A Mosfet attempts to be a perfect conductor. It can't, but it gets very close. Just like a wire does, it has some resistance, and will dissipate power as heat, depending on that resistance. This becomes a problem when the wire is too long (Voltage droop) or when the voltage + current is too high (wire heats up depending on resistance/cross section/size).
The IRF3205 states that it has a 8 milliohm RDS(on) (Resistance Drain to Source while Fully on). That's 0.008 Ohms. A dead short at 12V would produce
I = V/R or
12 / 0.008 = 1500 Amps. At 5 Amps as you are trying to drive, it should only have
V = I*R or
5 * 0.008 = 0.04 Volts. If we calculate power, that's
P = V*R or
0.04 * 0.008 = 0.00032 Watts. 320 MicroWatts. That shouldn't even register on your thumb.
I say should, as the RDS(on) has three caveats.
- The 8 mΩ is stated as VGS= 10V ID= 62A. You have at best 6 Volts VGS, and obviously only a fraction of the Current at the drain.
- Note 4: Pulse width ≤ 400μs; duty cycle ≤ 2%. You are not pulsing it, you have it at 100% duty cycle.
- Fig 4. shows the Normalized On-Resistance Vs. Junction (Not Case) Temperature. At Room Temperature 25°C it is 1 (Normalized 8 milliOhm? 1 Ohm? I'm not 100% sure) but at 125°C its 1.75 (Normalized, 14 milliOhms?)
The actual RDS(on) is likely higher at a 3~5 Volt VGS, 100% Duty Cycle situation.
Simply wire it as you want, and measure the voltage across the Drain and Source, and the current through the Drain. With those two stats, you can calculate the RDS you are seeing, and then calculate the Power in Watts that it is dissipating.
Note: You have a voltage divider of 1k/1k or half of the input voltage. At Best it's 6 Volts at the gate. And the NPN based Opto is switching on the high/middle side, which may bring that down more. Not an ideal setup.
The LED Strip Controller
The mosfets used in LED strip controllers is simply chosen for best conditions based on a known VGS and assumed 2 Amp ID. The duty cycle and pulse width vary, even at 100% ""light on"" it's not really 100% duty cycle (more like 90%, or less for red). If we guess worst case, crap RDS for those at say 0.5 Ohm, that's only 2 Watts, which can be handled with copper traces on the controller board. Ideally, the RDS should be a very small fraction of an Ohm. 0.5 is not good. 0.005 is 2 orders of magnitude better.