That amp with IRF530s was created back in 1999 and known as the "hints and kinks"
amp. When Wa2EBY tried to copy it he got your result blown transistors. Its unstable into
reactive loads (mismatches) and the transformers were poor. It also presents a rather
reactive load to the driver (QRP rig) and is known for making that final oscillate as well.
IF the low pass filter is heating either the design Q is too high or the harmonic content is.
assuming the cores ar T37 size. IF T50 something is wrong as I've run 60W though those.
FYI I also tried that amp (two IRF530s), it tried to blow! I used a current regulated power supply
so it could never exceed the blowup current. The only things that helped was a better output
transformer and a 3DB pad at the input to keep the input tame. I abandoned it.
The 100/150 ohm resistor are negative feedback in an attempt to stabilize the amp.
For the 'EBY amp 2W is max power as at that point that amp is cranking serious power.
That and more power will likely fry the gates.
Look at the WA2EBY amp (it is the two files [HF use fet] in files under my call). With 2W I see
at 80 and 40M about 55W at 28V using 2 IRF510 and about 37W at 10M. I got a board
for it from FARCircuits. I use it with a 4db input pad as most of the HB rigs I have do
about 4W. Do copy the mechanical layout as its critical to cooling and stability.
FYI I bilt mine back around 2006 and its still running despite various "operator errors".
One note. one must use very good care with parallel MOSFETs regarding layout.
They are known to become unstable and fry. Adding a 1-4 ohm SMT (1206) resistor
right at the gate before it goes into the part can calm them. FYI SHORT leads, the
board and the heatsink should nearly be one unit and the MOSFET leads short as possible.