When experimenting with the FIKAT board, I noticed that the inductances are tiny. For a DC-link voltage of 48V, 50% duty cycle and 24V output voltage at 150kHz, the current ripple is 61.5A. In reality, the duty cycle won't always be 50% and the output voltage at the filter capacitors isn't constant, so the average current ripple will be less on average, but just a bit, not by orders of magnitude.
This causes the semiconductors to heat up even at small loads. Running an induction machine at about 0.5A peak heated the chips to about 100 °C (thermal camera). Removing the filter capacitors alleviated the problem for this setup. The machine's inductance is quite large, so the current was sufficiently filtered, unfortunately the steep voltage slopes on the wire to the machine caused the connection to the debugger to break as soon as PWM was turned on. But the chips barely exceeded 40°C
For small loads, the 1.3uH inductors can be replaced by 33uH inductors in the same package. This reduces the current ripple to 2.4A, which is still fairly large, but acceptable. Unfortunately these inductors have a rated current of 5.5A, which is too small for larger loads. Alternatively, inductors can be added externally (given that the capacitors are removed), but that will cause the voltage at the sensors to jump as this creates an inductive voltage divider between the output voltage and the switched voltage.
When experimenting with the FIKAT board, I noticed that the inductances are tiny. For a DC-link voltage of 48V, 50% duty cycle and 24V output voltage at 150kHz, the current ripple is 61.5A. In reality, the duty cycle won't always be 50% and the output voltage at the filter capacitors isn't constant, so the average current ripple will be less on average, but just a bit, not by orders of magnitude.
This causes the semiconductors to heat up even at small loads. Running an induction machine at about 0.5A peak heated the chips to about 100 °C (thermal camera). Removing the filter capacitors alleviated the problem for this setup. The machine's inductance is quite large, so the current was sufficiently filtered, unfortunately the steep voltage slopes on the wire to the machine caused the connection to the debugger to break as soon as PWM was turned on. But the chips barely exceeded 40°C
For small loads, the 1.3uH inductors can be replaced by 33uH inductors in the same package. This reduces the current ripple to 2.4A, which is still fairly large, but acceptable. Unfortunately these inductors have a rated current of 5.5A, which is too small for larger loads. Alternatively, inductors can be added externally (given that the capacitors are removed), but that will cause the voltage at the sensors to jump as this creates an inductive voltage divider between the output voltage and the switched voltage.