An animated sequence of 20 1µs screenshots illustrates how the Pre-Switch AI actively learns from the initial startup under unknown conditions, then optimizes and adjusts the timing required to ensure a PWM input generates a current ramp to simulate the first part of a sine wave output.
Dual-pulse test data obtained from the 200kW inverter demonstrate that the Pre-Switch soft-switching platform, comprising the Pre-Drive™ 3 controller board powered by the Pre-Flex™ FPGA and the RPG gate driver board, reduces system switching losses by 90% or more.
At the start of the initial switching cycle (0), the Pre-Switch AI controller evaluates multiple inputs to determine the system's mode and then makes a safe, but not optimal, estimate of the resonant period required for soft switching. During the subsequent switching cycle (1), all AI inputs and the resulting outputs from switching cycle 0 are remeasured and accurately analyzed. A second conservative resonant synchronization period, similar to switching cycle 0, is emitted.
In switching cycle (2), the AI algorithm is now quantitatively reliable and can predict the optimized resonance time to ensure complete smooth switching, thus minimizing losses across all aspects of the system. Then, in switching cycle 3, the AI compares the system inputs and the results of previous switching cycles and adjusts the resonance time to fully optimize smooth switching with increasing load current. These inputs and outputs are stored along with the inputs from the previous switching cycle and the resulting resonance times to improve system accuracy and optimization.
In subsequent cycles, the smooth switching accuracy continues to be optimized, stored, and compared to the desired results needed to maintain accurate forced resonance smooth switching. Changes in system temperature, input voltage, output load current, and device degradation are taken into account and optimized within the pre-switch AI algorithm.
According to Bruce T. Renouard, CEO of Pre-Switch: “Pre-Switch is enabling customers to build systems with switching frequencies 4-5 times faster than their fixed-switch IGBT systems and 35 times faster than their fixed-switch SiC and GaN systems. The GIF demonstrates how our AI is working within the CleanWave200 to bidirectionally convert 800VDC to three-phase AC at power levels up to 200kVA with a switching frequency of 100kHz at 99% efficiency.”
The CleanWave200 evaluation system, reference design, and design files can be ordered from Pre-Switch.
