Should this circuit work?

[fubar.gr]

I wanted to pull a big current in a controlled manner from a car battery, so I made this circuit:



I built the R1 resistor out of a length of wire, using the 4 wire method and immersed it in water. The mosfet was a IRL540n and I controled the gate voltage with a 10 turn pot.

When the mosfet is fully turned on there should be a current of 25 Amps (there was quite some resistance in the wiring too limiting the current from the theoretical 30 Amps).

I attached a heatsink to the mosfet and dumped it into the water too.

It did work for a while, but then the mosfet died (shorted).

IIRC these mosfets can dissipate up to 50W with proper heatsinking and according to my calculations the max power dissipation in that circuit was about 70 W. I guessed that since it was heatsinked and immersed in water it would handle some beating.

Is there something fundamentally wrong with that circuit or I simply pushed the mosfet a little too hard?

Or was the mosfet a counterfeit? I bought them from ebay!

[Kremmen]

Your idea is basically "right" in the sense that this is a way to control the power throught the load. However it has problems which you have already encountered.

Problem #1 is that most power MOSFETs are designed and constructed for switching duty. Their safe operating areas may not be up to the task.
Problem #2 is related to above; today power is controlled not linearly but by switching action. The great benefit of that is that power is not dissipated in the pass element such as your MOSFET here. Instead energy is stored and controllably released using storage elements such as inductors and caps. The resulting circuits are considerably more complex but the losses are also significantly less.

Your circuit has failed because of the excessive dissipation in the pass element and it not really being designed for linear operation in such a regime.

[Simon]

Ah, you did not read all of the datasheet, the first page of most datasheets is bollocks! I've never had complaints of an IRL540N failing. the N version is meant for switching really, if you get the non N version it can take more heat. the IRL540N can do 36A if you keep the die under 25C at 100C it will do 26A and you should never work to the limit. I am talking internal temperature not that of the heatsink. Did you put a real genuine sil pad or thermal compound between the heatsink and mosfet ? I don't think your at a point where you need to ask if the part is genuine but if you understand what your requirements are and what the part can actually do.

[Seekonk]

Nothing wrong with the FET.  Take aq simple case.   If there is 5V across the resistor you have 12.5A.  12.5 x 7V = 88W.  You will never get the heat out fast enough. If you try to dissipate more than 25W you will be asking for trouble.  All those amps are for the saturation condition.  Typically 5 or more FETs in parallel would be used in that situation.

[free_electron]

threw it in water ... 

[Seekonk]

Remember SINCLAIR, like 1970's.  They had  some advertizing with the amp board heat sink just sitting in a bowl of water.  It said that was all you needed.

[fubar.gr]

Thanks guys!

Can you recommend a mosfet more suitable for linear operation and more than 50W dissipation?

threw it in water ... 

Why not? It was just for a short experiment

[janengelbrecht]

Trew it in water ? Eh just part of the heatsink....or the hole lot ? So that connections on MOSFET shorted ? wtf ?

[hamdi.tn]

Nothing wrong with it except too much power to dissipate , better use switching current controller i think.

[Simon]

clean water does not actually conduct very much, you will likely need a larger case mosfet to dissipate more heat, now are you actually trying to achieve something other than learn what mosfets can do by trial and error ?

[max_torque]

The problem is the thermal impedance between the junction and the die!  At those sorts of power levels, even a few degC / W are enough to fry the device even if you have a mythical "perfect" heat sink.  You could easily put a few of those mosfets in parallel that would help!