Author Topic: Do you have experience with SPICE simulation of IGBT switching, how realistic ?  (Read 977 times)

0 Members and 1 Guest are viewing this topic.

Offline MiyukiTopic starter

  • Frequent Contributor
  • **
  • Posts: 908
  • Country: cz
    • Me on youtube
Hi folks,

Do you have experience with simulating IGBTs switching behavior and losses ? I know it will be hugely affected by provided models. I want to use infineons devices.
I know they tell simulation dont replace measuring a prototype, but are results at least reasonable close to reality? Like +-10-20% ?
 

Offline fourtytwo42

  • Super Contributor
  • ***
  • Posts: 1218
  • Country: gb
  • Interested in all things green/ECO NOT political
Surprised nobody else answered you however simulations especially switching depend on how well you simulate all the other circuit elements, especially the driver, the load if its inductive, any current the diode is going to handle and any potential avalanche power. I have done plenty of sims and actuals with mosfets, less so with igbt's and I would not say the sims have been an order of magnitude out but as I say it depends how much effort you put into modelling. Tuning at the prototype is fairly easy unless you make a bad mistake (heatsink to small, bad layout resulting in excessive ringing etc).
 

Offline MiyukiTopic starter

  • Frequent Contributor
  • **
  • Posts: 908
  • Country: cz
    • Me on youtube
I know modeling can be tricky. But basically from layout data you can get all traces stray inductance, what is main issue
Traces coupling will be missing but it is more issue on small signal control loops 
Then is only question how models represent behavior of active devices
 

Offline fourtytwo42

  • Super Contributor
  • ***
  • Posts: 1218
  • Country: gb
  • Interested in all things green/ECO NOT political
To answer that you would have to inspect the model for the accuracy of the parameters that concern you. Most semi manufacturers mosfet/igbt models I have encountered are good enough for normal purposes however I have encountered some rubbish models on the general web through searching for component xyz so it just the same as the old saying "buyer beware".
 

Offline Yansi

  • Super Contributor
  • ***
  • Posts: 3930
  • Country: 00
  • STM32, STM8, AVR, 8051
Why do you need to model it? All you need is the Eon Eoff energy from the datasheet and the Vcesat(Ic) curves.
 

Offline MiyukiTopic starter

  • Frequent Contributor
  • **
  • Posts: 908
  • Country: cz
    • Me on youtube
Why do you need to model it? All you need is the Eon Eoff energy from the datasheet and the Vcesat(Ic) curves.
Eon Eoff are very layout dependent (stray inductance)
and Eon are for specific diode, yes it can be calculated to other diode by Crr to some degree

especially at fast IGBTs with tr 7ns and tf 21ns at Ic 25A 
at this speed other circuit components have huge impact
« Last Edit: April 25, 2020, 05:22:21 pm by Miyuki »
 

Offline Yansi

  • Super Contributor
  • ***
  • Posts: 3930
  • Country: 00
  • STM32, STM8, AVR, 8051
I think with IGBT (which are rather on the slow side of switching). parasitic ESL and C of the layout is hardly of concern.  But when talking MOSFET/SiC, that would be quite a different matter.

But a good point with the diode of course. Hence you should prefer IGBTs with integrated diodes: Both you get the correct Eon (that includes the recovery charge from the opposing diode) and you get much better chances at the layout.
 

Offline MiyukiTopic starter

  • Frequent Contributor
  • **
  • Posts: 908
  • Country: cz
    • Me on youtube
Yes with slow IGBT it is not a concern, but today there are many high speed IGBTs even for high current (higher tens of Amps), where they have better performance than 600V mosfets even at 50-100kHz range
And as they are much cheaper for same current rating it is interesting alternative
But this frequency range start to be tricky with layout

Also there are many typologies what rely on external diode, for example recovery of diode on second side of converter transfer to loss in primary side transistors
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf