0

I want to compare the Safe Operating Areas of two MOSFETs based on the graphs given by the manufacturers in the datasheets.

Our application is a 24 V inverter. What I concluded based on the information below is that device A can handle a higher current for DC operation, 10 ms operation and 1 ms operation. The values of current at 24 V for these three operating points are 10 A, 20 A and 100 A as compared to 0.22 A, 1 A and 17 A for device B.

However, for the 100 μs operating point device B has a higher current carrying capability as compared to device A and the value is 287 A as compared to 260 A for device A for 24 V.

Is my analysis correct, and what else you can identify from this information when comparing two MOSFETs?

Secondly, since I am switching at 20 kHz for my application, I want to see the behaviour at the operating point of 50 μs which is not given in this graph. Can anyone please help me in identifying the difference at the 50 μs operating point?

enter image description here

enter image description here

ocrdu
  • 8,705
  • 21
  • 30
  • 42
Alison
  • 295
  • 5
  • You say "switching", do you know what V and I are throughout the waveform? – Tim Williams Jan 31 '23 at 09:19
  • I'd just look at the 100 us lines, which would be worse than the 25 us lines that you really need for a 1/2 cycle of 20 kHz. If 287 A plays 260 A, those are exactly the same in engineering terms, certainly not sufficient difference to base your choice of device on. Now, don't run your device anywhere near these lines for reliability, and note that these diagrams only apply for a case temp of 25 C, which it will never be in practice! – Neil_UK Jan 31 '23 at 09:21
  • @TimWilliams in my case Voltage is 24V and Current is 40A max at 20khz – Alison Jan 31 '23 at 09:43
  • @Neil_UK I completely agree with you that these observations based on datasheet is far away from reality but still the nature of my activity is to do datasheet comparison for now. what would be your conclusion just looking at these SOA graphs. thanks – Alison Jan 31 '23 at 09:48
  • I ask, because if your "switch" has 40A and 24V applied to it simultaneously, it's not switching. Switching goes rapidly from the top-left region to the bottom-right edge of the SOA curve. The SOA curve is generally not of concern for SMPS in normal operation. – Tim Williams Jan 31 '23 at 09:52
  • @TimWilliams Could you please elaborate why SOA curve is generally not of concern for SMPS in normal operation. thanks – Alison Jan 31 '23 at 09:54
  • Plot the voltage (Vds) and current (Id) versus time for the MOSFET. Plot these on the SOA graph, along with approximately how much time is spent in each region. – Tim Williams Jan 31 '23 at 09:58
  • @Alison my conclusion is that there is no practical difference between these devices based on SOA and your application. Note Tim's comments about knowing what current and voltage waveform to expect **at the device terminals**. The SOA is about limiting the junction heating, so limiting the time you have high current and high voltage simulataneously. With a half cycle time of 25 us, you're unlikely to have high current and voltage for more than a few us, otherwise your efficiency is going to be rubbish. Do a switching simulation and see what you get. – Neil_UK Jan 31 '23 at 10:20
  • Just to add that for switching service the FOM most people care about is Cg, as that determines the gate charge losses, and Rds(on) as that determines the conduction losses, and possibly also the behaviour of the body diode which can matter in some designs. SOA matters when running a mosfet in linear operation where there is simultaneously significant voltage across the mosfet and significant current, most switching tries very hard to be an either one or the other sort of situation which makes little heat. – Dan Mills Jan 31 '23 at 11:09
  • @DanMills Thanks for your comment. is linear operation of Mosfet is same as DC operation – Alison Feb 07 '23 at 10:11
  • @Alison Not really, linear operation is where the mosfet is dropping a significant amount of voltage while passing a significant current, such as you might find in an audio amplifier, or a current source, the product of Vds and Id is a significant amount of power dissipated. In a switching design you try to stay OUT of the linear region, with the the thing spending most of its time either conducting with minimal Rds (Hence not dropping a significant voltage), or not conducting (hence dropping voltage but passing nearly no current), neither condition resulting in much power dissipation. – Dan Mills Feb 07 '23 at 12:09

0 Answers0