0
\$\begingroup\$

I would like to know which parameters make the difference between a high frequency IGBT and a low frequency IGBT.

For example, I read the AIKW50N65DF5(15 - 120kHz) datasheet and the AIKW50N60CT(0 - 30kHz) datasheet (both from Infineon) and I find that all the parameteres are similar.

So, what really makes the difference between them? Why an IGBT can work with a frequency that other can not?

\$\endgroup\$
  • \$\begingroup\$ I realized the rise/fall times are different but also the test conditions are. AIKW50N65DF5 is done with Rg = 12Ohm and Lsigma = 30nH and AIKW50N60CT is done with Rg = 7Ohms and Lsigma = 103nH. Thank you all! \$\endgroup\$ – ErnstOlch Feb 20 '18 at 9:40
2
\$\begingroup\$

It's about the rise time and fall time.

  • Rise time = time it takes for a signal to go from 10% to 90% of some value
  • Fall time = time it takes for a signal to go from 90% to 10% of some value

In the context of transistors, it's the drain/collector node. The schematic setup is just a pull-up resistor to the drain/collector if I'm not mistaken. You properly increase/decrease the voltage at the gate to make the transistor start conducting/stop conducting.


Here's the rise time and fall time for the AIKW50N65DF5:

  • Rise time = 12 ns
  • Fall time = 6 ns

Here's the rise time and fall time for the AIKW50N60CT:

  • Rise time = 29 ns
  • Fall time = 29 ns

So by the time the AIKW50N60CT has had its collector go from 0 V to VDD, the collector of AIKW50N65DF5 has been able to go from 0 V to VDD, stay there for 11 ns, and then go back to 0 V.

You won't change the voltage of the gate instantly though, you need to charge the parasitic gate capacitance above the threshold voltage, and in the real world there's only so much current you can force through the gate without damaging it. So if you got a driver IC that can only force 1A into the gate, then with less gate capacitance, you will reach your threshold voltage faster and therefore be able to reach higher frequencies.


Here's the capacitances for the AIKW50N65DF5:

  • Input capacitance = 2800 pf
  • Output capacitance = 65 pf
  • Reverse transfer capacitance = 11 pf

Here's the capacitances for the AIKW50N60CT:

  • Input capacitance = 3140 pf
  • Output capacitance = 200 pf
  • Reverse transfer capacitance = 93 pf

Then there's the gate charge that changes with gate-emitter voltage, in both of the transistors' typical usages, the gate charge for the fast transistor is about 1/3 of the slow one. So in a typical setup, the fast one should reach its threshold voltage about 3 times faster than the slow one.

There're probably many other things that I haven't thought of, but these things alone will be the main focus. I mean the fast transistor's max frequency is 120 kHz, the slow one is 30 kHz. The fast one can be used for a frequency that is 6 times greater than the slow one. The 1/3rd gate charge together with the less-than-half rise time + fall time => 3 × 2 = 6. And the capacitances will just make the fast one about 10% faster, which is probably lost somewhere on the way. So it makes sense.

\$\endgroup\$
3
\$\begingroup\$

I find that all the parameters are similar.

Not all of them: Turn-on delay time 21 / 27 ns

There are other differences as well. The differences are small though.

So one is slightly faster than the other.

My guess is that these devices are identical but that they are binned meaning, after production they are measured.

Then the "fast" ones ( Turn-on delay time = around 21 ns) are marked AIKW50N65DF5

and the "slow" ones ( Turn-on delay time = around 27 ns) are marked AIKW50N60CT

Then Infineon can charge a bit more money for the fast ones if they feel the need to do that.

Another possibility is that one device is a slightly improved version of the other device. Often small changes are made in the manufacturing process allowing for a "better" device. Again Infineon could charge a bit more money for the newer, faster ones if they feel the need to do that.

Some customers might have a design-in with the older, slower device and have no need for the newer faster device. These customers often do not even want the newer device as that would mean they would have to re-validate their product again (as something has changed). At Infineon they're not foolish, they want to keep their customers so they keep selling the older device as well.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.