I need to be able to switch high current (5 - 24vdc) for a very short time, up to 250ms max and as little as about 30-50ms. I suspect current will be 200-600 amps. I would need the chip to be able to handle the current every 10-30 seconds or so, and it won't be used repeatedly for prolonged periods - I suspect needing to do 5-10 bursts over a 5 minute period.

This is the chip I'm looking at:https://www.infineon.com/dgdl/irfp4368pbf.pdf?fileId=5546d462533600a40153562c61512015
I've also attached the PDF of the chip.

If this isn't the right one, can someone suggest somethign that would work? I don't mind running a few of them in parallel if that will work. I plan on controlling the timing with an arduino.

As far as a heat sink, I have some very large CPU heat sinks from server CPU's that should be more than adequate for this setup.

 

It looks like it can handle the current. But you don't want to drive it with an Arduino. From what I understand, you will need a driver stage.

I am not an expert. Stick around, there might be much better chips.

But no matter what you decide, you will want an isolation stage between switch and the Arduino.

That's a lot of current.

 

Hi RougueRose,
It looks like you the FET you have found would do your job. You might consider putting in a BIG resistor just to limit the current to make sure you dont try to take more than the rated spec of the FET (if you are making a capacitive discharged weld for example.)

I dont really see a need for isolating the FET from your Arduino, and you can also try driving the FET directly. You might consider putting a small value resistor between the arduino and gate connections, to prolong the switch on time and reduce the stress on the arduino output, and to reduce ringing and interference generation, but this will also mean your FET has to dissipate more power as it switches.

If you want very fast switching then you could simply use another smaller mosfet / BJT to switch / drive the larger one (the issue is the "gate capacitance" which you effectively have to charge from your arduino when you switch the FET off and on).

This data sheet:
https://toshiba.semicon-storage.com/info/docget.jsp?did=59460
Is very helpful about ways to drive FETs if you need to, and how to switch them quickly.

 

One of the issues is with gate charge, charging the gate C fast enough so
device spends minimal time passing thru active region where maximum power
is dissipated in device.. Look at the safe area curves to see limitations.

Driving this from an Arduino output w/o a high current gate driver is a recipe
for disaster. You want this device switched fast to minimize power dissipated
while device transitioning thru active region.

Is your load inductive ? If so some extreme voltages are in the offing. L is not
your friend in this design.

You then have the problem associated with PCB routing of high currents
and grounding. Hi current bus bars should be examined. http://espbus.com/
Just keeping PCB/vias together from thermal cycling will be a challenge.

Quite a challenge. Take the case off a plasma welder or tig/mig welder to see how
high current routing is managed.

What is the rep rate of switching ?

The Rdson Pdiss is ~ 666W instantaneous, huge amount of heat to get out of what is
a small package. I would compute Tj. At .3 C/W thats a 200 C rise above case.
Tjmax is 175 C. So if case at room, 25 C, then Tj is 225 C, out of spec.

I am thinking packages like these make more sense to get thermals down.

Or a design where several devices paralleled to distribute heat, current.

Consider contacting an FAE at one of the high power mosfet vendors and
discussing your requirements. This is a challenging design.

I had the pleasure in my career calling on 3 senior designers at a welding company,
I was handling IR as FAE. I kept my mouth shut and learned every trip in. These
guys were practicing an art, at 440 V lines, 100's of amps. I will never know all
the accumulated knowledge they had. They were designing both plasma cutters
and welders of a broad range of power and capability. Designing custom magnetics
to boot.

https://www.electronicdesign.com/boards/calculate-dissipation-mosfets-high-power-supplies

https://www.microsemi.com/document-...unting-considerations-for-the-sot-227-package


Regards, Dana.

 

The high current driver is also important to remove the effects of miller capacitance. With a weak driver like arduino there is a high possibility that during turn-off the inductive kickback could keep the transistor in the linear region for a very long time and that will easily destroy it.

 

Thinking out loud, it may be a good idea to use a gate driver for the mosfet so that it can turn on and off quickly, and an opto isolator between the gate driver and the arduino. This way if anything goes wrong on the mosfet that also affects the gate driver, the opto isolator would protect the arduino. With that much current and assuming any inductance in the load, as was mentioned above you could possibly get some really high voltage spikes with the sudden turn off. A flyback diode would surely help, but the opto isolator would be another layer of protection for the arduino.

 

Everything on the datasheet for currents above 195 A are for pulse widths less than 60 us. It also states that the bonding wires have a 195 A limit. For 600 A, I suggest 4 of these MOSFETS in parallel.

Also, the part has pc board pins. How are you going to get 600 A in and out of it? The packages shown in post #4 are expensive, but they do work.

ak

 

I have never driven a fet with a uP....and looking at the non bonded model....from the data sheet....

doesn't the fet require more voltage to completely turn on for that amount of current.....than the uP can supply?

 

Usually. There is a sub-class called a logic-level FET that is fully enhanced at 5.0 Vgs or less. The tradeoff is a slightly higher Rdson (all other things relatively equal).

ak