Design and Implementation of a Modular Low-Voltage Step-Down DC-to-DC Transformer with Galvanic Isolation

Transcription

1 Design and Implementation of a Modular Low-Voltage Step-Down DC-to-DC Transformer with Galvanic Isolation by Keegan Speidel Eric Collett Jordan Lucht Shujin Qiu Final report submitted in partial satisfaction of the requirements for the degree of Bachelor of Science Electrical and Computer Engineering University of Manitoba Faculty Supervisor: Dr. Carl Ho Spring 2018 Copyright by Keegan Speidel, Eric Collett, Jordan Lucht and Shujin Qiu, 2018

2 DC TO DC STEP DOWN TRANSFORMER i Abstract The purpose of this project was to design a modular system for stepping down input voltage while simultaneously providing galvanic isolation. To accomplish this, the project needed to consist of multiple modules connected each containing a 1:1 transformer. The reason for this design was to make a system that can be expanded or modified as necessary to meet various input/output voltage ratios. An application of this project is high-speed charging of electric vehicles by connecting to a high voltage source. By reducing the voltage from the source to the rated voltage of the vehicle, the current can be increased, allowing for faster charging times. To accomplish these tasks, the project was divided into multiple sections: transformer design, PET (Power Electronics Transformer) design, DSP (Digital Signal Processor) programming and PCB (Printed Circuit Board) design. Circuitry was designed around the transformer, including a current sensor to shut down the circuit if high current was encountered, opto-couplers to isolate electrical components from each other and boost converters to change input voltages to those required by the system. All components would be integrated together on a PCB to make a single module, and multiple modules are connected together based on the desired voltage ratio. At time of writing, we cannot comment on the results of this approach due to the late arrival of our PCBs.

3 DC TO DC STEP DOWN TRANSFORMER ii Acknowledgements We would like to specially thank the following people who provided guidance and contributions to this project. We especially want to thank Dr. Carl Ho as our wonderful supervisor of this project. We would not be able to reach our project s accomplishments without your sincerity and suggestions. We would like to thank the PHD students Dong Li and King Man Siu in the Power Systems group for supporting us and providing guidance for this project. We also would like to thank Ms. Aidan Topping for providing technical writing advice for our projects. We would like to thank Professor Daniel C. Card for project design assistance. We would like to thank Kenneth Beigun and Zoran Trajkoski for guiding us in the PCB design process of this project. We would like to thank Mr. Sinisa Janjic for supplying the components from the Electrical and Computer Engineering technical shop as well as for ordering components we required for the projects. Lastly, we like to thank VITEC Electronics Corporation for providing us the halfbridge transformers and inductors.

4 DC TO DC STEP DOWN TRANSFORMER iii Contributions Shujin Qiu: Transformer designer and final report editor (Section 1) Jordan Lucht: PET module designer (Section 2) Keegan Speidel: DSP programmer (Section 3) Eric Collett: PCB design and final report editor (Section 4) All team members were involved in the testing of individual modules and the full project.

5 DC TO DC STEP DOWN TRANSFORMER iv Table of Contents Abstract... i Acknowledgements... ii Contributions... iii List of Tables... vi List of Figures... vii Nomenclature... viii Project Summary... 1 Module Design Summary... 2 Section 1: Transformer Design Procedure : Specification of Design Parameters : Calculation of Volt-Amp Rating, S : Selection of Core Material, Shape and Size : Core Flux Density and Saturation Current : Magnetic Reluctance : Determining Losses Testing: Inductance Measurements: : Performance Section 2: PET Cell Design Process: Description of Section and Requirements : Initial Plan : Changes Made : Part Selection : Power Requirements : Results Section 3: DSP Programming : DSP Requirements and Selection : DSP Design Considerations : Current Sensor Design... 34

6 DC TO DC STEP DOWN TRANSFORMER v 3.4: Final Remarks Section 4: PCB Design : Board Layout : Input and Output of PCB Modules : PCB Trace Calculations : PCB Design Software Used : Final Remarks Conclusion: References: Appendix A: Final Project Cost Breakdown Appendix B: Full Project Schematic Appendix C: Testing Proto Board Setup Appendix D: DSP Code... 56

7 DC TO DC STEP DOWN TRANSFORMER vi List of Tables Table S - 1 Proposed Performance Metrics... 1 Table 1-1: Core Characteristics... 7 Table A - 1: G08 final project cost breakdown... 53

8 DC TO DC STEP DOWN TRANSFORMER vii List of Figures Figure S - 1: System Configuration... 1 Figure 1-1: Relative core loss versus flux density Figure 1-2: Transformer equivalent circuit Figure 1-3: Transformer equivalent circuit Figure 1-4: Primary side voltage Figure 2-1: Connection scheme for PET cells Figure 2-2: Initial design of the PET cell Figure 2-3: Design change to reduce current stress on resonant capacitor. [26] Figure 2-4: IR2110 datasheet gate driver implementation [21] Figure 2-5: DSP Output/Opto-coupler Input Figure 2-6: Opto-coupler Output/IR2110 Input Figure 2-7: Top(yellow) and bottom(blue) MOSFET gate signals Figure 2-8: MOSFET load voltage, 100 V square wave Figure 2-9: Measured voltage of transformer primary and secondary winding Figure 2-10: Rectified waveform Figure 3-1: Main Code Flowchart Figure 3-2: DSP Pin Diagram Figure 3-3: Current sensor test Figure 3-4: Current sensor electrical diagram [28] Figure 4-1: Final PCB layout Figure 4-2: Module connection diagram Figure 4-3: Top Copper Layer Figure 4-4: Bottom Copper Layer Figure B - 1: Full Project Schematic Figure C - 1: Testing Proto Board Setup... 55

9 DC TO DC STEP DOWN TRANSFORMER viii Nomenclature List of symbol types used in equations. Symbol Description Use A core core cross-section area core parameter for transformer A e effective magnetic cross section design parameter for transformer A L inductance factor design parameter for transformer reluctance A N Winding cross section design parameter for transformer A R Resistance Factor design parameter for transformer A w winding window area core parameter for transformer B Flux Density design parameter for transformer B sat Saturated Flux Density core parameter for transformer d Diameter copper wire parameter f Operating frequency design requirement F magnetic field force design parameter for transformer reluctance H magnetic field Strength design parameter for transformer saturation situation I L operating current design parameter for calculating losses of transformer I pri Rated rms primary current design parameter for transformer J rms current density in the conductor design parameter for transformer k cu copper fill factor design parameter for transformer conductor L inductance design parameter for transformer reluctance l e Effective magnetic path length core parameter for transformer ln average length of the turn design parameter for calculating losses of transformer L N Average length of the turn design parameter for transformer n Transformer turns ratio design requirement N Transformer winding turns design parameter for transformer N87 Core Material Core material for transformer design P cu copper loss design parameter for the losses of transformer P v Relative Core Loss core parameter for transformer R Reluctance design parameter for calculating losses of transformer R ac ac effective resistance of the conductor design parameter for transformer conductor

10 DC TO DC STEP DOWN TRANSFORMER ix R c core reluctance core design parameter for transformer R cu conductor resistance design parameter for calculating losses of transformer R dc dc effective resistance of the conductor design parameter for transformer conductor R g reluctance of the air gap design parameter for calculating losses of transformer R tot total reluctance design parameter for calculating losses of transformer S The Volt-Amp rating design parameter for transformer T a ambient temperature design parameter T s body temperature design parameter V pri Rated rms (root-mean squared) primary voltage design parameter for transformer ϕ magnetic flux design parameter for transformer reluctance Ω TOT total resistance design parameter for calculating losses of transformer μ r, μ e Relative effective permeability design parameter for transformer reluctance η PET cell efficiency design parameter for PET cell P in Power input to PET cell measured power input to PET cell P out Power output of PET cell measured power output of PET cell V out Voltage output of PET cell measured voltage out of PET cell R load Resistance of load resistance of load List of special characters used for physical constants. Symbol Description Value μ 0 permeability of free space 4π * 10-7 H/m

11 DC TO DC STEP DOWN TRANSFORMER 1 Project Summary The goal of this project was to design and implement a low voltage, modular, DC-to- DC step-down transformer with galvanic isolation. The system steps down 300 Volts at 1 kw power input to 100 Volts with a 90% power transfer efficiency. The system consists of three identical one to one DC transformer modules. The inputs of the modules are connected in series, and the outputs are connected in parallel. With this configuration, it is possible to modify the system to add additional cells for different input voltage levels. + + V S,1 PET Cell 1 + V O,1 + V S + V S,2 PET Cell 2 + V O,2 V O + V S,3 PET Cell 3 + V O,3 Figure S - 1: System Configuration Performance Metrics Table S - 1 Proposed Performance Metrics Single Module Proposed Metric Revised Metric Tested Metric Power Rating 333 W 166 W 56 W Efficiency 90% 90% 81% Input Voltage 100 V 100 V 100 V Output Voltage 100 V 50 V 44.3 V Temperature Range 50 C to - 30 C 50 C to -30 C 50 C to -30 C Input Current 3.33 A 1.66 A A Output Current 3.33 A 3.33 A A

12 DC TO DC STEP DOWN TRANSFORMER 2 Module Design Summary All modules are identical and consist of 5 parts: a DSP (Digital Signal Processor) controller, a resonant invertor, a high frequency transformer, and a full bridge rectifier. These are all designed and integrated together on a PCB (Printed Circuit Board). The resonant convertor switches the input DC signal and feeds the switched signal into a resonant tank to produce a high frequency AC signal that can then be transformed by the transformer. This signal has a frequency of 100 khz, based on the recommended transformer operating frequency (also 100 khz). The MOSFETS used for switching the signal require a gate driver circuit to increase the switching signal to a voltage level capable of switching the MOSFETS. The transformer included in each module is a 1:1 ferrite core transformer. This transformer is used due to the low core losses at a frequency of 100 khz. Each transformer takes a 100 khz AC wave as an input and outputs an almost identical wave to the full bridge rectifier. The full bridge rectifier receives the high frequency AC voltage wave from the transformer and converts it to a DC voltage with some slight ripple. The ripple of this rectifier is controlled by a large smoothing capacitor, further smoothing the output signal.

13 DC TO DC STEP DOWN TRANSFORMER 3 The DSP controls two main functions of the circuit; the switching signals for the MOSFETS and the current sensor. The DSP outputs two PWM (pulse-width modulated) square waves that are connected to the MOSFETS via opto-couplers. The opto-couplers protect the DSP from any shorts in the power circuit. The DSP also monitors the current entering the primary side of the transformer. A Hall Effect sensor is used to isolate the DSP from the power circuit. If the measured current enters a dangerous operating range, the DSP will disable the switching signals, thereby terminating all power input into the transformer. The PCB integrates all the above systems together and connects appropriate modules closely together to reduce noise in the circuitry. The PCBs in each module consist of two layers, with most copper traces placed on the top layer of the board. Components are placed to minimize the total distance of copper traces on the PCB.

14 DC TO DC STEP DOWN TRANSFORMER 4 Section 1: Transformer Design Procedure The magnetic transformer is an indispensable part of most power electronic converter designs. Due to flux density being inversely proportional to frequency, the operating frequency can be increased to reduce transformer size. High frequency power electronic transformers are used in power grids to carry out voltage transformation; isolating loads from the power grid and improving overall power quality. For this project the main purpose of the power electronic transformer is to achieve galvanic isolation [1] between electrical subsystems by which non-direct current can flow and possess different ground potentials. In summary, this allows for galvanic isolation of a DC source from its load. Transformer design considerations [2] consisted of the magnetic material selection, electrical loss calculations and performance testing. The high frequency transformer was designed and built based on the procedure summarised in this section. 1.1: Specification of Design Parameters The design parameters for the transformer are the following: a. Rated rms (root-mean squared) primary voltage: V pri = 100 V b. Rated rms primary current: I pri = P/V = W/100 V = 3.33 A c. Transformer turns ratio: n = 1:1 d. Operating frequency: f = 100 KHz e. Maximum body temperature, Ts of the transformer and the maximum ambient temperature (Ta = 50 C) The first four parameters were determined through design calculations for the power electronic converter circuit where the transformer is used. The maximum temperatures were determined by the other temperature-limited components, the diodes and MOSFETs

15 DC TO DC STEP DOWN TRANSFORMER 5 that are used in the circuit. The operating temperature range of the transformer [3] is -40 C to 125 C. 1.2: Calculation of Volt-Amp Rating, S The Volt-Amp rating, S, of the transformer is defined as the product of rated rms voltage V pri and current I pri. S = I pri V pri (1.1) The Volt-Amp rating of the transformer is S= (100 V)*(3.33 A) = VA. Voltage in equation (1.1) can be expressed in terms of transformer design parameters such as flux density, frequency, core area and number of turns in the primary winding. Current can be expressed in terms of current density and primary conductor cross sectional area if the influence of the skin effect is neglected. Taking the winding area and copper fill factor (kcu) into consideration, the Volt-Amp rating can be expressed as shown below [4]: S = I pir V pri = N pri A core w B 2 J rms A cu (1.2) S = I pri V pri = 2.22 k cu f A core A w J rms B (1.3) 1.3: Selection of Core Material, Shape and Size Core Material Based on the operating frequency of 100 khz high frequency, a ferrite material was chosen for the core. Ferrite materials are oxide mixtures of iron and other magnetic elements that have large electrical resistivity but low saturation flux densities [5]. Due to

16 DC TO DC STEP DOWN TRANSFORMER 6 the high electrical resistivity, there is no significant eddy current loss and only hysteresis loss in the core needed to be considered. Therefore, a ferrite core was chosen for operating at high frequencies (greater than 10 khz). Core Shape Two E shaped cores were selected to make up the transformer core based on prior experience with those shapes, their cost and availability. Wire Type The conductor windings for the transformer are made from copper. Selection of the conductor type in a transformer depends on the operating frequency and the importance of the eddy current losses (winding loss) in the windings. The reason why copper conductors are the first choice for the most inductors and transformer windings is due to the high ductility of the copper, which makes it easy to bend into tight windings around the magnetic core. Additionally, the high conductivity helps minimize the total copper volume and mass required for these windings. The copper fill factor can be calculated from the copper cross-section and winding window area. The copper fill factor is used for selecting the core size and calculating the winding losses. The volume of the copper windings and the operating temperature also affect the copper winding losses and magnetic losses. This is explained in more detail in Section 1.6. Core size Core size is chosen based on the Volt-Amp rating of the desired transformer. A table of core characteristics involved in transformer design is summarized in Table 1-1 shown below. These can be found in the datasheet of the transformer core [6].

17 DC TO DC STEP DOWN TRANSFORMER 7 Table 1-1: Core Characteristics Characteristics Symbols Value Units Coil Former Winding cross-section AN mm 2 Average length of the turn LN 96 mm Resistance factor AR = R cu (N) Bobbin bw bw 8.6 mm Bobbin height hw 36.8 mm Material Core Maximum core flux density B = 0.8B sat 320 mt Effective magnetic path length l e 127 mm Effective magnetic cross section A e 280 mm 2 Inductance factor (ungapped) AL = N87 L (N) nh Relative effective permeability (ungapped) μ e 1600 Relative core loss (ungapped) P v < 3.20 (100 mt, 100 khz, 100 C) Inductance factor (1.0 ±0.05 gapped) AL 393 nh Relative effective permeability (1.0 ±0.05 gapped) μ e 141 Copper wire 18 AWG MW 35-C HY [7] W/set Diameter d mm The selected core ensures the value of S computed by equation (1.3) is larger than the designed Volt-Amp rating of VA.

18 DC TO DC STEP DOWN TRANSFORMER 8 From equation (1.3), S = I pri V pri = 2.22 k cu f A core A w J rms B = k cu R dc R ac The copper fill factor kcu can be determined based on the core and copper wire selection (0.136 unitless). Thus, the selection of the core and copper wire meet the requirements since the calculated S MAX is greater than required S. 1.4: Core Flux Density and Saturation Current Transformer performance is constrained by the magnetic flux limitation of the core. For ferromagnetic core transformers to work properly, the flux needs to be continuously changing. This change in flux induces a voltage in the secondary winding. However, ferromagnetic materials cannot handle infinite magnetic flux densities. The flux tends to saturate at a certain level, meaning that a further increase in the magnetic field does not lead to proportional increase in magnetic flux [8]. Furthermore, the flux stops increasing with increasing primary magnetic current. When the primary winding receives excessive applied voltage the magnetic current, the flux in the core reaches saturation level of a peak AC sine wave cycle. Hence, voltage induced in the secondary winding does not remain sinusoidal due to harmonics created in the secondary side of the transformer. These undesired harmonics will cause problems in the power system including overheating, power loss, reduced efficiency and even damage to the transformers. Therefore, it is important to eliminate the saturated magnetizing current during transformer design to ensure the transformer is able to function properly under the operating current. The equations used to determine the transformer core saturation magnetizing current are shown below. Choosing an air gap length of 1mm ± 0.05mm and the number of the turns actually wound of N = 52 for both the primary and secondary side of the transformer.

19 DC TO DC STEP DOWN TRANSFORMER 9 Magnetic Field Strength: H = B μ e μ 0 = A m (1.4) Saturation Current: I = H l e N = 4.41 A (1.5) The saturation current can also be calculated from the effective magnetic cross section A e and inductor factor AL, found in the data sheet of the core, as shown below. The saturation current calculated is the same as the current calculated above based on the magnetic field strength. Saturation Current: I = B A e = 4.41 A (1.6) A L N The calculations for the saturation current indicate that this transformer can support a 4.41 A current with a 1.0 mm air gap. This current meets the requirements of a 3.33 A operating current required for the design. 1.5: Magnetic Reluctance Magnetic reluctance is defined as the ratio of the magnetic field force (F) in a magnetic circuit to the magnetic flux in both AC and DC fields. The definition can be expressed as follows: R = F Φ (1.7) The reluctance of a uniform magnetic circuit can be calculated as: R = l μ 0 μ r A (1.8)

20 DC TO DC STEP DOWN TRANSFORMER 10 Multiple different sized air gaps [9] can be used in the core of the transformer to reduce the effects of saturation [10]. However, an air gap increases the total magnetic reluctance. This enables it to store more energy before N 2 core saturation becomes dominant, which effects transformer efficiency. From equation (1.8), the core reluctance is: R c = le μ 0 μ r A e = H With an air gap, l g, of 1 mm the reluctance of the air gap is: R g = l g μ 0 A e = H The total reluctance is: R tot = R c + R g = H Hence, the magnetic inductance of the transformer can be calculated: Inductance: L = N2 R tot = μh (1.9) The magnetic inductance can be calculated instead using the inductance factor given in the datasheet of the core. The AL value of a core configuration is supplied by the manufacturer. The relationship between inductance and AL in the linear portion of the magnetization curve is defined to be: Inductance: L = A L N 2 (1.10) The inductance if no air gap is present is calculated using the following. Inductance: L = N2 R C = A L N 2 = mh

21 DC TO DC STEP DOWN TRANSFORMER : Determining Losses The two main components for power electronic transformer losses that need to be considered are no-load losses and load losses [11]. No-load losses occur in the transformer core whenever the transformer is energized, it is also called iron losses or core losses. The no-load losses are composed of hysteresis losses and eddy current losses. Hysteresis losses are associated with hysteresis in the transformer core. It is caused by the movement of magnetic domains in the core being magnetized and demagnetized by alternation of the magnetic field where not all the energy of the magnetic field is returned to the power circuit when the magnetic field force is removed [12]. This energy is normally dissipated in the core as heat. These losses depend on the type of the material used to build a core, the AC flux density and the operating or switching frequency. When the core is energized, the changing magnetic fields induce circulating loops of current called eddy currents. As these current loops flow perpendicular to the magnetic axis and produce the I 2 R losses in the magnetic material known as eddy current losses [13]. The eddy currents cause the secondary magnetic field produced to oppose the applied primary magnetic field. These opposing fields tend to screen the interior of the core from the applied primary field, resulting in the total magnetic field in the core decays exponentially with distance towards the centre of the core, also known as skin effect limitations [14]. Similar to the hysteresis loss, eddy current loss also increase the temperature of the magnetic material. The magnitude of the current can be reduced by splitting the solid core into thin laminations in the plane parallel to the magnetic field. Eddy current loss can also be reduced by using a magnetic material with high resistivity such as silicon, steel, or ferrite. For this reason, most magnetic cores tend to be made of ferrite for power electronics applications.

22 DC TO DC STEP DOWN TRANSFORMER 12 Core manufacturers provide detailed information about core losses [15], usually in the form of graphs of specific loss type as a function of flux density at a given frequency and temperature as shown in the Figure 1-1. Figure 1-1: Relative core loss versus flux density The core loss specified in the data sheet of the core is smaller than 3.2 W for an ungapped core under 100 khz operating frequency and 100 mt flux density. Load losses include heat loss and eddy current loss in the primary and secondary conductors of the transformer. These losses are commonly called copper losses since most transformers use copper as the conductor [16]. The heat loss in the transformer windings is caused by the resistance of the copper conductors. These losses can be reduced by increasing the cross-section area of conductor or by reducing the winding length. However, the skin effect also occurs in the conductors used in transformer windings similar to the

23 DC TO DC STEP DOWN TRANSFORMER 13 skin effect limitation for the core. Eddy currents flow in the opposite direction to the applied current, resulting in the current density being highest at the surface and decaying exponentially inside of the conductor. This unequally distributed current under high frequency operation increases the effective resistance, thereby causing eddy current losses in the windings. The copper loss can be calculated from the resistance factor located in the data sheet. The relationship between resistance per unit length of conductor and resistance factor AN is defined as below: Resistance: R cu = A N N 2 (1.11) The total resistance of the copper windings can be calculated using the average length of the turn ln, also called mean turn length. Total Resistance: Ω TOT = A N l N N = Ω (1.12) Hence, the copper loss can be determined from the operating current. Copper Loss: P Cu = I L 2 Ω TOT = 1.57 W (1.13) Copper losses can be reduced by using materials with higher electrical conductivities like copper, using Litz wire to avoid the skin effect, increasing the crosssectional area of the conductor and improving the winding technique. 1.7 Testing: The main testing for the transformer to confirm it is functioning properly for use in the design can be divided into two parts. One part is testing the leakage inductance and magnetizing inductance of the primary side. These two inductances affect the design of the

24 DC TO DC STEP DOWN TRANSFORMER 14 resonant converter. The other part is testing the power transfer efficiency of the transformer. 1.8 Inductance Measurements: The transformer equivalent circuit is shown in figure 1-2. Figure 1-2: Transformer equivalent circuit This can be simplified to a transformer with leakage inductance and magnetizing inductance (primary and secondary inductances) for testing purposes. Figure 1-3: Transformer equivalent circuit-2

25 DC TO DC STEP DOWN TRANSFORMER 15 The equipment for testing the inductance was the HM8118 Programmable LCR bridge. First, the primary winding was tested with the secondary winding open. The secondary winding open means the primary winding was in series with the mutual inductance. The measured inductance is Ll1 + Lmp. Next, test the primary winding with the secondary winding short-circuited. The secondary winding shorted means secondary winding is shorted with mutual inductance, therefore the measured inductance is just the primary inductance, Ll1. When the first transformer was wound, the primary winding was wound first, then the secondary winding was wound on the outside of the primary winding. This resulted in the measured magnetic inductance being very large. A possible reason for this is the winding technique used created parallel wire, which caused the magnetizing inductance to behave more like a capacitor. When winding the second transformer, it was wound in a way to ensure the primary winding and the secondary winding were side-by-side. The un-gapped leakage inductance value measured was H and the magnetizing inductance was mh, which matched the calculated values of the core. In order to get a reasonable inductance value for the resonant converter design, our group received a sponsor from VITEC Electronics Corporation, which supplied us with the transformers and inductors [17]. This was due to multiple failed trials using our own handwound transformers. The inductance values from the hand-wound transformers were not suited for the project. The sample transformers from VITEC were successfully tested, and our group decided to use this transformer for the system integration. The leakage inductance value measured was 2.6 H and the magnetizing inductance was H using same measurement method described above.

26 DC TO DC STEP DOWN TRANSFORMER : Performance The two key tests for transformer performance are testing and verifying the ratio of the transformer turns and transformer performance with an AC source at our design frequency. The equipment used for testing the inductance was a function generator and AC amplifier. The function generator generated an AC sinusoidal signal that changed from low frequency to the specified design frequency, 100 khz. The signal enters the amplifier, is amplified and enters the transformer. The output of the amplifier is seen as an AC source and is connected to the primary side of the transformer. By attaching a load to the secondary side of the transformer, the performance can be tested. Figure 1-4: Primary side voltage In Figure SF4, the signal input from the function generator is in yellow. The signal for the transformer primary side voltage is in blue. The signal for the transformer secondary side voltage is in pink. Based on these testing results, we can say that the transformer turns ratio is 1:1. During testing, it was observed that the transformer can properly transfer power between primary and secondary sides of the transformer with 1 A between 60 Hz and 100 khz.

27 DC TO DC STEP DOWN TRANSFORMER 17 Section 2: PET Cell Design Process: Description of Section and Requirements The power electronic transformer (PET) circuit for our project required a modular topology capable of increasing the step-down ratio by increasing the number of cells. Each cell needed to be connected in series on the input side and in parallel on the output side. + + V S,1 PET Cell 1 + V O,1 + V S + V S,2 PET Cell 2 + V O,2 V O + V S,3 PET Cell 3 + V O,3 Figure 2-1: Connection scheme for PET cells. The PET cells were constructed identically and were designed around four requirements: 100 V DC input, 100 V DC output, 3.33A input/output current and 100 khz resonant frequency. The efficiency goal for each cell was 90%. 2.1: Initial Plan Our initial design for the PET cell was a MOSFET half bridge inverter, resonant tank, hand-wound one-to-one high frequency transformer, full bridge rectifier, and a DSP controller. A diagram is included below.

28 DC TO DC STEP DOWN TRANSFORMER V S,1 V O,1 Inverter HF Transformer Rectifier Controller Figure 2-2: Initial design of the PET cell. A half bridge inverter was the simplest way to create a 100 V square waveform, which would be smoothed into a 100 V AC waveform by the resonant tank circuit. An AC waveform would be induced on the secondary side of the transformer and converted back to DC by the full bridge rectifier. A full bridge rectifier was chosen to reduce output voltage ripple as much as possible. 2.2: Changes Made There was one change made to the initial design of the PET cell in figure 2-2. After preliminary testing at high current, module efficiency was discovered to be significantly below expected performance. Upon further research, it was discovered that a different topology could increase our efficiency [26]. An additional resonant capacitor was connected between the source and transformer primary winding. This reduced current stress on the resonant tank circuit. By reducing the size of capacitors by 50%, the resonant frequency was not affected. [26]

29 DC TO DC STEP DOWN TRANSFORMER 19 Figure 2-3: Design change to reduce current stress on resonant capacitor. [26] The control system was also changed and required multiple iterations. Initially, the inverter MOSFETS were controlled directly by a micro controller which was programmed to output two square waves with a 180⁰ phase difference and 50% duty cycle. There were multiple problems with this design. The first issue encountered was that the top MOSFET could not be driven properly during conduction. As soon as conduction begins, the source voltage becomes ~100 V, meaning the gate signal of the top MOSFET must be Vth + 100V. This introduced a need for a gate driver circuit capable of supplying ~110 V to the top MOSFET, which we found in the IR2110 [21] with a boost-trap-cap configuration. Figure 2-4: IR2110 datasheet gate driver implementation [21]

30 DC TO DC STEP DOWN TRANSFORMER 20 The IR2110 chip is supplied by 15 V DC (VCC) and 5 V DC (VDD) supplies, high and low control signals, and a common (COM) signal. The capacitors (a 100 uf electrolytic capacitor and 0.1uF ceramic capacitor in parallel) between VB and VS charge to 15 V, boosting VB to 115 V during top MOSFET conduction. HO will pulse voltage VB reference to COM when HIN receives a pulse. Using this technique, the top MOSFET can be controlled regardless of the source voltage. We also realized that using a gate driver circuit could address the difference in ground voltages across multiple modules. For example, in the third (top) module, the power ground is 200 V. By setting the COM pin on the IR2110 to 200 V, the HO pin can pulse the required 315 V to the top MOSFET. In addition to the gate driver circuit, the difference in ground voltages made an isolation circuit a requirement. Connecting the DSP controller to a non-zero ground was not an option for our choice of controller, therefore we needed to design a circuit that would allow us to reference the DSP to any of our power grounds. The Fairchild FODM8071 optocouplers [22] were used for both high and low MOSFETs to accomplish this. The five-pin device is supplied with 5V logic signal and zero ground from the DSP, along with 5 V DC and a reference voltage (power ground). The output is a 5 V logic signal referenced to power ground, which is then sent to the gate driver circuit. The addition of the gate driver circuit and isolation circuit also added a requirement for lower voltage supplies. Three converters were implemented to meet the voltage requirements: a 100 V DC to 15 V DC converter, a 15 V DC to 5 V DC converter, and a 120 V AC to ±15 V DC converter. The gate driver circuit requires 15 V DC and 5 V DC, the isolation circuit requires 5 V DC, and the current sensor circuit requires ±15 V DC, referenced to absolute ground.

31 DC TO DC STEP DOWN TRANSFORMER : Part Selection The requirements of 100 V DC, 3.33 A, and 100 khz resonant frequency were the main constraints which the parts of the PET cell were picked around. MOSFETs: The MOSFETs selected needed to be able to handle 100 V DC, 3.33 A, and switch at minimum 100kHz without significant losses. We selected the Infineon IPW60R07C6, which exceeds requirements in all three metrics. It is rated for 650 V DC, 34 A continuous operation, and can switch at a rate of 50 V/ns, or a 2ns switching time in our application. With each switching period being 10 us, this gives us negligible switching time. [18] Resonant Tank: The resonant capacitor value of 15 nf [19] was selected based on typical configuration schematics provided by VITEC, the manufacturer for the transformer windings. VITEC also provided the inductor which has a value of 52 uh. Combined with the transformer magnetic inductance value of 275 uh, the resonant tank has a total inductance of 327 uh. Diodes: To construct the full bridge rectifier, four diodes were used. The diodes selected were required to handle 100 V and 3.33 A at 100 khz operation. The diode we selected was the PI LXA20T600, which exceeds each metric. It can handle a peak repetitive reverse voltage of 600 V, and average currents of 20 A. The reverse recovery time is 26.5 ns at 25⁰C, much shorter than our 5 us half period. The LXA20T600 was also used in our gate driver circuit. [20]

32 DC TO DC STEP DOWN TRANSFORMER 22 Gate Driver Chip: The requirements for our gate driver circuit were the ability to pulse 15 V with a changing reference at 100 khz. As mentioned above, the IR2110 with a boost-trapcapacitor configuration can achieve this. The chip can handle up to 500 V, much greater than our 100 V requirement. A worst-case rise time of 35 ns is acceptable [21]. Opto-coupler: The isolation device selected needed to be able to receive logic signals referenced to zero ground (DSP ground) and output logic signals referenced to power ground with minimal delay. The FODM8071 met our requirements with a minimal but not negligible worst-case propagation time of 55 ns [22]. 2.4: Power Requirements The PET has three different power requirements; a 100 V DC source input, 5 V DC for the opto-couplers and gate driver chip, and 15 V DC for the gate driver. The 5 V DC and 15 V DC are stepped down from the 100 V DC source using DC to DC converters. First 100 V DC is converted down to 15 V DC, referenced to power ground. This was achieved by using the XP Power RDD08110S15 DCDC converter. This device was chosen due to its relatively low price compared to other 100 V DC step down converters ($63) and acceptable efficiency (83%). Although this efficiency is not ideal, the power requirements of the gate driver circuit are comparatively low [23]. To meet the 5 V DC requirement, the converted 15 V DC was again converted, this time down to 5 V DC. The XP Power 1205SA was used to accomplish this. Converting from 15 V DC instead of 100 V DC was desirable due to the excessive cost and poor efficiency of 100 V DC to 5 V DC converters. The XP Power 1205SA has a rated efficiency of 72%, which

33 DC TO DC STEP DOWN TRANSFORMER 23 is acceptable for this application. Like the 100 V DC to 15 V DC converter, the power used is negligible [24]. Compared to the power circuit, the power consumption of the gate driver circuit is negligible (less than 1%). The measured power consumption by the gate driver and optocouplers combined was 45 mw. To power the overcurrent protection circuit, ±15 V DC was required. The overcurrent protection circuit is covered in depth in section 3. Due to it being connected directly to the DSP, it must be isolated from the power circuit. The sensor selected [28] is an isolated hall effect sensor which prevents high power from flowing to the DSP. The sensor is grounded to neutral (DSP ground), not power ground. The sensor is powered by the Recon Powerline 120 V AC ±15 V DC converter. This converter was chosen because of the availability of 120 V AC, the acceptable efficiency of 76%, and ease of use compared to other converters [25]. 2.5: Testing Breadboard testing was conducted to ensure each stage of the design was functioning properly. After the DSP was tested to be supplying switching signals as intended, the gate driver circuit could be tested and verified. The first testing conducted was to verify the functionality of the opto-couplers. The opto-couplers receive two inverted 3.3 V PWM signals from referenced to DSP ground, with a 660 ns rest time to prevent simultaneous conduction of the MOSFETS. The opto-couplers output 5 V PWM signals referenced to power ground.

34 DC TO DC STEP DOWN TRANSFORMER 24 Figure 2-5: DSP Output/Opto-coupler Input Figure 2-6: Opto-coupler Output/IR2110 Input The next step was to verify the gate driver chip was sending the correct signals to the MOSFETs. The gate driver receives the two 5 V inverted PWM signals from the opto-

35 DC TO DC STEP DOWN TRANSFORMER 25 couplers and sends the appropriate pulsed threshold voltage to each MOSFET gate. The top MOSFET receives pulsed ~110 V and the bottom MOSFET receives pulsed ~15 V. This creates a 100 V square wave as shown below. Figure 2-7: Top(yellow) and bottom(blue) MOSFET gate signals After verifying the gate signals were correct, the output voltage was measured.

36 DC TO DC STEP DOWN TRANSFORMER 26 Figure 2-8: MOSFET load voltage, 100 V square wave After confirming a square wave of amplitude 100 V, we could begin testing the resonant tank circuit and transformer windings. Figure 2-9: Measured voltage of transformer primary and secondary winding Our created resonant waveform was not ideal and therefore contributed to the low efficiency of the PET cell. The poor performance of the resonant tank circuit contributed to the variation in the output waveform. This variation was the primary limiting factor for being unable to reach the efficiency performance metric set at the start of the project. Possible modifications to the resonant tank include changing components used and setup of the resonant tank circuit. After confirming an alternating waveform, the functionality of our rectifier was tested. Testing was done using four 171Ω resistor banks in parallel, for a load of 42.75Ω.

37 DC TO DC STEP DOWN TRANSFORMER 27 Figure 2-10: Rectified waveform As seen in the above figure, the rectifier is able to properly rectify the input signal. 2.6: Results We calculated our PET cell s efficiency during testing: η = P out P in = V 2 out R load 56.2 = V 42.75Ω 56.2 = 81% (2.1) Due to the resonant tank s poor performance, we did not meet our original performance metric of 90%. In addition to the resonant tank, the lack of PCB testing likely decreased efficiency. We also decided to reduce the PET cells output current to 1 A due to heat dissipation issues in several components.

38 DC TO DC STEP DOWN TRANSFORMER 28 Section 3: DSP Programming 3.1: DSP Requirements and Selection The DSP fulfills 2 purposes in the design. The first purpose is supplying switching signals to the MOSFETS, the second purpose is to monitor current in the power circuit. Each module contains 2 MOSFETS, each requiring a switching signal and the project consists of 3 modules. This requires the DSP to output 6 switching signals, 2 to each module. Should this system be expanded to include additional modules, each new module would also require 2 switching signals. The DSP requires an analog to digital convertor port to monitor the each transformer, requiring 3 ports total to monitor all modules. When the monitored current exceeds a dangerous operation level (4.7 A) the DSP stops switching the MOSFETs, thus stopping the power flow to the transformer. Based on the requirements, Dr. Ho recommended the TI Launchpad F28377S board. The F28377S board has 4, 20 pin connector banks, a 200 MHz clock, a 12 Bit Digital to Analog convertor, PWM output, and 1 MB flash memory. These features are sufficient for the power control system for the design [27]. The current sensor must be able to accurately measure the high frequency AC current passing through the transformer, and output a readable signal to the DSP. Additionally, the current sensor must also isolate the DSP from the power circuit. Initially we chose the LAH-25 NP Hall Effect current sensor. The LAH-25 NP current sensor fit all the requirements necessary for the design. After further investigation, we switched to the L18P005D15 current sensor instead. When comparing the 2 sensors the L18 current sensor performed similarly to the LAH-25 current sensor; our decision to change the sensor was based on cost. Per unit cost of the L18 was $13.40/sensor whereas the LAH unit cost was $32.50/sensor. Due to the need to purchase multiple sensors, the L18 was chosen

39 DC TO DC STEP DOWN TRANSFORMER 29 due to similar performance to the LAH while costing sufficiently less (Digikey prices as of Feb ). The combination of the TI Launchpad board and L18 current sensor allow for proper monitoring of the circuit at the 100 khz design frequency. 3.2: DSP Design Considerations There were several considerations made regarding commutation failure of the MOSFET switching circuit. The switching signals output from the DSP are square waves with a frequency of 100 khz. Each module requires 2 switching signals, requiring the 2 signals to have a phase difference between them. Due to the nature of the circuit, the phase difference is ~180 degrees. Similarly, the duty cycle of the square wave should be ~50% to allow for alternating between which MOSFET is active. The final duty cycle ended up being less than 50% due to several practical limitations. Firstly, the gate driver signal has rise and fall times, meaning a 50% duty cycle would result in times when both MOSFETS were active, which is undesired. Secondly, the software used for programming the DSP, Code Composer Studio had limitations in the PWM code. CCS is a free C++ compiler included with the DSP. Due to these limitations, the duty cycle was chosen to be 43.75%. This allows for a resting time between signals of μs, with commutation time for the MOSFETs of 83 ns [18]. By incorporating a phase shift between signals and a modification to the duty cycle, the chance of MOSFET commutation failure is significantly reduced. By isolating the DSP from the power circuit through the use of opto-couplers, the DSP can be powered externally by either a USB port or a 120 V AC supply. The F28377S board has a micro USB port that can be used to provide power. Another method of powering the DSP was through the use of a 3.3 V DC source. This possibility was explored using the input voltage source and some additional components to reduce the 100 V DC input to 3.3 V. This approach was not implemented for 2 reasons: 1. The need to power the DSP from input voltage was not deemed a high priority 2. The additional components were

40 DC TO DC STEP DOWN TRANSFORMER 30 incredibly costly. The choice to power the DSP externally was based on reduced cost and the ability to ensure isolation using our opto-couplers. Use of the opto-couplers in the circuit required additional changes to be made to the duty cycle of the switching signals. The opto-couplers used are active low, meaning when supplied with a high signal, the opto-couplers output a low signal. A similar situation applies when the opto-couplers are supplied with a low signal. For an opto-coupler to output a 43.75% duty cycle, the DSP must supply a 56.25% duty cycle signal. This was implemented in the DSP PWM code, due to the relative ease of modification compared to additional circuitry to handle the change. TI (Texas Instruments) Code Composer studio was used for programming the Launchpad F28377S board, and compiled using a C++ compiler. The base code used in this project was modified based on the examples provided by TI in the Power Suite program. When the 6 PWM signals are being output from the DSP, the blue LED on the F28377S is lit. When the blue LED is not lit, the PWM signals are not being transmitted, and the DSP needs to be reset before further use. A flowchart for the main code is provided below.

41 DC TO DC STEP DOWN TRANSFORMER 31 Figure 3-1: Main Code Flowchart The functions and methods used to program the DSP were modified from the code supplied by TI to fit the design parameters of the project. Examples of code used in this project are based on the PWM_Trip_zone, ADC_soc, and LED_blink examples. Pin assignment for the DSP was based off recommended pins included in the F28377S datasheet. Due to the sample code from TI using specific pins, the decision was made to connect the recommended pins to the rest of the system. This decision was made so that the supplied code was changed as little as possible.

42 DC TO DC STEP DOWN TRANSFORMER 32 TOP J1 J3 J4 J V 5v V Gnd V Gnd EPWM 2A 18 4 ADCIN EPWM 3A ADCIN EPWM 3B 14 8 ADCIN ADCIN J5 J7 J8 J V 5v Epwm 1A 0V Gnd V Gnd Epwm 1B Epwm 2B Figure 3-2: DSP Pin Diagram The PWM signals output from the DSP are generated using interrupts. As the interrupts are being processed, all connected current sensors are polled. The resolution of the analog to digital convertor included in the DSP is 12 bits. Time required for polling a current sensor is 75 ns, with polling of all 3 current sensors taking 225 ns. This results in a worst-case scenario shut-off time of 225 ns. An alternative solution was to use a 16-bit resolution analog to digital convertor, however the time required for each ADC cycle increased to 320 ns, resulting in a worst-case conversion time of 960 ns. This time was deemed acceptable when compared to the alternative circuitry which could be used as a fuse in the event over currents occurred. When using the ADC, time between samples had to be taken into account. In order to accurately sample the signal, the sampling frequency had to be twice the frequency of the highest frequency component (Nyquist Sampling Theorem). The frequency of the AC

43 DC TO DC STEP DOWN TRANSFORMER 33 voltage was 100 khz, meaning that the sampling had to occur at a frequency greater than 200 khz. From our worst-case scenario of 225 ns, the sampling frequency was found to be 4.44 MHz, much more than necessary to satisfy the Nyquist theorem. Figure 3-3: Current sensor test From Figure 3-3, we can see the DSP ADC inputted with a step function to represent an over current. The PWM continues for one cycle before shutting off. This gives a measured shutoff time of 11.6 micro seconds which is deemed acceptable for the project application. The PWM interrupt is based off the internal clock on the F28377S board. By modifying the code, the interval between consecutive clock triggers can be changed. To generate the 100 khz signal, the internal 200 MHz clock on the F28377S must be divided by pre-scalars. F osc = 200 MHz (3.1) Num To achieve an oscillation frequency close to 100 khz, the Num is chosen to be 16. This choice of Num results in an oscillation frequency of khz. The PWM interrupt

44 DC TO DC STEP DOWN TRANSFORMER 34 counts up and outputs a high signal if the counter is above a certain threshold. If the counter is below that same threshold, a low signal will be output. The flowchart for the interrupts can be seen below. As discussed prior, each pair of PWM signals on each board has a 180 degree phase difference to generate a sine wave as input to the transformer. Once the output of the transformer is rectified, some slight ripple is produced. If all 3 pairs of switching signals were in phase with each other, the resultant ripple at the output would be tripled. To minimize the ripple at the output, each pair of switching signals is offset from the other switching signals by 120 degrees. This changes reduce the output ripple of the system by a factor The code used to program the DSP is included in the appendix of this report. 3.3: Current Sensor Design The current sensor consists of 2 sides that are electrically isolated from each other. The primary side of the sensor is placed on the primary side of the transformer. The secondary side of the current sensor is powered by an external ± 15 V source and is connected to the same ground as both the DSP and ADC.

45 DC TO DC STEP DOWN TRANSFORMER 35 Figure 3-4: Current sensor electrical diagram [28] The initial placement of the current sensor was on the secondary side of the transformer. The theory was that since the transformer turns ratio was 1:1, the secondary current would be the same as the primary current. The change in location was made upon realizing that the transformer had some loss, reducing the current on the secondary side compared to the primary side. Moving the current sensor to the primary side ensured 2 things: full galvanic isolation between primary and secondary sides of the circuit and true measurement of current entering the transformer. The output from the current sensor ranges from 0 to 4 V, which is based off the measured current from 0 to 5 Amps. The ADC scales this voltage to a value between 0 and 4095, with 0 representing 0 Amps and 4095 representing 5 Amps. The DSP reads this converted number and shuts down the circuit if the number exceeds 3860 which represents a 4.7 A peak current. Should the current reach this level, the DSP disconnects the input power from the transformer since we deemed this current unsafe for operation. The current sensor requires 15 ma to operate. Total power consumed by the current sensor when powered by ± 15 V is Watts per module. This power loss will be considered in the final efficiency calculation due to the sensor being powered manually.

46 DC TO DC STEP DOWN TRANSFORMER : Final Remarks The choice of DSP and current sensor are sufficient for the tasks they are required to perform. Section 4: PCB Design 4.1: Board Layout The PCB in our project was required to have dimensions less than or equal to an 8.5 x 11 inch sheet of paper. The final dimensions of our PCB were 241 x 174 mm (9.49 x 6.85 inches), well within our stated design parameters. The final PCB is shown below. Figure 4-1: Final PCB layout

47 DC TO DC STEP DOWN TRANSFORMER 37 The colour coding in the above diagram is explained below. Red lines are copper traces on the top layer of the PCB. Green lines are copper traces on the bottom layer of the PCB. Note: due to image size, some of the smaller traces may not be visible. Red patches are surface mount areas for surface mounted components. Yellow rings describe the size and position of through holes for mounting through hole components. Thin blue lines approximate the physical size of a component on the PCB. Finally, the yellow holes in each corner of the PCB are mounting holes for attaching casings or supports to. Component placement and trace widths will be discussed in detail later in this section. In brief, components were placed to minimize total copper trace length on the PCB while remaining as close as possible to their neighboring components. During the project, our group discussed whether we would place all three identical modules on a single PCB or on multiple PCBs, and then connect the PCBs together to achieve the desired result. Our final decision was to use multiple PCBs and connect them together. Upon choosing multiple PCBs, a method of connecting all modules together to achieve our design specifications was required. To connect all modules together, the decision was made to use screw terminals. This will be discussed in further detail later in this report. The main advantage of this decision was the ability to print multiple identical PCBs. This allowed us to print smaller PCBs thereby distributing the weight from heavier components such as the transformer across multiple PCBs. In addition to weight concerns, the minimum number of PCBs we could order for printing was five. If we had decided to use one PCB for all three modules, the extra four PCBs would not have been used and the cost of printing them would be wasted. One of the driving factors for this choice was the time constraint, as the project was nearing completion. Due to the changes made throughout the project, our PCB was finalized later in the project than initially planned in our project timeline.

48 DC TO DC STEP DOWN TRANSFORMER 38 The disadvantage of printing multiple identical PCBs and connecting them together was isolating the different power grounds on each PCB from each other. This was due to the different ground levels required in each PCB. Each PCB requires a high voltage power ground, and an absolute ground power ground. The high power ground is required for components not referenced to absolute ground, to ensure that the input terminals on each PCB for stepping down our voltage could be connected together in series. At the time of writing, the PCBs have arrived, however due to their late arrival they could not be tested in time for this report. The physical layout of components on each board was designed around the gate driver chip. The reason for this was that the gate driver had the most components connected to it of all components in the circuit. Minimizing the distance from the gate driver to the components connected to it helps minimize total trace distance. The larger components on the board (the DSP, transformer and inductor) also required special considerations for where they were placed. The result of this approach was a reduction in total copper trace length on the PCB. This ensured more intact copper layers on both sides of the PCB. When placing the DSP, which pins were being used in our design had to be taken into account. Similar to the rest of the circuit, the DSP was oriented in such as way as to minimize the total trace lengths on the board. All modules of the project were driven by a single DSP. In order to connect the DSP to all modules, all PCBs must be identical and allow for selection of which pins were needed on each module. Our solution to this problem was to connect like pins together (such as the pins driving the opto-couplers), then placing a resistor between the copper trace and DSP. This allowed for selection of which pin was in use by soldering a low impedance resistor to the proper trace, and open-circuit the other paths to prevent mixed

49 DC TO DC STEP DOWN TRANSFORMER 39 signals. An added benefit of this approach was reducing the number of traces on the board, making it easier to read. The transformer and its accompanying inductor could unfortunately not be placed close together. The reason for this was that the inductor had connections to other components on the board, and placing the inductor close to the transformer would cause these other traces to become much longer than their final version. Therefore the decision was made to place the inductor farther away from the transformer in order to minimize trace lengths for components connected to the inductor. 4.2: Input and Output of PCB Modules As part of the project, our group had to determine how to connect a voltage source to the circuit and obtain the output signal. To do this, we had to determine what input signals were required and how to supply them. We concluded that 2 input voltage levels were required to power all components in our circuit. These voltage levels were selected based on output voltage to input voltage ratio of each module (1:1), and the external source required to power additional circuitry. The 2 voltage levels we decided on using were 100 V DC and 120 V DC. The 100 V input was used to supply the input voltage to be transformed, and the 100 V to 15 V boost convertor. Similarly, the 120 V input was used to power our 120 V to V boost convertor. This second boost convertor is used specifically to power the overcurrent sensor while isolating it electrically from the rest of the circuit. The boost convertors isolate their input voltage from output voltage, allowing us to decouple a voltage source from its input ground level. In order to connect the DSP to the rest of the circuit, we needed to add connectors for the DSP. The DSP has a total of 80 pins spread between 4 banks of 20 pins each. There are other pins on the DSP that are not in the banks of 20 pins, but

50 DC TO DC STEP DOWN TRANSFORMER 40 due to their positioning they did not need to be incorporated into the PCB footprint. A total of 10 pins were used in our design, specified in the above DSP section. Each of the connector banks on the DSP was connected to the PCB by a corresponding open-top, 20 pin male-to-male connector. This allows the DSP to sit on top of one PCB, so that all components on our PCB would be visible at one time. The DSP is connected to other modules through additional wires and cabling. The DSP footprint in particular had to be created based on physical measurements taken by the group, since the manufacturer datasheet did not include a spacing diagram showing the spacing of pin banks on the DSP. This was the only component where the footprint was not based on readily available supplier data. To connect external sources to the circuit on our PCB, we needed to decide which type of terminals to use. There were 2 different types of connectors we considered using: coax connectors and screw terminal connectors. The screw connector was chosen for this project. A major benefit of the coax connector was that it allowed for easier connection to external sources, since the voltage sources we were using had coax connectors on them. In the case of a single PCB containing all 3 modules of our transformer unit, the coax connector would have been the ideal choice. This is because the coax connectors would not require any intermediary connectors, and would reduce the complexity of input and output connections. However, since we decided on separate PCBs for each module we decided against using the coax connector. Connecting multiple modules together would be more difficult using coax connectors compared to screw terminals within the context of this project. To connect multiple modules together using coax connectors, multiple coax connectors would be required for input and output from each module. Using screw terminals allowed for easier connection of multiple modules together, since screw terminals can be connected together by use of wires. Therefore we could easily connect the inputs of our modules in series by running a wire from

51 DC TO DC STEP DOWN TRANSFORMER 41 the bottom of one screw terminal on one module to the top of the screw terminal of the next module. Use of these terminals allowed for similar connection of our PCB outputs in parallel. A module connection diagram is included below. Figure 4-2: Module connection diagram As shown in figure EF2, the input terminals on each module are connected together in series, so that each module takes in 1/3 of the total input voltage (300 V in this project). The outputs are connected in parallel to provide 100 V output. However, the change in ease of connecting modules together meant that extra circuitry was required to change the input signal (carried by coax cable) into a form that could be run through wires. Both of the types of connectors we considered had advantages, but we ultimately decided on using the screw terminals. We made this choice based on perceived ease of connecting modules together within the limited time remaining on the project.

52 DC TO DC STEP DOWN TRANSFORMER : PCB Trace Calculations An important design choice during this project was determining the proper trace width for all current paths in our circuit. The maximum current allowable in our circuit is 4.7 A, and the high current trace widths were calculated based on this. We decided that wide traces were not required throughout the whole circuit, since only certain paths would use high current (such as the current sensor). Note that 1 mil = mm. The trace widths chosen were 3.5 mm for high current paths and 0.5 mm for lower current paths. These widths were based on the following equations [29]: Area(mils 2 ) = ( 1 Current(amps) k (Temprise[degrees Celsius]) b) c (4.1) Where k = 0.048, b = 0.44, and c = for IPC-2221 PCB board material. Using this area, width was found based on the equation: Width[mils] = Area[mils 2 ] Thckness[ounces] 1.378[ mils oz ] (4.2) These calculations were made under the assumption that the thickness of the copper used in our PCB was going to be 1 oz/square foot, a 60 C ambient trace temperature and a maximum increase in trace temperature of 10 C. Based on these constraints, our minimum trace thickness in air was calculated to be 2.03 mm (80.0 mils) for our high current traces. Upon recommendation from Daniel Card, our traces were made approximately equal to 1 mm per Amp. This was to protect from possible over-currents in the circuit, and to reduce temperature increase under normal operation modes. Wider traces also have lower resistance and contributed inductance than thinner traces, so wider traces help reduce copper loss due to traces.

53 DC TO DC STEP DOWN TRANSFORMER 43 The layout of the circuit was designed to limit the total trace length on the PCB. However, not all components could be placed close to their neighbors, so traces were placed to try and minimize the amount of power ground they cut up. Figure 4-3: Top Copper Layer

54 DC TO DC STEP DOWN TRANSFORMER 44 Figure 4-4: Bottom Copper Layer Figures 4-3 and 4-4 show the copper coverage on the top and bottom layers of the PCB. White areas are locations where no copper is placed, and black areas are where copper is placed. When comparing the two layers, the bottom copper layer has fewer traces placed on it, allowing it to stay more intact than the top copper layer. This choice was made based on the recommendation of Daniel Card to keep one ground plane as intact as possible. In order to achieve this, the majority of traces were placed on the top layer of the PCB. In cases where traces would cross each other, one of the traces was placed through a via to run on the bottom layer of the board. The choice of which trace would be run on the bottom layer of the board was based on which trace would take less space on the bottom layer.

55 DC TO DC STEP DOWN TRANSFORMER : PCB Design Software Used The software used in the design of our PCB was Kicad, a free, open source development package including a circuit simulator, PCB layout environment, and a Gerber file generating component. The reason for using Kicad as opposed to another premium PCB design software such as Altium was the inability to make Altium work properly on a virtual machine. An attempt was made to install and run Altium on a Mac running a virtual machine, but various issues with this approach caused it to be unsuccessful. Kicad was chosen due to ease of use and setup, and potential to run on any operating system in case the PCB files needed to be passed between group members. As part of the design process, the footprints of almost all components needed to be modified to match their manufacturer specifications. In order to make these changes, a footprint was chosen that most closely matched the required component and the spacing and size of through holes or surface mounts were modified to most closely match manufacturer recommended layout. This was a disadvantage of using Kicad, the component library was not as comprehensive as other premium PCB design packages so component footprints were left up to the discretion of the PCB designer. 4.5: Final Remarks With the aforementioned details about the design process of the PCB, the resultant PCB is not the ideal PCB for this circuit. There are several reasons for this, including extensive use of modified component footprints, and non-ideal location of components in the circuit. Future work to optimize the PCB could include the use of design algorithms to place components and lay traces. By doing this, the PCB could be made smaller, and closely connected components could be located closer together, helping to improve the coherence of the PCB.

56 DC TO DC STEP DOWN TRANSFORMER 46 Conclusion: In conclusion, we were unable to meet the original specified performance metrics. Due to the current limiting characteristics of some components, we were unable to meet the original current specification. Based on our preliminary testing, we observed that the circuit did not perform as intended when supplied with greater than 1 A. A possible improvement to the project includes placing all components on the PCB as initially planned. Due to the time of PCB arrival in relation to the time of report submission, testing the circuit on a PCB was not feasible. Additionally, overall module efficiency could be improved through changes to the resonant tank components and topology. The approach used during this project indicates that this design is capable of power transformation with galvanic isolation.

57 DC TO DC STEP DOWN TRANSFORMER 47 References: [1] Jason A. Zengel. (2003, June). DC-DC Power Conversion with Galvanic Isolation, Naval Postgraduate School, Monterey, California [Online]. Available: [Feb. 12, 2018]. [2] N. Mohan, T. M. Undeland and W. P. Robbins. (2003) Design of Magnetic Components in Power Electronics Converters, Applications, and Design, 3rd ed. B. Zobrist, C. Cervoni, Eds. U.S: John Wiley & Sons. Inc pp [3] VITEC Electronics Corporation. Half-Bridge Transformer 75P8113. Carlsbad: VITEC Electronics Corporation, [4] Magnetic Component - Design and Optimization Class note for ECE7440, Department of Electrical and Computer Engineering, University of Manitoba, [5] Magnetic core in Wikipedia, Magnetic core [Online]. Available: [Feb. 12, 2018]. [6] EPCOS AG. (2017). " Ferrites and accessories ETD Core and accessories" [Online]. Available: [October 20, 2017]

58 DC TO DC STEP DOWN TRANSFORMER 48 [7] American wire gauge in Wikipedia, American wire gauge [Online]. Available: [Feb. 12, 2018]. [8] Saturation (magnetic) in Wikipedia, Saturation (magnetic) [Online]. Available: [Feb. 12, 2018]. [9] J.O. Aibangbee and S.O. Onihaebi. (2017, August. 24). Improving Current Transformers Transient Response and Saturation Effects Using Air-Gapped Core, Bells University of Technology, Ota, Nigeria [Online]. Available: [Feb. 12, 2018]. [10] R. Clarke. (2011, May. 26). Air Gap Magnetic core [Online]. Available: [Feb. 12, 2018]. [11] A. Bagglni. (2016, August). Power Transformers - Introduction to measurement of losses. INTAS, European [Online]. Available: [Feb. 12, 2018]. [12] K. Daware. Transformer - Losses and Efficiency [Online]. Available: [Feb. 12, 2018].

59 DC TO DC STEP DOWN TRANSFORMER 49 [13] Types of Losses in Transformer. Circuit globe [Online]. Available: [Feb. 12, 2018]. [14] H. Lloyd. (2003) Eddy Current Losses in Transformer Windings and Circuit Wiring. Texas Instruments Incorporated, Texas, Dallas [Online]. Available: [Feb. 20, 2018]. [15] EPCOS AG. (2013). " EPCOS DATA BOOK Ferrites and accessories" [Online]. Available: rrites-and-accessories-db pdf [October 20, 2017] [16] Copper loss in Wikipedia, Copper loss [Online]. Available: [Feb. 12, 2018]. [17] VITEC Electronics Corporation. Inductor 51P5368. Carlsbad: VITEC Electronics Corporation, [18] Infineon. (2010, Feb. 9). 600V CoolMOS C6 Power Transistor IPW60R070C6 [Online]. Available: en.pdf?fileid=db3a ff5e aa644ac5 [Nov. 25, 2017]. [19] Nichicon. (No date available). QXP Metallized Polypropylene Film Capacitor [Online]. Available: [Jan ].

60 DC TO DC STEP DOWN TRANSFORMER 50 [20] Power Integrations. (2010, Jan.). LXA20T600 [Online]. Available: [Nov. 25, 2017]. [21] International Rectifier. (2005, Mar. 23). IR2110 [Online]. Available: 7e [Dec. 4, 2017]. [22] Fairchild. (2014, Dec.). FODM8071 [Online]. Available: [Dec. 28, 2017] [23] XP Power. (2017, Oct.). RDD Series [Online]. Available: [Jan. 24, 2018] [24] XP Power. (2011, Jan.). IW Series [Online]. Available: [Jan. 24, 2018] [25] Recom. (2015). Powerline AC/DC Converter [Online]. Available: [Jan. 24, 2018] [26] ST Microelectronics. (2008, Sept.). An introduction to LLC resonant half-bridge converter. ST Microelectronics. [Online]. Available:

61 DC TO DC STEP DOWN TRANSFORMER 51 7/b7/ad/9f/4d/dd/CD pdf/files/CD pdf/jcr:content/translations/en. CD pdf [27] Texas Instruments (2014, Aug.) TMS320F2837xS Delfino Microcontrollers. [Online]. Available: [Jan. 24, 2018] [28] TAMURA. (2012, Mar.). Hall Effect Current Sensors [Online]. Available: [Jan. 24, 2018] [29]: Advanced Circuits, PCB Printed Circuit Board File Creation Calculator Advanced Circuits. [Online]. Available: [Feb. 17, 2018].

62 DC TO DC STEP DOWN TRANSFORMER 52 Appendix A: Final Project Cost Breakdown The final cost breakdown of the project is specified in table B-1 included on the following page. A total of $ was spent for the project, which is within the original proposed budget. Of the $945.76, $400 was supplied by the Department of Electrical and Computer Engineering and $ was sponsored by Dr. Carl Ho. The changes to the budget were associated with changes to design topology and the testing procedures used. The microcontroller and material for transformer windings were borrowed from Dr. Carl Ho. The transformer and inductors were sponsored by the VITEC Electronics Corporation. All sponsored components will be returned to Dr. Carl Ho after the final demonstration on March 23, 2018.

63 DC TO DC STEP DOWN TRANSFORMER 53 Table A - 1: G08 final project cost breakdown

64 DC TO DC STEP DOWN TRANSFORMER 54 Appendix B: Full Project Schematic Figure B - 1: Full Project Schematic

65 DC TO DC STEP DOWN TRANSFORMER 55 Appendix C: Testing Proto Board Setup Figure C - 1: Testing Proto Board Setup