A Physical Approach to High-Bandwidth, Low Insertion Impedance Current Measurement for High-Frequency Switching Wide-Bandgap Devices

Since its invention by Julius Edgar Lilienfeld in 1926, the field-effect transistor FET led the way to the invention of the metal-oxide-semiconductor field-effect transistor (MOSFET) in the 1960s. Contrary to their poor performance in the beginning, MOSFETs soon improved their switching characteristics and switching is what they are mostly known and used for in power electronics today. Until the introduction of MOSFETs based on silicon carbide (SiC) in 2011 and high electron mobility transistors HEMTs based on gallium nitride (GaN) in 2010, Si and other semiconductors with band gaps <2eV were the only options for power electronics designs. Since then, transistors based on WBG semiconductors became the focus for new developments in the field. Compared to traditional Si, GaN and SiC transistors offer improved electrical features due to their wide bandgaps of 3.4eV and 3.26eV. The most prominent improvements are the higher blocking voltage, faster switching speed, and superior thermal characteristics. The improved switching behaviour of GaN and SiC transistors is the key ingredient for more efficient power converters and thus helps to save substantial amounts of energy in power conversion on big and small scales. This thesis focuses on the improved switching performance of GaN transistors, which is mainly a result of a) the high electron mobility of the GaN substrate, hence the very low R_DS−ON and b) the reduced parasitic capacitance and inductance. The latter is a result of the compact package and improved connection between package and chip. GaN devices allow switching within 1ns. To achieve such fast slew rates, the rest of the electrical circuitry must be dimensioned accordingly. The copper traces of the commutation and driver loop must feature lowest possible impedances. When developing power converters, it is essential to know the voltage and current across the switch. Their product is the power loss of the switching transistor, which has the most impact on the correct function and efficiency of the power converter. Additionally, voltage and current indicate if the switching process is precise, without transient oscillations or even ringing after switching. The evaluation is done by measuring the voltage v_DS and current i_DS across drain and source of the transistor. Voltage sensing is straight forward and can be done with some custom made test points at the PCB. The current measurement requires a sensor to be fitted into the commutation path, which inevitably adds insertion impedance for the measured current. The amount of inserted impedance highly depends on the sensor type and implementation design. Conventional sensors and shunts usually come with high parasitics and a bandwidth which is too low to capture the fast transient. Furthermore, galvanically isolated current sensing would be preferable. To evaluate current sensing for WBG semiconductors, this thesis takes a look at suitable physical phenomena and current sensors based on these effects. Custom made PCBs for the so-called double pulse test with two GaN transistors in a half bridge configuration were developed and tested in three different variations. These variations feature two different current sensors, firstly the established coaxial shunt (CSR) and secondly the novel Infinity Sensor, a specifically optimized Rogowski coil, developed and sold by the University of Bristol. The third variant features both sensors in combination, enabling measurement verification. All three variants were used for double pulse tests, a common evaluation procedure for transistor switching. Measurements were taken, post-processed, and assessed for the technical feasibility of the sensors as well as their limitations for the intended purpose. Further technologies for current sensing were considered theoretically and partially adapted to the existing PCBs, but not tested yet. These are described in the outlook and future work. A miniaturized version of the coaxial shunt was proposed, which is called ultra fast current shunt (UFCS). Another promising technology is the fibre optical current sensor (FOCS) which uses the Faraday effect, and combines galvanic isolation and very low parasitic impedance. Until now, this technology is only used for high voltage applications with very high currents. Again, miniaturization could lead the way for this technology to the application in power electronics as mFOCS.