Hi Plecs team,
I am fairly new to plecs and I have developed a three-phase, two-level inverter in PLECS using sinusoidal PWM. For the loss and thermal analysis, I am using the Fuji Electric IGBT module 7MBR50VD120-50, with the IGBT and diode thermally modeled based on the parameters provided in the datasheet.
After performing the steady-state analysis in PLECS, I calculated the conduction and switching losses and compared them with the results obtained from the Fuji Electric online simulator under what I believe are equivalent operating conditions. However, I am observing a significant discrepancy between the two sets of results.
I would appreciate any guidance on what aspects of my PLECS model might need adjustment (for example, device parameters, switching loss modeling, modulation assumptions, thermal setup, or operating conditions) in order to obtain loss values that are closer to those reported by the Fuji Electric online simulator.
attaching the simulation and xml files for reference
Thank you very much for your time and support. Any suggestions or best practices for aligning PLECS simulations with manufacturer-provided simulation tools would be greatly appreciated.
Best regards,
Yashaswini
fuji_typeg.plecs (94.7 KB)
new_fuji.xml (171.8 KB)(igbt xml file)
was not able to add diode xml file earlier adding it now
new_fuji_diode.xml (134.1 KB)
the plecs model losses are:
IGBT
Conduction : 1.74W
Turn on loss : 0.4744W
Turn off loss : 1.7815W
DIODE
Conduction loss : 0.2726W
Reverse recovery loss: 0.1617W
Fuji electric online simulator losses
IGBT
Conduction : 1.54W
Turn on loss : 0.47W
Turn off loss : 1.15W
DIODE
Conduction loss : 0.45W
Reverse recovery loss: 1.07W
the input parameters for both are same
Thanks for sharing the model and the XML files.
I had a look at the setup and I don’t see any major issues in your PLECS model. The device modeling, loss calculation approach, and thermal setup all look reasonable and consistent.
One minor point is that you are constructing the PWM generator manually instead of using the built-in PWM blocks available in PLECS. While this is perfectly valid, it does introduce an additional source of potential differences. In particular, details such as dead-time implementation, carrier type (sawtooth vs. symmetrical), comparator logic, and any additional gating logic can all influence switching behavior and therefore the calculated losses.
It is also important to note that the internal assumptions and implementations used in the Fuji Electric online simulator (e.g., PWM strategy, dead-time handling, current ripple assumptions, switching loss interpolation, or internal derating factors) are not publicly documented. For this reason, it is not possible to make further detailed comments on the simulator’s results or to guarantee a close numerical match.
In practice, some discrepancy between PLECS and manufacturer-provided simulation tools is expected unless all modulation, operating conditions, and loss modeling assumptions are matched very closely. If tighter agreement is required, I would recommend carefully aligning PWM details (especially dead-time and carrier definition) and double-checking that the electrical operating points (DC-link voltage, modulation index, current amplitude and phase, switching frequency, and temperature) are truly identical in both tools.
Hope this helps.
Best regards,
Reto
Thank you for taking the time to review and greatly appreciate the confirmation.
Based on your suggestion, I will rework the model using the built-in PWM blocks in PLECS and align the carrier definition and dead-time handling as closely as possible, to check whether this reduces the discrepancy in the calculated losses.
I also wanted to ask if there are any recommended best practices or checks within PLECS to ensure that the loss estimation error stays within a reasonable bound, say ±10%, when comparing against manufacturer-provided simulation tools. For example, are there specific parameters, custom tables, assumptions, or validation steps that you would typically focus on to improve confidence in the loss results?
Best regards,
Yashaswini
Dear @yashaswini
To clarify, we do not have insight into the internal implementation or assumptions of the Fuji Electric online simulator. Since its PWM strategy, loss modeling, and internal settings are not documented and not accessible to us, we cannot comment on or explain its results.
Best regards,
Reto
Thank you very much for your help