Development and Test of a Quadrature Power Pickup for Inductive Power Transfer Systems. Priyanka Shekar

Transcription

1 Development and Test of a Quadrature Power Pickup for Inductive Power Transfer Systems Priyanka Shekar Department of Electrical and Computer Engineering University of Auckland, Auckland, New Zealand Abstract The investigation of a controller (for power regulation) for a novel power pickup configuration in Inductive Power Transfer (IPT) systems are detailed in this report. The quadrature pickup has a feature of good lateral tolerance (ability to move in the lateral domain), beneficial to vehicle applications. The controller has been explored through design, construction and testing of a system in the lab. Using this system, aspects of circuit behaviour and power flow were analysed. Most results were as predicted by technical theory, however some of the power flow results require further investigation.. Introduction. Motivation for project In an effort to improve lateral tolerance in IPT systems, the quadrature pickup configuration was proposed by the University of Auckland Power Electronics research group fairly recently. Lateral movement is particularly beneficial in Automated Guided Vehicles (AGVs) in industrial settings, and Roadway Powered Vehicles (RPVs). Investigation into the quadrature pickup thus far has been guided by this potential applicability. This project aims to supplement work done by last year s Part IV students, as there is need for investigation of a controller for the pickup.. Objectives The motivation for this project has been realised through the following objectives: To develop a computer simulation to model the circuit of the quadrature pickup-controller system. This was used to gain an initial understanding of circuit behaviour, and may be used later for design. To design, build and test two controllers with different tuning topologies, namely LC (with series and parallel tuning) and LCL topologies. 3 To study circuit behaviour of the controllers. 4 To investigate power flow in these controllers with respect to lateral displacement of the pickup. This report outlines technical concepts on the incorporation of IPT systems in AGV technology, then discusses the quadrature pickup system as a solution. The development of a computer simulation to model the system, and design and construction of LC and LCL controllers for the quadrature pickup are described. Experimental results are presented, followed by conclusions drawn.. Inductive Power Transfer in Automated Guided Vehicles. Background IPT systems transfer power through inductive coupling to galvanically isolated moveable pickup coils []. This has advantages of being environmentally clean, safe, unobtrusive, robust and maintenance-free (unaffected by chemicals, no mechanical wear) [, ]. Thus it has found industrial application including clean rooms, rail transport, AGVs and battery charging. In concept IPT systems behave like transformers, however they have a much lower magnetic coupling (greater than 9% in transformers compared to less than % in IPT systems) which makes power transfer more difficult []. Therefore, these systems require close lateral and vertical alignment to the power source to be more efficient []. As the motivation for this project mentions, the proposed pickup seeks adoption in AGVs for industrial applications. AGVs are driverless systems that transport materials and have been in existence for the past fifty years in manufacturing and warehousing and environments [3, 4].. IPT principles To illustrate the principles of IPT with a topdown approach, consider the system housed in an AGV in Figure : Power transfer occurs between a track (power source) and pickup (galvanically isolated sink) through mutual inductive coupling []. In most systems, power is transferred continuously as the pickup moves (along with the AGV) on the track. The track is an elongated loop carrying very low frequency (VLF, typically -4 khz) constant AC current, powered by mains supply. In an AGV application, it is usually buried below the floor. Pickups comprise of a coil wound on ferrite material, located on the base of the

2 IBB LBB Controller DC out ~M~ Track Pickup AGV and aligned closely with the track []. A voltage is induced on the coil due to the coupling with the track. This is boosted and regulated by a controller, made up of compensation (tuning), rectifier and switched-mode controller circuits. An output DC voltage is produced by the system to power the motors and controls of the AGV..3 Quantification of power transfer As previously discussed, a voltage is induced on the pickup coil due to the mutual inductance (M) with the track current (I ), of frequency ω. It is called opencircuit voltage (V OC ) and is given by: V OC = jωmi () Another quantity is the short-circuit current of the pickup, limited by the self-inductance of the coil (L ): MI Compen sation Rectifier Switchedmode controller Figure IPT system components in an AGV I SC = () L The uncompensated power of the pickup (S u ) is the power of the pickup alone (without the controller), found by: M Su = VOC I SC = ω I (3) L The maximum power transferable from the pickup is.5 S u. To improve this figure, compensation (tuning) is implemented which increases the uncompensated power by a quality factor (Q). The maximum compensated power transferable to the load (P) is: M P = Q Su = Q VOC I SC = Q ω I L (4).4 Lateral displacement restrictions In standard pickups either the vertical or horizontal component of the magnetic flux generated by the track is captured using a single coil aligned correspondingly (see Figure ). However, the uncompensated power is directly related to the magnetic coupling (refer Eqn 3) and is not constant as the pickup is laterally displaced. In fact, displacements exist where no power is transferred called nulls. This can be seen in Figure 3 for a single-phase bipolar track (the track type in Figure ), where the pickup is centred to the track at mm. Consequently, to maintain adequate power transfer there is little lateral tolerance. Figure 3 Uncompensated power profiles for vertical and horizontal coils (retrieved from []) In AGV and RPV systems freedom for lateral tolerance is desirable. Current AGVs are programmed to move on a fixed longitudinal path. However when turning there is insufficient magnetic coupling, and they rely on widely curved tracks and loops in tracks to improve this. RPVs have had limited commercial development as the lateral tolerance required for human drivers cannot be met. Vehicles that have been commercialised employ automatic steering control []. Current solutions for greater lateral tolerance including the following: Pickup: oversized or multiple pickups Track layout: wider tracks, three-phase tracks, meander tracks, sequential excitation of track sections Energy storage: on-board batteries to manage fluctuations in power []. 3. The quadrature pickup system Currents Vertical flux add (b) E Pickup Lateral axis Horizontal flux (b) Flat Pickup Figure Standard pickups (adapted from [])

3 3. The quadrature pickup concept As explained earlier, lateral tolerance of IPT systems need to be improved for ready adoption in vehicle applications. The concept of the quadrature configuration is that by observing the horizontal and vertical coil power profiles have complimentary nulls and peak power points, combining the two will eliminate the nulls (see Figure 4). Thus, lateral tolerance is extended. Figure 4 Uncompensated power profiles for vertical, horizontal coils and sum (retrieved from []) This is achieved by mounting both coils in space quadrature on a single flat-e pickup (see Figure 5), and designing a controller that sums the power transferred Vertical coil Horizontal coil Flat-E type ferrite Figure 5 Quadrature pickup (adapted from []) by each coil. The perpendicular arrangement of coils allows no mutual coupling between them, and therefore they operate independently allowing a simple summing of power. A flat-e pickup is used because in vehicle applications the track is usually buried so the pickup cannot extend into it, unlike the E pickup in Figure (a). Because of this the flat-e pickup has a reduced magnetic coupling factor, approximately half that of the E pickup. However, with coils in quadrature it can provide lateral tolerance more than four times that of current pickups []. 3. Past work on quadrature pickups A quadrature pickup was constructed of four flat-e cores taped together. Cores selected are easily available and commonly used commercially. However, they are designed to capture vertical flux, and therefore the power profile for the vertical coil is slightly greater than that of the horizontal coil. To achieve a uniform power profile such as that of Figure 4, the core needs to be reshaped. Using this pickup, the effects of different track layouts, angular displacement of the pickup, and height of the pickup above the track were investigated. A prototype LC topology controller was also constructed for 3 V, W output power. 3.3 Suitable controllers A pickup controller consists of compensation (tuning), rectifier and switched-mode controller circuits. As previously discussed, for a quadrature pickup the controller must sum the power output of each coil: This is done by separate tuning and rectification of the induced voltage on each coil, and then summing the DC currents to input to the switched-mode controller (see Figure 6). The switched-mode controller can use either fast switching or slow switching methods to regulate output voltage. In slow switching, the resonant voltage in the compensation circuit is completely collapsed each switching cycle. However, with fast switching this Horizontal Compen sation Controller Rectifier Figure 6 Quadrature pickup system Switchedmode controller Vertical Currents Compen Rectifier add sation resonance is only partially collapsed producing less ripple in the output voltage, but switching losses can be greater [5]. The focus of this project was to investigate two suitable controllers for the quadrature pickup with different compensation topologies, these are detailed in the following sections LC topology The LC tuning topology for a single coil is shown in Figure 7, note that the circuits for each coil sum together after the rectifier.

4 V O = V V F π ().9V after neglecting diode voltage drops V F, and then combining Eqns 5 and : Figure 7 LC tuning topology for a single coil The compensation block is essentially a tank circuit tuned to resonate at the track frequency. This is to boost the uncompensated power transferred by the pickup by the factor Q as given by Eqn 4 previously. The functions of the circuit components are as follows: C : for series tuning, it can be thought of as boosting the short-circuit current of the pickup by the current Q factor (Q I ), giving current I flowing through the pickup coil. Short-circuit current is limited by the pickup inductance, and hence this is compensated by a series capacitance [5]: I = Q I I SC (5) Where Q I is: Q I ω LC = ω L C (6) C : for parallel tuning, it is tuned with the total reactance of the pickup and C (X): ω = XC (7) Where X is: X L = ω + (8) ωc P = V I VO Q.9 I I SC () Note that the power transferred by the quadrature system is calculated by adding the power output of each coil s circuit found by Eqn. Major advantages of this topology are: It is fairly easy to control over a large range of output voltages and loads [6] Input voltage and current can be individually designed by separate Q factors, unlike the traditional parallel-tuned topology where they are fixed by the pickup inductance [6] Disadvantages include: Non-unity input power factor of the circuit this reflects a reactive impedance on the track which also causes reactive current to flow in the circuit, causing more losses [6] The large DC inductor causes harmonic currents in the circuit which also reflect onto the track causing EMI/RFI [6] 3.3. LCL topology In this topology the compensation block is a symmetrical network, also resonating at the track frequency (see Figure 8) [6]. L DC : ensures continuous current flow through the rectifier, it can be seen as a choke input filter to the switched-mode controller [6] The voltage Q factor (Q V ) is: Q V V = (9) VOC and the total Q factor for the circuit is: Q = Q Q I V () The maximum output power (P) of the system can be found from the formula for the output voltage (V O ): Figure 8 LCL tuning topology for a single coil The functions of components C and C are identical to the LC topology. The functions of the other components are as follows: L 3 : ensures continuous current flow through the rectifier [6], given by:

5 ωl V 3 O (3) X V OC where X is the same as in Eqn 8 C 3 : compensates the reactance of L 3, the rectifier and the load to maintain a symmetrical network as seen from the input and output sides of the compensation block [6]: ω C3 = (4).4674ωL X 3 The current and voltage Q factors of the circuit are: Q I = ωc ωl (5) VC Q V = (6) VOC The total Q factor is found as in Eqn. The maximum output power of the system is: P V OV πx OC = (7) Major advantages of this topology are: Unity input power factor, thus a real load is reflected on the track. This results in more efficient power transfer [6] This along with the symmetry of the network means that the voltage and current are in phase through the rectifier diodes, hence they switch when the current is zero [6] The large, expensive DC inductor has been eliminated for a smaller AC inductor [6] software. The main purpose of the simulation was to assist in understanding of the quadrature system, prior to designing a new controller. A few of the limitations of the model are listed: There is no track the only effect of the track modelled is the open-circuit voltage induced on the pickup. This was modelled as an ideal voltage source in series with the pickup inductance. In reality reactive loading of the track will induce reactive currents in the pickup circuit The simulation does not account for inductor core losses and capacitor ESR losses, as these were modelled as ideal components. Inductors were represented with external series resistances Virtually any Q factor is achievable. The practical limit of approximately due to bandwidth limitations and circuit sensitivity to component tolerances were not observed. The open-circuit voltage frequency in the model is constant, therefore the circuit cannot go out of tune, unlike in reality where the track frequency may drift 4. Experimental verification Current and voltage measurements were compared between simulation and experiment at various key points in the circuit, considering both steady-state and transient quantities. The experimental setup used was that described in Section 5.3. The results of the verification indicated that most of the simulation values fell within 5% of experimental readings. Examples of this can be seen in Figure 9. On this basis it can be recommended that this simulation may be used as at tool for quadrature LC controller design in the future. Vertical Coil Compensation - Resonant Tank Voltage Disadvantages include: A larger number of components are involved in tuning, this makes the tuning process more difficult The total number of components for tuning is greater than the LC topology: for a quadrature pickup controller, there are two more capacitors and the single DC inductor is replaced by two AC inductors. Peak voltage (V) Lateral Displacement (mm) Simulated Experimental 4. Simulation Model Development 4. Simulation design A simulation model of the existing pickup and LC prototype controller was created in LTSpice

6 Average Current (A) DC Inductor - Transient Overshoot Current (Tuned Circuit at Resonance) 4 7 Lateral Displacement (mm) Simulated Experimental Figure 9 Graphs showing comparison of experimental and simulated current and voltage 5. Quadrature Controller Development 5. Specifications and constraints LC and LCL topology slow-switching controllers were designed using principles described in Section 3.3. A slow-switching strategy was used because the LCL topology cannot implement fast-switching, so as to use the same type of switching for both designs. The design specification was for output of 3 V, 5 W over a lateral range of ±7 mm. This output voltage was chosen because it is a standard level used commercially. This output power was decided upon because commercial systems range from approximately kw upwards, and so this power is significant without being too dangerous to work with in the lab. It was also important to design a power profile as uniform as possible around the rated load, in order that excess available power is not wasted in over-design of the system. Constraints were the use of the existing quadrature pickup and a single phase khz 5A (RMS) track supply in a bipolar arrangement. Wampfler, a company producing commercial IPT systems for AGVs, originally gave both the pickup and track supply. 5.. LC topology design details The LC controller was designed using Eqn. Possible values for Q I were experimented with using a spreadsheet displaying the maximum output power profile for any Q I input. A safety factor of.333 was placed on the specified output power, giving a designed output power of W. This was done to cover for all losses in the circuit LCL topology design details The LCL topology controller design had an additional constraint placed, in that the component values of C and C be left identical to those of the LC controller. This also constrains the power output as given by Eqn 7, and therefore the LCL controller was designed for a reduced power output compared to the LC controller. Components L 3 and C 3 were designed using Eqns 3 and 4 respectively. 5. Construction Switched-mode controller Rectifier LC and LCL tuning Figure Quadrature controller implemented The controller designs were implemented on two printed circuit boards (PCBs) one holding the two compensation topologies (LC and LCL), and one holding the rectifier and switched-mode controller circuits (See Figure ). A full list of components used is given in Appendix A. Some of the major implementation decisions made are as follows: Double-layered PCBs were used for simpler routing of tracks and to allow greater current carrying capability Layout of tracks and component placement on the PCB to accommodate either an LC or LCL topology, so that two copies of the same PCB design can be produced for building each controller Note that ultimately both compensation topologies were constructed on the same PCB in a way that they could be switched in and out. This was done due to time constraints of the project and ease of debugging and adjustments to the circuit, in the sense that the remaining components were shared 5.3 Experimental setup

7 Ø regulated output voltage. When the switch turns off, allowing the tank current to flow to the load, both voltages exhibit transient oscillations due to the underdamping of the tank. Figure Setup of track, pickup and controller All experimental work was performed with the setup in Figure : The track conductor spacing was mm and the track diameter was mm. The height from the base of the ferrite on the pickup to the top of the track conductor was mm, achieved by using wood spacers. The pickup was laterally displaced in the directions shown by the green arrows in the figure. All power flow measurements were taken at mm intervals over a range of ±5 mm. These were performed at the maximum output power capability of the system, and therefore the switched-mode controller was not able to switch. All voltage and current waveforms were captured at the following important displacements: mm horizontal field null, vertical field peak 4 mm both fields approximately equal 7 mm - horizontal field peak, vertical field null mm - both fields approximately equal but much weaker than at 4 mm 5.4 Circuit behaviour observations Some examples of voltage and current waveforms captured to understand circuit behaviour are presented below with comments LC controller waveforms Figure 3 LC Resonant tank voltages, mm On Figure 3, at the horizontal field null and vertical field peak, both tank voltages still reach the steady-state clamped voltage. However, the horizontal voltage builds up monotonically in the transient period due to the very low field strength. Current (A) Time (μs) Figure 4 LC Rectifier currents, 4 mm Again at equal field strength in Figure 4, the current waveforms through each rectifier are of approximately equal magnitude and almost square in shape, as expected. The spikes observed on the edges of the waves are caused by the reverse recovery of the rectifier diodes. The two waveforms are continuous due to the effect of the DC inductor, and sum after the rectifier to produce a smoothly rippled DC current through the inductor. Figure LC Resonant tank voltages, 4mm Referring to Figure, at equal field strength, the voltages on the resonant tanks for both horizontal and vertical coil circuits are approximately equal, this is why the horizontal tank voltage waveform is sitting directly behind the vertical on the trace. This behaviour is expected, according to Eqn whereby the tank voltage is clamped by the approximately constant

8 5.4. LCL controller waveforms 5.5 Power flow results Power link budgets were performed on the quadrature pickup with both controller topologies Notes on measurements and approximations. Input power transferred from the track to the pickup was measured indirectly by: Pin = I VOC DPF (9) As can be seen in the close-up of Figure 5, the horizontal field peak and vertical field null, both tank voltages take on approximately the same steady-state value, similar to the LC topology. The transient behaviour due to the different field strengths is also similar. At the horizontal field peak and vertical field null in Figure 6, the currents through the rectifiers tend towards sine waves which is expected, and as they would if field strength was sufficient and they were being held in continuous conduction. By the design implemented, the horizontal current should be on the edge of continuous conduction at approximately this displacement. The waveform does in fact appear as the sum of a sine wave and triangle wave as it should at this point. The vertical current is very small in magnitude due to the low field strength, directly proportional to the induced voltage, and is given by the following equation: I Figure 5 LCL Resonant tank voltages, 7 mm Current (A) E E- V X OC 6.9E- 6.9E- 6.9E- Time (s) 6.9E- 6.E- 6.E- Figure 6 LCL Rectifier currents, 7 mm Rectified Sum Horizontal Rectifier Vertical Rectifier O ( AC ) = (8) It is also discontinuous as expected by calculation. The two currents sum to produce a rippled DC current. Where DPF is the displacement power factor and takes into account the phase difference between the fundamental components only. The phase angle between I and I was measured, and since the phase between V OC and I is known from Eqn, the required phase angle was determined from this. V OC was obtained previously by open circuiting the pickup coil, as it cannot be measured when the remaining circuit connected. Power (W). Back-worked efficiency is an approximation on true efficiency, and refers to: Where P loss is the total power lost across the circuit, found by measuring the currents in resistive components. This is contrasted in the results with measured efficiency: η = η = P P P out in P out + P out loss Pickup Displacement (mm) () Figure 7 LC maximum output power profile 5.5. Maximum output power Horizontal Input Vertical Input Total Input Output () Calculated Output

9 As can be seen from Figure 7, the maximum output power of the LC controller system meets the design specifications, and null points do not exist in the output power profile. Unfortunately, some time during experimental work the displacement ruler on the test jig had moved by approximately 5 mm. This is why nulls do not show up in the individual coil input powers, as they exist between two consecutive displacement measurements. The total input power was found by simply summing the inputs of the individual coils, which were measured according to the method described earlier. It is apparent from these individual coil profiles that the vertical coil contributes greater power, this is because the pickup was originally intended to capture vertical flux. The measured output power follows the calculated output very closely for the majority of displacements. Towards the ends of the graph, the true output is lower than expected. A possible reason for this may be that the Q factors in these regions are well above the practical limit of. Due to the inverse proportionality of Q and bandwidth of the tuned circuit, the frequency the circuit was tuned to may now be on the edge of the band. Power (W) Pickup Displacement (mm) Horizontal Input Vertical Input Total Input Output Designed Output Figure 8 LCL maximum output power profile The LCL controller system also meets the specifications as can be seen in Figure 8, although the design was constrained to a reduced power output compared to the LC controller. The discrepancy between the measured and calculated output starts to occur at a smaller displacement than that of the LC controller. This may be due to the high Q factors as with the LC design. In addition, the currents through the rectifiers are discontinuous in this region, but the design calculations assume continuous conduction Efficiency Efficiency Efficiency Pickup Displacement (mm) Figure 9 Measured efficiency comparison Pickup Displacement (mm) LCL Design LC Design LCL Design LC Design Figure Back-worked efficiency comparison A visual comparison of measured efficiencies for the LC and LCL controllers in Figure 9 showed no apparent difference. This is an unexpected result, as the LCL controller is known to have better efficiency due its unity input power factor. Observation of input power factors of the coils for each controller showed that power factors were between.95 and for most displacements, with little difference between the two designs. Efficiency falls off towards the fringes, and power factor follows the same pattern. Reduced efficiency may be partly because there is constant current in the resonant tank, causing losses, however the throughput to the load is greatly reduced at these displacements. Further investigation is needed in this area. The measured efficiency data plotted is not as smooth as the calculated efficiency in Figure. This is most probably due to the measurement of phase angle between currents to find the input power that could have been somewhat inaccurate with current probes available. Note that the back-worked efficiency is higher because inductor core losses, capacitor ESR and diode commutation losses were neglected.

10 5.5.4 Input power for individual coils Input Power (W) Measured Pickup Input Power Pickup Lateral Displacement (mm) Figure LC input power profile As shown in Figure, the input power to the pickup when both coils were connected in quadrature and operated is identical to each coil s individual operation, by disconnecting the other coil from the controller. Measurements were taken at the important displacements. The results verify the theory discussed earlier, that there is no mutual coupling between the coils and thus they operate independently. This allows the total input power of a quadrature pickup to be predicted easily, by summing the individual coil power inputs. 3. Conclusions Conclusions derived from this project work are as follows: A simulation model was created for the quadrature pickup with prototype LC controller. Due to its verified accuracy, it can be recommended as a tool for future controller design LC and LCL controllers were successfully built and tested to meet design specifications Upon examining controller circuit behaviour, most results were as theory predicted Preliminary investigation of power flow in the controllers was successfully performed. However, some areas need further study. 4. Acknowledgements Vertical (both coils) Horizontal (both coils) Vertical (single coil) Horizontal (single coil) I would like to acknowledge my Supervisor Assoc. Prof. Grant Covic, and Second Examiner Prof. John Boys for their direction and advice in this project. I would also like to thank PhD student Stefan Raabe for allowing us the use of his quadrature pickup, as well as his ongoing assistance. 5. References [] Boys, J.T., Covic, G.A. and Green, A.W., Stability and control of inductively coupled power transfer systems, Electric Power Applications, IEE Proceedings-, vol. 47, pp ,. [] Raabe S., Elliott, G.A.J., Covic, G.A. and Boys, J.T. "A quadrature pick-up for inductive power transfer systems" to be presented IEEE Conference on Industrial Electronics and Applications, ICIEA 7, Harbin China, 3-5 th May, 7 [3] The Basics of Automated Guided Vehicle Systems. Retrieved nd May 7 from [4] The Basics of Automated Guided Vehicle Systems. Retrieved 3 rd May 7 from [5] Covic, G.A. (7) Inductively Coupled Power Transfer (ICPT) & pick-up control. Lecture handout, Power Electronics, the University of Auckland. [6] Boys, J.T., "A new pick-up controller for parallel tuned pick-ups", Auckland Uniservices Ltd, HID/IPT research report December 4-6 Auckland, New Zealand, pp Elliott, G. A. J., Boys, J. T. and Green, A.W., Magnetically Coupled Systems for Power Transfer to Electric Vehicles in Proceedings of the International Conference on Power Electronics and Drive Systems, vol., Singapore, -4 Feb 995, pp Appendix A: LC and LCL Controller Component List LC Controller (nominal values) L (horizontal) =7.6uH L (vertical) =59.uH C (horizontal) =6nF C (vertical) =33nF C (horizontal) =nf C (vertical) =97nF Switch: IRGPB6PD IGBT Rectifier diodes: HFA6PB LC Controller (nominal values) L 3(horizontal) =.mh L 3(vertical) = 843uH C 3(horizontal) = 54nF C 3(vertical) = 69nF Other components were identical to the LC controller.