Hello, I am exploring the thermal simulation calculation principle of MOSFETs in PLECS. I have set up turn-on & turn-off losses, conduction losses (gate on & off), and used a Cauer model in the Thermal Chain with defined thermal resistances and capacitances. I also created an Rgon table and an Rgoff table in Custom Tables. In the turn-on & turn-off loss settings, I selected “Lookup table and formula” with the formula:
E*(i>0)*lookup('Rgoff table', Rgoff, T)/lookup('Rgoff table', 5.1, T).
I would like to know: when calculating losses in PLECS, is it simply multi-dimensional linear interpolation? That is, when the simulation conditions are not exactly the values defined in the .xml file, does it draw a line between two adjacent points and take the interpolated value?
Also, if the selected Rg exceeds the range defined in the Rgon table and Rgoff table, what is the extrapolation strategy?
When calculating losses in PLECS, is it simply multi-dimensional linear interpolation?
That’s correct, as noted in the documentation.
Also, if the selected Rg exceeds the range defined in the Rgon table and Rgoff table, what is the extrapolation strategy?
The same extrapolation strategy is used. Two notes:
- In the Variables tab of the thermal editor you can define Min/Max values for input variables to alert the user if the entered Rg is outside of the allowed range.
- Speaking more generally, one can also construct the look-up table so the extrapolated value would be the first/last value by adding a repeated data point at the end of series. In the case of Rg scaling this wouldn’t correspond to the device behavior, but it can be very useful in other cases, especially when using look-up tables in control applications.
OK.Thanks for your reply.
“I have another related question and hope to get your reply.
During PLECS simulation, are all the lookup tables in the Thermal Description using linear interpolation? For example, Cond.Loss (Gate On & Gate Off) and the Rgon Table & Rgoff Table defined in Custom Tables – do they also use linear interpolation, or other interpolation methods?”
Yes, linear interpolation is used in this case as well. If the loss characteristic is non-linear, as it often is, then it can be broken down into smaller linear sub-regions.