Having read through this thread and looking at the schematics, I would highly recommend testing having an opto-coupler between the ATMega and the Heater FET's. From what I read in the specs of the IRFZ44N FET's at 5V they aren't all the way on yet, which helps account for some of the heat they are producing. By adding the opto-couplers, you can drive it easily with a full 12 Volts which will saturate the FET (they measure the full current capabilities @ 10V).
It shouldn't take a whole lot to jury rig one in with the supporting resistors instead of the FET resistor, and then jumper in the +12V & ground for a test. Then compare the FET temp after the same runtime to see if it's worthwhile to integrate it in in a future version. To save space, something like an LTV-827 can be used, but you still need to fit in 4 additional resistors as well to drive both.
reference [www.irf.com] with special attentiont to Continous Drain Current on pg 1, Figure 3 Typical Transfer Characteristics chart on page 3, and Figure 6 Typical Gate Charge vs Gate-to-Source voltage on page 4.
It shouldn't take a whole lot to jury rig one in with the supporting resistors instead of the FET resistor, and then jumper in the +12V & ground for a test. Then compare the FET temp after the same runtime to see if it's worthwhile to integrate it in in a future version. To save space, something like an LTV-827 can be used, but you still need to fit in 4 additional resistors as well to drive both.
reference [www.irf.com] with special attentiont to Continous Drain Current on pg 1, Figure 3 Typical Transfer Characteristics chart on page 3, and Figure 6 Typical Gate Charge vs Gate-to-Source voltage on page 4.