IGBT maximum allowable current pulse width

[sabu31]

Hi all,

I am using IGBT H-Bridge for purpose of pulse generation using a high voltage pulse transformer. The pulse duration is 10us and the repetition is 250Hz.

The datasheet of IGBT is attached and its shows Icrm is 3*Inom. What is the pulse duration considered typically for these datasheet values. When my primary current exceeds around 520 A, the Vce sat protection is limiting pulse width. Is there a way to increase the limit without causing damage to device. I am using Skyper 32 driver board.

 

[KlausST]

Hi,

the datashhet isn´t very informative about SOA.

Thus I expect that the manufaucturer provides additional informations.
Did you check on this?

Klaus

 

[Easy peasy]

you do not want to run a single igbt much above 350A pk, even for 10uS burst, 250 times every second...

 

[FvM]

You are asking two questions in one, 1. pulse rating of the IGBT, 2. programming of the Skyper 32 desaturation threshold. Desaturation threshold can be adjusted as described in the datasheet.

SOA can be constructed from dynamic transient impedance curve, unfortunately output characteristic above 400 A isn't specified. A rough calculation of switching and on-state losses reveals, that switching losses are dominant with 10 µs pulse duration. The total power energy and power dissipation at 250 Hz repetition rate can be handled by the device.

The other point is thermal wear and lifetime, addressed by Easy peasy.

 

[danadakk]

https://www.semikron.com/dl/service-support/downloads/download/semikron-application-note-thermal-resistances-of-igbt-modules-en-2014-11-30-rev-01/

SOA https://library.e.abb.com/public/78...a/5SYA2057_IGBT diode safe operating area.pdf

You should also talk directly to Semikron tech support on the characterization data
for that specific module and your application.


Regards, Dana.

 

[Easy peasy]

also - why are you not paralleling IGBT's ?

 

[sabu31]

You are asking two questions in one, 1. pulse rating of the IGBT, 2. programming of the Skyper 32 desaturation threshold. Desaturation threshold can be adjusted as described in the datasheet.

SOA can be constructed from dynamic transient impedance curve, unfortunately output characteristic above 400 A isn't specified. A rough calculation of switching and on-state losses reveals, that switching losses are dominant with 10 µs pulse duration. The total power energy and power dissipation at 250 Hz repetition rate can be handled by the device.

The other point is thermal wear and lifetime, addressed by Easy peasy.

Thankyou FvM for the reply. Even though its mentioned 3*Icm in the datasheet. What is your recommendation limit current for for 200 A IGBT device for such type operation.

--- Updated Nov 23, 2021 ---

also - why are you not paralleling IGBT's ?

I considered based on datasheet that it has limit for 600A. What is the typical frequency and pulse duration considered for evaluating peak repetitive pulse.

 

[Easy peasy]

I considered based on datasheet that it has limit for 600A

Have you calculated the die temp at the end of this pulse ?

 

[D.A.(Tony)Stewart]

Parasitic ESL stored energy must be limited by layout from 1/2ESL*I*2 going into IGBT parasitic C (pF) by mJ limits. Rce is around 7 mohms so the total current of reactive and real current both contribute to losses but they do not indicate repetitive SOA just single pulse. So as long as you can have some margin for mJ and SOA pulse, to allow for Rce rising slightly with temperature. Thus to avoid a thermal runaway condition thermal power resistance must be lower than the added electrical power dissipated per deg C to have a loop gain <1 for temperature. So to add margin, water cooling may be needed or parallel IGBT's and use Litz wire for interconnects.

You might also need dynamic snubbers. Simulation with various RLC parameters can also help. I suggest searching Mitsubishi site for latest-generation technology. 8G? Even the 7G IGBT's 600A will be better for thermal resistance which use solid copper to emitter where you monitor with a thermocouple. Old generations use tiny gold wire bonds which are the weakest link to fusing and thermal cycle induced shearing.

The Infineon ones are 4G (4th generation)

 

[sabu31]

Hi all,

I need one clarification. Wheter using two 200A devices((SKM200GB12E4) ) in parallel is better or one 300A((SKM300GB12E4) (as its Icm is 900A) my peak pulse will be 550A. Or should I go to 400A device.(SKM400GB12E4) All of them have same footprint.