I am designing a half-bridge plus H-bridge inverter. I need to calculate the loss of the switch tube (turn-on, turn-off and conduction loss). My control is variable modulation, and I see that the tube loss is always changing. How should I calculate it? I don’t know if it is right or wrong. For the module that comes with plecs, the unit of loss is W. If I want to calculate it as energy, how should I do it?
The first line is switching loss, the second line is conduction loss. I want to know why the loss is periodic. How to calculate a cycle or a switching cycle?
The Switch Loss Calculator block will output the losses in Watts averaged over a specific interval that you provide. If you use a standard Probe block, you can access the switching energy signals directly, as you show in the image of your third post above. But then to calculate these over a defined (switching) cycle, you will need to use Averaging blocks. This is already done for you with the Switch Loss Calculator block, and we do recommend you use that!
Do you mean that I can calculate the loss using the switching loss module, and then I see that I need to set a time. Is this time the switching cycle of the switch tube?
Often the averaging period relates to the switching frequency, yes. But for AC systems, you might average over the AC line cycle frequency. Based on your description of your system, I think the AC frequency is most appropriate here.
I have encountered a new problem, that is, my model has a loss of about 310w at 40k, and a loss of about 330w at 50k, but the difference is 10w at 10k, 20k, and 30k. I discussed with others because there is a problem with my model, and I don’t know how to solve it. Can you help me take a look?
CHB_IGBT_up.plecs (497.4 KB)
IKQ75N120CS6_diode.xml (5.6 KB)
IKQ75N120CS6_igbt.xml (7.0 KB)
I used the other number:IKQ40N120CH3.
The result is similar in 50khz, i want to know whetaher it is wrong.
I do see there is non-linear behavior in the losses for the IGBT at 50kHz, although at 49 and 51kHz this is not the case. So there is some timing issue in your model most likely. I’m curious why you implemented your controls with the C-Script blocks in a more complicated than necessary way? I’d suggest you try to use the included pre-built modulator blocks. But actually, the main thing I noticed is that you are using the fixed-step solver and I think this could be the main factor. Have you tried the variable-step option? This is certainly what we recommend first and foremost. Please revisit the results with that setup and come back if you continue to have issues!
I use C-script to modulate to reduce losses.
I also use variable step length, but the result is not right, and the waveform has a big problem.
I would like to ask whether the general IGBT switching frequency is before 50Khz, or what range.
I have a problem now. I use GaN to replace this IGBT, and the output waveform is not five-level. Why is this?
This has been answered in your other post.