consideration the load whenfrom EM fields in the tissue. The other is non-thermal tuning the receiving coil. Designers have a degree of freedom in selecting a duty cycle value. The other components are hence evaluated in [] as: R dc = R in () losses, the topology has a high output power capability as the diodes are stressed to the input current during conduction and stressed to the output voltage when reverse biased, giving good semiconductor utilisation. i in D i D v D D i D v D C f i Cf R dc The experiments investigated the efficiency of the selected topologies under several input resistance designs. Cree SiC schottky diodes (CD6) were used for high power operation at 6.78 MHz. The evaluation of R ac,ser was made using the information I dc Fig.. Fig. Current. Class-D Driven Class-D Half Wave Half Wave Rectifier Rectifier [] V dc

effects, caused by the stimulation of muscles, nerves and sensory organs. Working at 6.78 MHz means that both thermal and non-thermal effects need to be taken account of when designing such IPT systems. A European Union (EU) Directive [7] was adopted on 6 June by the European Parliament and the Council of Europe, which sets out limits on the exposure of workers to EM fields. This Directive is to be transposed into UK law by July 6. Prior to the passing of this Directive, there have been no statutory limits on EM fields for both workers and the general public in the UK. The safety limits described in the directive are based on the 998 and ICNIRP limits [8], [9]. The Directive defines both exposure limit values (ELVs) and action levels (ALs). The ELVs (which ICNIRP calls basic restrictions) are quantities that are directly related to established health effects (i.e. tissue heating and nerve stimulation). These quantities, which are generally difficult to measure, must not be exceeded. The particular ELV which relates to thermal effects is the Specific Absorption Rate (SAR), whilst the ELV for non-thermal effects is the internal electric field induced in the body. At 6.78 MHz, the SAR limit is. W kg (whole body, averaged over 6-minute period and g of tissue) and the induced internal electric field limit is 576 V m (peak). Because ELVs are difficult to measure directly, the Directive also defines ALs (referred to as reference levels by ICNIRP). These external quantities, which can be measured, are external electric field and external magnetic field. At 6.78 MHz, the magnetic field ALs are µt for non-thermal effects (induced internal electric field) and. µt for thermal effects (SAR). Compliance with these ALs ensures compliance with the respective ELVs. However, if the ALs are exceeded, it does not necessarily follow that the ELVs will be exceeded as well. In these cases, further tests are needed to prove compliance with the ELVs such as performing D EM simulations. In reality, depending on the power requirements of the application, it may not be possible to deliver enough power to a load whilst at the same time keeping magnetic field levels within the EU Directive AL limits. Hence, it would be necessary to define an exclusion zone, outside of which it would be safe for humans to be physically present. Nevertheless, these EM field limits and regulations imply that it is important to design IPT systems with high link efficiencies, so the required level of power can be delivered to the receiver load with minimal magnetic field. VIII. EXPERIMENTAL RESULTS A 6.78 MHz mid-range IPT system capable of transferring W of power across a distance of cm with a DC-to-load efficiency of ~7 % has been implemented and demonstrated. This system was designed with the principles introduced in this paper. The experimental setup is shown in Fig.. The transmitter coil is a cm -turn air-core coil made from copper piping. A Class-E inverter in semi-resonant operation was selected to drive the transmitter coil, with the IXYS RF Fig.. Experimental setup of 6.78 MHz mid-range IPT system IXZDFN combined gate driver and MOSFET module used as the switching device. The receiver coil is a cm 5-turn air-core coil, also made from copper piping. The chosen rectifier topology is the voltage-driven Class-E low dv/dt rectifier, utilising the Cree CD7 SiC Schottky diode. These methods have resulted in a reduction in losses and an improvement in efficiency of the power electronics. Consequently, this mid-range IPT system is able to operate feasibly in the MHz region, leading to an increase in coil Q- factors, link efficiency and overall DC-to-load efficiency. IX. CONCLUSION The design of a lightweight and portable IPT system calls for the use of air-core coils in favour of coils with ferrite cores. The weak coupling of air-core coils suggests that the operating frequency should be increased to the multi-mhz region to maximise link efficiency. However, the efficiency of the power electronics will tend to decrease with frequency, unless suitable high frequency power converters are utilised. Inverters that are appropriate for multi-mhz frequencies include the Class-E semi-resonant inverter, the Class-E inverter with a saturable reactor, and the Class-EF or Class-E/F inverter. Class-D or Class-E rectifiers may be used to rectify the high frequency coil voltage. Different types of semiconductor devices need to be considered, including GaN, SiC and specialist high-speed RF Si devices. Receiver load emulation may be needed to maintain maximum link efficiency by emulating the optimal load seen by the rectifier. The regulations on human exposure to EM fields may also influence IPT system design and usage scenario, and ultimately encourage the design of a highly efficient system. By following the link efficiency-led design principles in this paper, a mid-range IPT system can be designed with maximum link efficiency and overall DC-to-load efficiency. ACKNOWLEDGMENT The authors would like to acknowledge the Department of Electrical and Electronic Engineering, Imperial College London for financial support.

REFERENCES [] M. Chigira, Y. Nagatsuka, Y. Kaneko, S. Abe, T. Yasuda, and A. Suzuki, Small-size light-weight transformer with new core structure for contactless electric vehicle power transfer system, in Energy Conversion Congress and Exposition (ECCE), IEEE, Sept, pp. 6 66. [] K. van Schuylenbergh and R. Puers, Inductive Powering, Basic Theory and Application to Biomedical Systems. Springer, 9. [] G. Kkelis, D. C. Yates, and P. D. Mitcheson, Comparison of current driven class-d and class-e half-wave rectifiers for 6.78 MHz high power IPT applications, in Wireless Power Transfer Conf. (WPTC), 5 IEEE, May 5. [] D. C. Yates, A. S. Holmes, and A. J. Burdett, Optimal transmission frequency for ultralow-power short-range radio links, IEEE Trans. Circuits Syst. I, vol. 5, no. 7, pp. 5, July. [5] J. Lawson, M. Pinuela, D. C. Yates, S. Lucyszyn, and P. D. Mitcheson, Long range inductive power transfer system, Journal of Physics: Conference Series, vol. 76, no., p. 5,. [6] (5, January) IXYS RF IXZDFN RF Power MOSFET & DRIVER. [Online]. Available: http://ixapps.ixys.com/datasheet/ ixzdfn.pdf [7] (5, January) International Rectifier IRFB8PbF HEXFET Power MOSFET. [Online]. Available: http://www.irf.com/product-info/ datasheets/data/irfb8pbf.pdf [8] (5, January) Cree CD7H - Silicon Carbide Schottky Diode. [Online]. Available: http://www.cree.com/%7e/media/files/cree/power/ Data%Sheets/CD7H.pdf [9] M. Pinuela, D. C. Yates, S. Lucyszyn, and P. D. Mitcheson, Maximizing DC-to-load efficiency for inductive power transfer, IEEE Trans. Power Electron., vol. 8, no. 5, pp. 7 7, May. [] S. Aldhaher, P. C.-K. Luk, and J. F. Whidborne, Tuning class E inverters applied in inductive links using saturable reactors, IEEE Trans. Power Electron., vol. 9, no. 6, pp. 969 978, June. [] Z. Kaczmarczyk, High-efficiency class E, EF, and E/F inverters, IEEE Trans. Ind. Electron., vol. 5, no. 5, pp. 58 59, Oct 6. [] S. D. Kee, I. Aoki, A. Hajimiri, and D. Rutledge, The class-e/f family of ZVS switching amplifiers, IEEE Trans. Microw. Theory Tech., vol. 5, no. 6, pp. 677 69, June. [] S. Aldhaher, G. Kkelis, D. C. Yates, and P. D. Mitcheson, Class EF inverters for wireless power transfer applications, in Wireless Power Transfer Conf. (WPTC), 5 IEEE, May 5. [] G. Kkelis, J. Lawson, D. C. Yates, M. Pinuela, and P. D. Mitcheson, Integration of a class-e low dv/dt rectifer in a wireless power transfer system, in Wireless Power Transfer Conf. (WPTC), IEEE, May, pp. 7 75. [5] A. Ivascu, M. K. Kazimierczuk, and S. Birca-Galateanu, Class E resonant low dv/dt rectifier, IEEE Trans. Circuits Syst. I, vol. 9, no. 8, pp. 6 6, Aug 99. [6] C. H. Kwan, J. Lawson, D. C. Yates, and P. D. Mitcheson, Positioninsensitive long range inductive power transfer, Journal of Physics: Conference Series, vol. 557, no., p. 5,. [7] European Union, Directive /5/EU of the European Parliament and of the Council of 6 June on the minimum health and safety requirements regarding the exposure of workers to the risks arising from physical agents (electromagnetic fields) (th individual directive within the meaning of Article 6() of Directive 89/9/EEC) and repealing Directive //EC. in Official Journal of the European Union, vol. 56, no. L79, June, pp.. [8] International Commission on Non-Ionizing Radiation Protection, Guidelines for limiting exposure to time-varying electric, magnetic, and electromagnetic fields (up to GHz), in Health Physics, vol. 7, no., 998, pp. 9 5. [9] International Commission on Non-Ionizing Radiation Protection, Guidelines for limiting exposure to time-varying electric and magnetic fields ( Hz - khz), in Health Physics, vol. 99, no. 6,, pp. 88 86.