A SPICE Model for IGBTs and Power MOSFETs Focusing on EMI/EMC in High-Voltage Systems
Round 1
Reviewer 1 Report
Review Report on electronics-1444043
The MS entitled “SPICE Model for IGBTs and Power MOSFETs Focusing on EMI/EMC in High-Voltage Systems” -by B. Khvitia, A. Gheonjian, Z. Kutchadze, and R. Jobava presents two models of Power Transistors the IGBT and MOSFET that in the paper were used for analysis of electromagnetic interference (EMI) and electromagnetic compatibility (EMC). Both models were behavioural and model parameters were obtained from datasheets and adjusted using results from measurements. The results obtained in simulations were compared with measured turn-on and turn-off waveforms for different types of IGBTs and the models validated comparing with a brushless DC electric motor test setup.
The paper is well organized, and its English is satisfactory enough.
Results in the paper are sound, validated comparing with measurements and, certainly, they are of interest for researchers working in the field.
My two main questions are the following:
- A behavioural model is based on equations and cannot be better than these equations. Both power MOSFET and IGBTs are extensively studied in literature and a certain number of non-behavioural models based on exists. Please, could authors discuss how the new models compare to the existing ones?
- Was temperature taken into account in the model? In a power device, modelling temperature is of paramount importance. If not considered, authors must discuss how to temperature effects could be implemented in a future version of the model.
Author Response
Dear reviewers!
I want to thank you for attention, interesting questions and considerations.
I hope changes that we made in our article will fulfill your considerations and questions.
-A behavioral model is based on equations and cannot be better than these equations. Both power MOSFET and IGBTs are extensively studied in literature and a certain number of non-behavioral models based on exists. Please, could authors discuss how the new models compare to the existing ones?
We agree that non-behavioral models exist. In the introduction of our article we discusses papers that describe these issues in detail. For example, "A Review of IGBT Models" reviews transistor models in the literature before year 2000, analyzing, comparing, and classifying them by type of mathematics, objectives, complexity, accuracy, and speed.
Our goal was to create as simple model as possible illustrating only the basic parameters of transistors but capable of modeling EMC-EMI problems.
- Was temperature taken into account in the model? In a power device, modelling temperature is of paramount importance. If not considered, authors must discuss how to temperature effects could be implemented in a future version of the model.
Since the use of our models is recommended to detect parasitic resonances in cable systems, the temperature dependence in these models has not been taken into account. However, in the first model which is more accurate, the temperature dependence can be checked by entering its temperature in the parameters of the PNP transistor. Temperature is not provided in the second transistor model.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The work presents a interesting topic, that of refined Spice model of IGBTs, to reliably predict electrical phenomena related to EMC. The discussion is supported by experimental results taken from two different IGBTs.
The quality of the work is i my opinion good and I have a couple of considerations.
1) General, regarding results. it would be beneficial spend more words commenting the results, both when comparing simulation models and also when including experimental data.
2) Line 247, measurement setup. Measurement of waveforms. Could you clarify if you used standard oscilloscope probes, and how the ground reference was selected, including the problem of using multiple probes at the same time and the possible influence of probe cable routing.
3) Line 312-313 and elsewhere. Please, quantify the accuracy of the model, such as maximum error from measurement results, but also specifically focusing on portions of the waveforms ( such as rise time and overshoot) that are the most relevant once interpreted in frequency domain for EMC purpose.
Author Response
Dear reviewer!
I want to thank you for attention, interesting questions and considerations.
I hope changes that we made in our article will fulfill your considerations and questions.
1) General, regarding results. it would be beneficial spend more words commenting the results, both when comparing simulation models and also when including experimental data.
Your consideration was taken into account in revised article. Please look line 354-364.
2) Line 247, measurement setup. Measurement of waveforms. Could you clarify if you used standard oscilloscope probes, and how the ground reference was selected, including the problem of using multiple probes at the same time and the possible influence of probe cable routing.
During 100 amp pulsed currents, voltage drops (up to 200-300mv) occur in the GND of the PCB, which distorts the measured signals. Also a problem is the influence of a direct magnetic field, the influence of which is quite difficult to neutralize in high-altitude standard probes. The measuring PCB design is made in such a way that the diameter of the power loops is minimal, the capacitors must have a minimum internal resistance and inductance, the resistors must also have a minimum of parasitic inductance, and so on. All probes used for the measurements are located on the PCB. Their design was designed so that they took up as little space as possible on the PCB, had minimal impact on the measurement parameter, and their output impedance was 50 ohms. By doing this we have minimized the additional interference caused by the use of standard probes.
3) Line 312-313 and elsewhere. Please, quantify the accuracy of the model, such as maximum error from measurement results, but also specifically focusing on portions of the waveforms ( such as rise time and overshoot) that are the most relevant once interpreted in frequency domain for EMC purpose.
We took note of your comment and added more detailed images of the Fall and Rise signals. For optimized We took note of your comment and added more detailed images of the measurement and modeling waveforms Fall and Rise shapes (Fig.9, Fig.12 and Fig.10, Fig.13). For optimized models, in our estimation, the error of the complex model is ≈5-10%, of the second simple model ≈10-20%.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have satisfactorily responded to all my questions and made the necessary changes to the manuscript.