Current limiter: should not be needed save for running on bench supply. If used put it in the radio and set it to about 2.5A. For 8W out at 13.8V your current should never be close to that. The amp is about 40% efficient and for 6W out your doing max 16W input power ( less than 2A at 13.8V) in and the difference is in the form of heat. I try to not use one as they tend to limit on voice peaks and the audio suffers badly.
My experience with IRF510s is extensive and much different. I rarely kill one even during experimenting.
I use them for RF power though 6M. How much 8 in a parallel 4x4 push-pull at 6M running 225W! My favorite is two in a push pull amp (K500r/wa2eby design) with 28V running about 55W on 40m and 37W at 10M and after 6 years of wrong antenna and all sorts of usual havoc its still running with the same pair. The two device amp uses a heatsink with 4x6 inches of surface and fins nearly an inch tall. This is a sample of two amp I've used for a while and I have others.
Many things kill the IRF510. Heat, the usual heat sink seen is not big enough. Bad construction causing instability. Excessive voltage its a 100V device but if the combined DC+RF exceeds that its gone. The last one is sneaky that is far to much gate voltage. Exceed 20V and the get punches through and the part goes away. How can that happen put 30V on the drain and usse a gate drive that is not sufficiently stiff and low enough in impedance and the RF voltage coupled back by the Gate to Drain capacitance will finish it off fast!
The part is rated 100V max and 5.6A max but only as a low speed saturated switch. The power limit is 43W dissipation but that is on a large heatsink with derating for heating. The derating is .29W for every degree centigrade above 25C for the DIE temperature the case will be cooler than that due to thermal resistance. The TO220 package has terrible thermal properties so if its heating the heatsink to 50C the die inside is near its failure point.
If the heatsink gets more than just warm its too small or insufficient airflow.
Bias, there is a right value and if high the heatsink better be good! Also the bias set the gain to a point so if the amp is unstable at 100ma look at layout or wiring.
The transformer is not 4:1 its 1:4 the drain to get in the 5-8W region at 12V needs to be at about 12.5 Ohms impedance not 200 (under a watt possible)! The transformer matches the Drain to the design load 12 or so ohms to the load of of 50 ohms. The low pass filter is symmetrical is there for harmonic cleanup but a few I've see use asymmetric values to translate from less than 50 ohsm to 50 ohm as well. This also means if the antenna is poor and the load impedance is lower than 50 ohms the reflected impedance to the IRF510 is even lower and the power it will try to generate will be higher but if the cooling is not adequate it may die!
The impedance at the gate is not a high impedance as that only occurs at DC, for RF its roughly the gate capacitance to ground plus lead inductance. That means way less than 150 Ohms is typical even at 40M. If the source is not stiff impedance wise then instability due to Gate to Drain capacitance is probable.
Long leads. Just don't. Especially the Source lead. My favorite layout is very short source to ground (near the case) cut the drain lead and use the tab with the part flat on a heatsink. The inptu on one side of the part and the IRF510 between the long way and the output on the other. Works even at 6M.
Resistance in the source lead. works at DC but at RF it tends to result in the source elevated above ground (even with a capacitor) make the stability at RF poor. It can make a fine power oscillator.
As the device is very capable of high gain keep the input as far as possible from the output. Typical gain biased as AB1 linear amp is better than 13DB at 50mhz!
Bipolar such as the 2n1969 cannot sub direct as the bias circuit needs to be much different for good thermal and RF stability as well as acceptable linearity. Its a good part but apples and mangos.
RF power fets like the RD or MRF parts have the same or very similar constraints. Input and output impedance needs to be controlled by the surrounding circuits.