How to model an N-channel IGBT in Simscape using output characteristics available in datasheets accurately?

I'm doing loss calculation of an infinion IGBT, and i want to parameterize the N-channel IGBT block using look up table approach (giving known Vge, Vce, and Ic values to model it.)

My gate voltage driver switches b/w +15 & -15 at 444 Hz, and my Vcc is 810 V.

But in datasheet output chara corresponding to Vge of 8,9,10,12,15,and 20 V are only given.

Also Vce changes from 810 V to 1.9V during switching but in datasheet only values from 0 to 5V are given.

How can I find Ic corresponding to other values of Vge( -15 to 8V ) and Vce (5 to 810 V) so that my model accuracy increases?

Answer 1

[Joel Van Sickel]

Hello Nikhil,

the data being asked for is only for steady state operation for when the device is switched on. You do not need to capture VCE values higher than 5, because conditions when VCE is greather than 5 are durring turn on and turn off events, as well as when the device is not conducting, in which case only the dynamics are relevant.

The particular points of interest you need, to make sure the device is simulating correclty when turned off, would be the approx 0.6V VCE location, which is where Ic first starts to rise (aka, the device starts to conduct), and the point at 0V for VCE. You should also add an extra point for VGE at 0, and say all the currents at this point are 0 (or if you are being very accurate, the corresponding leakage current)

Regards,

Joel

[nikhil james]

Hello Joel, thankyou for your reply.

So, the dynamics is not affected by the look up table values and is only determined by the junction capacitance values that we enter, right?

Also which approach gives the most accurate model, 'fundamental non-linear equations' or 'look up table' ?

[Joel Van Sickel]

Hello Nikhil,

you are correct, the dynamics will be dictacted by the junction capacitances.

There are two ways to look at it. If you are concerned about having a device that matches a data sheet specifically for the graph provided (IV curve), the lookup table approach is the most accurate. However, as most devices are operated at a single VGE, it is not necessarily that important to match any curves other than the VGE that you are going to use, and even that will have a limited portion of the curve for standard operation. In this case, the best performance can actually come from parameterizing the non-linear equations and tweaking them to match well at your specific VGE. The IV curves might not line up as well, but you'll get better dynamic performance for your specific operating conditions.

Regards,

Joel