I’ve been working with a three-phase inverter circuit provided directly by the company, which can be implemented in LTspice. While attempting to calculate switching losses, conduction losses, and the overall efficiency of the motor, I encountered convergence errors in the simulation.
Could you please suggest a reliable approach for calculating these losses? Also, at what simulation instant or time interval should these losses be measured for accurate results?
Lastly, would this same circuit be compatible with Qspice?
@physicboy when talking about loss calculation, you have more experience than I.
@Vikas2 Let clarify your situration first. You have a circuit can run in othe Spice platform but encounter convergence issue when implement in Qspice? OR, the circuit can run in Qspice but perform your math of calculating switching loss, it has convergence issue?
About compatibility, assume not encrypted model, LTspice has its unique device (A-device) which cannot be ran by other SPICE platform.
I am able to run the circuit in LTspice. The circuit is attached in the above post.
I now understand the circuit cannot be implemented in Qspice.
Due to convergence issue not able to run the circuit.
The goal is to calculate power losses.
Could you please suggest a reliable approach for calculating these losses? Also, at what simulation instant or time interval should these losses be measured for accurate results?
I can execute the netlist from Qspice by making three modifications:
Adding .option fastmath=0 to utilize 80-bit math.
Including a 100ns maxstep in the .tran directive.
Substituting all subckt ROHM MOSFETs with Qspice’s built-in MOSFETs, for example, UF3C120040.
The first two adjustments aim to address convergence issues. Regarding the last modification, I am currently investigating as the ROHM model (SCT4045DE_L3) is not functioning correctly in Qspice.
Would it be possible for you to share the QSPICE circuit file?
I was able to reproduce the same waveforms in LTspice, but I believe QSPICE makes the process much easier. It might also simplify the power loss calculations switching and conduction right @physicboy ?
That netlist file is for Qspice to run. However, if you prefer a graphical schematic in Qspice, you can draw the schematic, and it is not difficult. Simply avoid using ROHM MOSFETs until a solution is found. The subcircuit libraries “SS I-V3P.lib” and “SS PWM-MOTOR-3P3PM.lib” both function effectively.
Please refer to this Github page of mine to obtain the MOSFET power losses estimate based on ideal switch and losses lookup table structure (which is the same idea as you can do in PLECS or PSIM).
and you may refer to this article to understand the math a bit better.
