LTSPICE simulation on switching and conduction loss on three phase inverter

Discussion in 'The Projects Forum' started by yue.yao, Feb 22, 2013.

  1. yue.yao

    Thread Starter New Member

    Dec 5, 2012
    The purpose of this project is to provide an accurate simulation of the conduction and switching losses inside a three phase inverter under different driving schemes and ultimately quantify how different parameters (e.g, dc voltage, drain current, switching frequency, rise and fall time, blanking time, load-side resistor and inductor) affect each component of power loss.

    Reffering to the model in the attachment, the three phase inverter is composed of 6 IRF7832 MOSFETs and 6 anti-parallel diodes in hex-bridge configuration. The load (brushless DC motor) is modled by resistances and inductances joined in star configuration.

    The pulse width modulation waveforms for driving the three half-bridges is obtained via comapring 3 sets of shifted sine waves (shifted by 120 degrees from one another) whose fundamental frequency is 60Hz with a repeting triangular wave whose switching frequency is 10KHz. This is implemented in LTSPICE using behaviour voltage sources.

    The blanking time is necessary so that the 2 MOSFETs within a half-bridge do not simultaneously turn on. This is implemented in LTSPICE using resistors, capacitors and AND gates.

    With the DC link voltage set to 12V and a switching frequency of 10KHz, averaged gate loss + conduction loss + switching loss in MOSFET 1 (top left one) is 1.485W and 1.524W in MOSFET 2 (button left one). In the antiparallel diode, the sum of averaged conduction and reverse recovery losses is 123.5mW in diode 1 (top left schotty diode) and 114.4mW in dopde 2 (button left schotty diode).

    From theory, the conduction loss in the MOSFET is derived by multiplying the square of RMS drain to source current with the on-state resistance. Given drain current is at maximum 20A (according to datasheet) and Rds(on)=4 milliohm, the total conduction loss in the inverter is therefore 4.8W.

    I'm hoping to separate the loss components in LTSPICE into MOSFET conduction, MOSFET switching, diode conduction and diode reverse recovery loss, but so far no luck. Also, I'm uncertain whether the losses in the free wheeling diodes are correct, since each MOSFET should have an internal body diode which may or may not be accounted for in LTSPICE. So for example, when MOSFET 1 is in the off-state, the current can pass both the internal body diode and the anti-parallel diode. I don't know if putting a diode in series with the MOSFET will block the current through the MOSFET and thus force the current into the free-wheeling diode.

    Please leave comments and suggestions on how to improve it!