DFIG Controller

The goal of this project is to design and implement a rotor side converter (RSC) that monitors a doubly-fed induction generators (DFIGs) torque, current and voltage output in order to command an input waveform that will stabilize the DFIG with the grid at 60Hz. The completion of this project will aid in the advancement of wind energy systems by allowing further research to be conducted on doubly-fed inductions generators.

Background
As the principal generator used in Type III wind energy conversion systems, the doubly fed induction generator (DFIG) is widely accepted in today's wind energy industry. As society shifts toward renewable energy, these systems are becoming more prevalent. This brings an increasing need for research into how these systems interact with the grid. To achieve this end, the University of Idaho has created a wind turbine simulator which uses an AC induction motor to drive a DFIG at variable speeds and monitor power delivery and stability. In order to use this test bench, a full featured controller is needed operator the DFIG properly. That is the motivation for this project. The DFIG is essentially a wound rotor induction generator in which the rotor circuit can be controlled by external devices to achieve variable speed operation. The stator of the generator is connected to the grid through a transformer, whereas the rotor connection to the grid is done through power converters, harmonic filters, and the transformer. The stator of the generator delivers power from the wind turbine to the grid. In the rotor, the power can be delivered from the rotor to the grid and vice versa through rotor-side converter depending on rotor speed, and power factors. Two of the main components in the control system of the DFIG are the grid-side converter (GSC) and the rotor-side converter (RSC). The GSC is responsible for managing power delivery to the grid and generating a stable DC rail value from the grid to power the inverters. The RSC is responsible for controlling the power applied to the rotor in order to facilitate smooth, clean, and efficient power generation. The scope of this project is assembling and testing the components to create the RSC to be used with the wind turbine simulator test bench that the University of Idaho has built.

Deliverables
The focus for the first semester is understanding the systems involved and getting components ordered and tested. The process for this is the following: Once hardware procurement and testing is completed, the plan for the second semester is the following: This outlines the process, and the final deliverables will be:
 * Learning to understand systems involved and components required
 * Developing and verifying a design for the system using simulations
 * Developing specifications and selecting hardware components based on design and verification
 * Ordering components and testing individually to ensure expected function and specification compliance
 * Perform component integration and subsystem tests
 * Develop control code and test sub routines involved
 * Complete any fabrication of custom components
 * Mount components to power electronics stand
 * Final integration and full scale tests
 * 1) A capable Microcontroller programmed with the required algorithms to run on the wind turbine simulator
 * 2) Power electronics components including Gate Driver, IGBTs, snubbers, rectifiers, etc.
 * 3) System connection components including any breakout boards and cables required to connect components
 * 4) Isolation components, in this case Optic Fiber links for noise immunity and isolation of high and low voltage components
 * 5) Sensors connected and tested in the final layout

Specifications and Constraints
Specifications for the project are listed below:
 * Prototype must be modular
 * Prototype must be robust
 * Utilize startup and fail safe procedures in code
 * Incorporate adequate isolation

Constraints of the RSC design were as follows:
 * Integration must be designed to handle a DC bus voltage value up to 340V
 * Must be able to stabilize a generator operating in a range of 0-60Hz with the grid (ie. 60Hz)
 * Read sensor inputs and command the 3-phase inverter at a minimum frequency of 20kHz

Project Outline
Below you will find the schedule that was followed to complete this project. The gantt chart shown is a broad outline of the design steps taken. A detailed version can be found in the additional documents section of this page.

= Design Evaluation and Verification =

Space Vector Modulation and Controller
Evaluation of the Space Vector Modulation (SVM) algorithm began with studying and understanding the basic concept during the learning phase. Equations for our implementation of SVM were collected and used to create a more detailed simulation that accurately tested the implementation of SVM that will be used for the physical test bench. Essentially, this test consists of utilizing SVM in current control mode, determining the PWM scheme to use, devising an algorithm for causing the fewest switching events possible, and finally testing how this interacts with an inductive load. To test active current control, a PI controller was implemented in the simulation. The three phase current measured in the load was converted to Direct-Quadrature (dq) axis rotating reference frame in order to have stable value (i.e. current values are essentially constants), then it is not difficult to implement a PI controller that can maintain the currents very close to the command, even as the command changes. Next, we decided to use center aligned PWM with three channels (one for each phase leg) with six outputs (three positive and three negative or complementary). This makes applying the derived vector straightforward and provides protection for power electronics by maintaining the switch configuration in one of the eight possible safe states. Using center aligned PWM also naturally allows us to reduce the number of switching events (by ensuring that all states are not toggled to on/off at the beginning of each cycle) while also staying in each state for the correct amount of time. In each cycle, the SVM algorithm generates a time for each of three states. These consist of two adjacent non zero states and a zero state (the remaining time for a cycle). There are two zero states where there is no net current flow through the balanced three phase load (i.e. states 000 and 111). By properly utilizing these zero states, we are able to ensure that only one phase leg changes state at a time. With the PWM waveform generated, the simulation tests how this waveform would interact with an inductive load. To test this, we generate a transfer function for each phase of the load. We feed the voltage waveform generated by PWM through the transfer function in order to obtain a current estimate. This is done in each phase and is used to calculate the overall current in the dq rotating reference. This overall current is compared to the command to generate an error function which is used by the PI controller to actively adjust the command to control the current phase and magnitude. The results of the simulation are shown in the following figures.

After completing this simulation and gaining a reasonable understanding of the control system in general, it was possible to develop a list of requirements for the microcontroller to use in the project. The notable requirements are:
 * Clock speed >100 MHz with PWM capable timers at >20 MHz in order to provide at least 9 bits of resolution at a fast switching frequency of 40 KHz (for standard PWM). *Note: 9 bits = 512 steps = max 0.66 V/step @ 340 V DC rail
 * Advanced PWM capability including 6 channels (automatic 3-phase complementary a bonus) and center aligned mode.
 * At least 6 simultaneous ADC channels with at least 12 bits of resolution
 * Various communications protocols including I2C, SPI, CAN for interfacing with other modules
 * Hardware math accelerators such as FPUs and MACs (usable without assembly wrappers is a bonus)
 * Free IDE with plentiful documentation, code samples, libraries, and developer community

Power Electronics
The first step in the design evaluation process from the power electronic side of this project was to investigate the integration of the gate driver with the IGBTs. Research during this portion led us to two possible solutions for our design: If we were to utilize the first proposal the end design would have been more compact and would demonstrate higher performance. However, this idea also required a large portion of our budget to be placed towards buying equipment in order to mount the six pack IGBT module since it was a newer IGBT housing design. After consultation with our client the second solution was chosen for implementation due to costs, as well as keeping the final design modular. Along with the choice of implementing a three-phase gate driver with three two pack IGBT modules we also decided to purchase a high-side low-side gate driver in order to conduct small scale tests without risking damaging our three-phase gate driver. After deciding on a model for gate drivers and IGBTs, simulation models for each device were able to be plugged into LTSpice simulations to verify our design before purchasing the components. Simulations conducted to validate our choices consisted of creating a single leg circuit and H-bridge circuit. The schematics used for these tests can be seen below: Following research involved with the gate driver and IGBTs the next power electronics design revolved around incorporating the protection circuits and components. Protection for our design focused on two main topics: snubbers and MOVs. Snubber circuits were chosen in order to minimize the switching losses of our design while also protecting the IGBTs by minimizing the losses of each IGBT and therefore reducing heat transfer in the devices making the design more robust. The calculations seen below were used to determine the snubber capacitor and resistor values and the inductor value was chosen by referencing the principle of duality: After calculating values the components were integrated into the original single leg Spice simulation to verify functionality and many reiterations were conducted after making changes to the inductor, capacitor, and resistor values. Simulations were ran until the voltage and current waveforms of the IGBT produced results displaying acceptable power losses along with current and voltage values well within the rating of the IGBT. Results from these tests showing the turn on and turn off characteristics can be seen below: These tests resulted in the following finalized values for the snubber circuit: Alongside snubbers the other means of protection mentioned were the use of MOVs. MOVs were utilized as an extra layer of protection for the IGBTs in order to act as a path for current to flow if an unexpected spike were to occur. MOVs were selected based on the following parameters: The last two design components to be simulated for verification and understanding purposes were the three-phase inverter and the high voltage (HV) side power supply. An ideal circuit had to be implemented for the three-phase inverter in order for Spice to run in a timely manner due to the complexity of timing involved with the fully integrated design of the three-phase inverter. A schematic for the ideal three-phase inverter and its results can be seen below: This simulation served as a means of understanding and having a visual representation of the end goal of this project and reference for tests in the lab. Concluding the power electronics verification stage was a simulation to validate component selection for the HV power supply which consisted of: A schematic and results for the Spice simulation regarding the HV power supply can be seen below: This simulation utilizes a voltage source to represent the voltage seen from the output of the 120:12V transformer and a current source depicting the total load current being drawn by the devices that it will be supplying power to.
 * implementing a three-phase gate driver with a six pack IGBT module
 * implementing a three-phase gate driver with three two pack IGBT modules
 * 50 Ohm resistors
 * 1.2uH inductors<>
 * .12uF capacitors
 * Varistor voltage
 * Continuous rms voltage
 * Peak pulse current
 * 120:12V transformer
 * Full wave rectifier and its smoothing capacitors
 * Two DC-DC converters

= Hardware Procurement = Various new pieces of hardware were required for satisfying the requirements of the project and to replace out of date or broken components from previous projects. This section details the hardware we picked, gives some of the critical specifications and the justification for choosing the specific module. Information detailing the smaller components utilized for the snubbers, full wave rectifier, noise filtration, and all other components has been left out to avoid clutter. If interested in viewing information regarding these components the datasheets for all of the components can be found in our portfolio.

IGBTs
The final choice for IGBT was the FF200R06KE3 by Infineon. The main parameters of the FF200R06KE3 that led to our choice were:
 * its 600V collector-emitter voltage
 * its low collector-emitter saturation voltage of 1.6V
 * its 200A continuous collector current
 * its timing parameters that corresponded to a dead time of approximately 1us

Gate Driver
The final choice for gate drivers were the 6EDL04I06PT by Infineon as the three-phase gate driver and the IR2101(S) by Infineon as the high-side and low-side (single phase) gate driver. Main features of the three-phase gate driver that dictated our choice were: The choice of the IR2101 stemmed from our decision of the three-phase gate driver we chose. Since it would be serving as a research tool to familiarize ourselves with the gate driver and IGBT interactions of our design, the IR2101 was chosen due to its similarities to Infineon's 6EDL04I06PT gate driver.
 * its voltage rating of 620V
 * its capability of separate control for each driver
 * its extra built-in protection and signal interlocking capability of every phase.

Notable Features
Unfortunately, this microcontroller only contains 5 ADCs that can be run simultaneously. This board can also be used with Simulink to run and debug Simulink models and convert them directly C/C++ code that can run independently on the microcontroller. This requires STM32CubeIDE/STM32CubeMX along with MATLAB + Simulink with MATLAB Coder, Simulink Coder, and Embedded Coder. However, we tried to go this route and found that Embedded Coder plugin is not available for purchase on a personal/student/home license and would have to be purchased through the University to be used. If successful though, this would offer an excellent option for researchers to skip the coding and use Simulink models that can be tested in simulation and Processor-in-Loop (PIL) beforehand to program the microcontroller.
 * 32-bit ARM Cortex-M4 CPU (@170MHz)
 * Includes FPU, Real-time Accelerator, hardware trigonometric function evaluation
 * 512KB flash memory, 96KB SRAM
 * 107 GPIO pins
 * Various types of high resolution timers
 * 6 channel automatic complementary PWM generation with center aligned mode, emergency stop, automatic dead time insertion
 * Free IDE with various examples demonstrating peripherals usage, standard debug features
 * CAN, I2C, UART, SPI communication interfaces

Notable Features
Along with this ADC module, a Samtec breakout adapter was purchased to convert the compact Samtec header on the evaluation module to standard sized pins (2.54 mm pitch) to allow direct connection, testing, and prototyping with the microcontroller without making another custom PCB with the Samtec connector.
 * 8 simultaneous sampling channels at up to 200Ksps/channel
 * True Bipolar inputs support +-5V or +-10V depending on configuration
 * High input impedance to allow direct connection to sensors and devices without buffers
 * Fast parallel interface
 * Evaluation module includes standard breadboard header for connecting to analog inputs as well as SMB connectors
 * Precision Host Interface (PHI) module adapts board to connect to PC computer and run free Texas Instruments software to test the module. Software is available at the product page.

Transmitters & Receivers
The transmitters and receivers chosen were the AFBR-2624Z models by Broadcom Limited. These were chosen since they were able to transmit signals at a rate up to 50MBd and utilized TTL input logic.

Fiber Optic Cables
The fiber optic cables chosen were the IF 1L1M-0-5 model produced by Industrial Fiber Optics and were chosen because they were cheap and compatible with the transmitters and receivers we went with. = Implementation and Testing =

=Final Design=

=Future work= Future work needed to be completed for this project can be broken down into five topics, all of which have been partially completed but have some advancements needing to be implemented. These topics include PCB fabrication, high voltage side stand configuration, low voltage side stand configuration, coding, and system testing. Tasks involved with the future work associated with PCB fabrication are as follows: The PCB board needing to be printed is for the gate driver and its external components utilized in the protection abilities that the gate driver device provides. Steps needing to be completed to finish the high voltage side of the prototype include: Low voltage side stand configurations tasks needing to be completed are: In terms of coding the only task needing to be completed is to write and verify the field oriented control algorithm. Following these tasks the last item involved with future work is to run tests utilizing the wind turbine simulator and the RSC design. It is advised to run a few small scale tests verifying system integration before performing tests with the wind turbine simulator.
 * PCB printing (gerber files provided)
 * RCin port capacitor chosen
 * ITrip port capacitor chosen
 * ITrip port resistor chosen
 * Mounting the PCB and making gate driver connections (mounting hardware already in place)
 * Twist soldering snubber components (snubber components already in place)
 * Drilling extra holes on the heatsink for proper IGBT mounting
 * Mounting the positive rail
 * Connecting power supply outputs to the optics and gate driver(devices already mounted in a layout that provides easy interconnections to be made)
 * Mounting uC and making connections to the optics and sensors
 * Creating the breakout boards that will aid in the connection process of the uC (schematics for these boards are provided)

=Team Members=

=Additional Documentation=

Gantt Chart
Gantt Chart