Even with the gate driver UCC5350-Q1, S or M version bipolar supply cannot be used. The datasheet only talks about unipolar operation and no mention of bipolar supply. Even the typical application circuits show unipolar supply with VCC2 connected to 15V with respect to GND connected to VEE2. M version consists of CLAMP pin to avoid false turn ON.
The expected power loss should be around 4 watts.
Another thing is on top you mentioned that Vds voltage is 115V but the MOSFET is rated for only 60V maximum.
We have a setup with 7 switches connected in parallel per arm, and there are 6 such arms in total. A prototype has been built and tested. During testing, a current of 375A was applied for 30 seconds, resulting in a measured temperature of 92°C. This corresponds to approximately 53.57A per switch.
However, we’re facing difficulty validating the loss values from the simulation. The simulation indicates a loss of 28W per switch.
One method I considered was to measure the voltage drop across a single switch and multiply it by the current through that switch to estimate the power loss. However, since the voltage drop across each switch is quite low, I’m unsure whether this method is practically viable or accurate.
Could you please suggest any reliable methods to experimentally validate the loss values per switch?
It is basically two different unipolar supplies, it is okay as per the circuit shown. The one connected on the bottom side will connect to the gate in reverse direction hence acting like a negative voltage across gate to source. It looks okay to me but finally will have to look at your circuit to confirm.
If it is showing 0.5V then it is quite high, if 0.5V is dropping across drain and source terminals then it means ON resistance of 6.25 mili ohms which corresponds to 40W of power loss but datasheet says max ON resistance of 1 mili ohms only.
Something is wrong in your simulation, I would have to look at your circuit and the MOSFET model characteristics.
Can you probe the voltage across drain and source terminals of the MOSFET using a voltmeter on the actual board while the MOSFET is conducting current. Let me also know the value of the current being conducted and the voltage across gate and source terminals of the MOSFET.
I have shared the mosfet data sheet which used for the prototype MOSFET Datasheet
The resistance shows are 6.7mohm.
Is practically possible to testing potential drop using voltmeter across mosfet during high current passing?
I was referring to the previous MOSFET datasheet. 40 watts power loss with 6.7mili ohms and 80A is possible. You will need to somehow measure the actual voltages to verify if the MOSFET is working as expected.
Just FYI, I just posted this on our LinkedIn page and tagged Juliano Mologni of Ansys. Here’s what he says:
There are many things in here that I don’t think you are considering for an accurate electrothermal simulation:
1-You are considering a “uniform” loss computed in PSIM. You have a loss map distribution (just like temperature) that is not uniform
2-I can’t tell for sure, but not sure if you are including the PCB design in your thermal simulation (Icepak supports all industry formats such as ODB++, IPC2581)
3-There are DC and AC losses. You are using only a DC RMS current, which is not accurate. You should consider harmonic losses (real current waveform)
4-When temperature rises, it changes the resistivity/conductivity of the materials. That’s the reason you need a two-way iterative flow (losses > Temperature, and update the losses and recompute the temperature until you reach convergence)
5-Ansys Q3D and Ansys Icepak provides all of the above (compute losses map, two-way coupling, harmonic losses with real world current waveform). We are also shipping some Infineon switching devices now with the install to get accurate current waveform.
I would suggest you reach out to the Ansys Engineering support team, they can definitely help you understand all the details I mentioned above so you can get an accurate results.
