ISSUE 012 Winter 2021 Sigma Powertrain EMAX transmission dossier l In conversation: David Hudson l 48 V systems focus l 2021 Battery Show North America and Cenex-LCV reports l Everrati Porsche 911 digest l Switching insight l Motor laminations focus

The unit has been tested on a two- level, 400 V inverter from BorgWarner/ Cascadia Motion (the PM100DX) with a control board including an OLEA T222 FPCU from Silicon Mobility and a Semikron IGBT module. The motor is an HVH250-090SOM product from BorgWarner. The tests show an increase in efficiency of the IGBT-based switching system from 82 to 87% at low load and from 91 to 93% at higher loads. Pre-switching Another approach is pre-switching, where the timing of the switching is varied according to changes in the inductance and capacitance of the inverter. These vary with the load, so the timing of the switching is constantly changing. “If you look at a Tesla, the drivetrain is the biggest loss in the system until 55 mph, then wind resistance becomes the biggest loss,” says Bruce Renouard, CEO at Pre-Switch. “This means you have a different efficiency at 5-6% load and at 80% load, but the efficiency of pre-switching is flat across the load.” This pre-switching is achieved by a combination of a new switching topology and a controller called CleanWave, which uses machine learning to identify the patterns in the voltage and current across the combination of the inverter and the motor, and across the whole load range. The controller pre-distorts the sine wave that controls the switching of the transistors so that a perfect sine wave is recreated in the motor. “We add two IGBT switches to a half-bridge to handle the current for a pulse rating with zero switching losses so that they can switch faster,” Renouard says. “That means they don’t need to be big, so you can reduce the number of switches, from four or six units in parallel to each other. You also shrink the DC-link capacitor dramatically. “We gave one design a 0.5% gain in peak efficiency but at the 5% load it’s a big difference, it gives 5-12% more range. “We are giving people a full inverter with motor control that we will license, but we have also developed a 200 kW inverter that is 32 mm tall, 150 mm (6 in) wide and 200 mm (8 in) long including all the cooling and DC-link capacitor,” he says. This smaller size comes mainly from reducing the size of the DC-link capacitors, as they have to handle less current. “The CleanWave2 design uses the cheapest IGBTs and cheapest discrete SiC cascode devices,” he adds. These cascode devices, from UnitedSiC, combine a low-voltage silicon MOSFET and high-voltage SiC JFET (junction FET) in the same package. Their advantage comes from the fact that they can be controlled using ordinary MOSFET gate-drive signals generated by an ordinary MOSFET gate driver. “You can’t usually use these in inverters, as they are not gate- controlled, so you can only use them in a soft switching topology. But with our topology there is no damping, as switching occurs with zero voltage and zero current. This design is for frequencies from 50 to 150 kHz with no external DC-link capacitors. “By doing all the motor control we were able to improve the efficiency of the whole system. Working with other people’s inverter designs, we had to react to the PWM input and map the performance over a period of time to learn the system. “Now we know what the system needs and how it works. For example, this means at peak load we can slow the switching down to 10 kHz, where the edge is transitioning the fastest across the zero voltage, and that’s how we reduce the losses.” Constructing a power module for inverter switching (Courtesy of onsemi) Winter 2021 | E-Mobility Engineering 61 Deep insight | Switching