I dont think you could use the VNP49N04. The problem in particular is the section that states When the current limiter is active, the device operates in the linear region, so power dissipation may exceed the capability of the heatsink.

This means that rather than turning off altogether, the gate-source voltage (the signal that turns the MOSFET on and off) is reduced so the MOSFET is operating in the linear region. The linear region means that the MOSFET is neither fully on nor fully off. Only when it is fully on or off is it dissipating reasonable amounts of power. When it is half on and half off, the power dissipation can be hundreds of times larger. Then the temperature will rise rapidly, and the MOSFET temperature protection will shut it down. Im not sure this is a nice way to operate the device, and will certainbly reduce its reliability.

I have designed a current limiting circuit, but its built into a speed controller design. You can find it in section 10 here: http://homepages.which.net/~paul.hills/SpeedControl/SpeedControllers.htmlhttp://homepages.which.net/~paul.hil...ntrollers.html
This could be redesigned to reduce the PWM ratio based on the current. I havent time to do this at the moment Im afraid, but Ill bear it in mind for the future.

Have a go at using the 40N04 by all means, and report your results. Good luck!