MOSFET / IGBT selection : SOAR

[marc426]

Hi everyone,

Been recently looking over transistor datasheets trying to identify a contender for a medium-voltage (120V) high-amp (phase : 100A cont. 300A peak) controler.
So I've made some simple conduction losses (not completely up to speed on switching losses yet) and obviously the MOSFET is often the preferred choice unless you run >90% throttle all the time.

So there a couple of low RDSon (>5mOhms) MOSFET but the problem is that I don't know what to consider when I look at the SOAR graphs...
First, their main "continuous" rating is definitely not representative since it can only sustain that many amps at very low voltage.
But it also varies a lot depending on the time it is on (I guess that's the tON of the PWM? Am I correct?)...
So what to take as a value for a worst case scenario (except short-circuit obviously)?

If we take a 32kHz PWM ==> T=31,24µS, how long would the PWM be at 100% ==> What's the longest the transistor will be on at a time?

Thanks for your input!

Marc

 

[marc426]

Hi everyone,

To further illustrate my message :
MOSFET : IXFK-FX170N20T
http://ixdev.ixys.com/DataSheet/DS100131(IXFK-FX170N20T).pdf

on fig. 12 it gives me a safe rating only ~50A @ 100V for 1mS. but more than 200A @100V for µS...


On the other hand :
IGBT : FGPF4536
http://www.mouser.com/ds/2/149/FGPF4536-278415.pdf

on fig.10 you have to go below 10µS to hope to get >100A. And @100µS, the SOAR is only around 10A... Pretty lame!!

 

[marc426]

Anyone ?

 

[peters]

Hi,
MOSFETs operate in switch mode in this application, normally there is no high current and high voltage at the same time, just during transition.
The SOA pulse ratings figures are for single pulse only (it is noted on the figures), not related to the tON of the PWM. So need to use the continuous rating for conduction loss even for the peak controller current.
For example at DC current the conduction loss is Id^2*Rdson. At a certain PWM duty cycle it is Id^2*Rdson*duty_cycle. If the current is sinusoidal then need to calculate with its RMS and it becomes more complex.
This is a useful app note about the losses:
http://www.ti.com/lit/an/slpa009a/slpa009a.pdf
Note that parameters change by temperature, for example Rdson roughly doubles by 100°C increase. For rough 'bad case' calculation just double the highlighted datasheet value...
Max. Id is also derated by temperature.

IRFP4568 is a good choice and lower price then others in this category, but you need at least 2 of them in parallel for 300A peak current, and buy from a reliable distributor (farnell for example), because there are many fake parts from Chinese webshops. For the controller's peak current rating you need to ensure the junction temperature remains below its maximum rating, and depending on the cooling effectiveness you can expect a few seconds or a few minutes of peak current.