Rich got it right.
These devices were originally intended for low frequency switching; they can
handle large currents, but only for a very short time. Their use at RF DEPENDS on
finding a very small gate voltage area which allows linear operation.
Sometimes that area does not exist, even on the originals and true copies.
Some later ones were especially designed for use at RF. They are not MOSFET
So using a IRF MOSFET is always a case of possible pig in a poke. Some of
them will not work no matter what one does to them. Fortunately, most will work,
but tuning the gate voltage is absolutely critical. At around 5v, there is a sudden release
of ALL available current at the drain to the source. This is the so-called "avalanche" condition.
It must be avoided at all times if using the device for RF (which means that it is harder to
use them when a full-time carrier is involved -- but it van be done).
The other major consideration is the heat generated. These MOSFETS have very poor heat transfer
characteristics. And there is a very small die area where the "switching' work is done. Heat transfer
must be maximized and strictly controlled. Else, poof! There it goes (in milliseconds). It also
very easily breaks into oscillation, especially VHF oscillation. That is controlled by impedance matching of
the drain to the output.
The best tuning is to set the gate voltage low (2v?) without a drain connection. Then connect the
drain supply. Then slowly increase the gate voltage until a SMALL current increase is seen in the drain.
That is getting as close to the linear region as one is able to do with these devices. This is true
(and a characteristic) of all switching MOSFETS. (The major differences are in the peak voltage, the current
each can handle, and the gate capacitances; these vary tremendously according to their die arrangement).
Then connect and adjust your RF source to the drive transistor...this is also just about the recommended procedure
for the BITX40.
These devices work much better at voltages higher than 12v. Those especially designed
for RF use (like the RDHHFxx) devices ARE designed to be used at 12v. Not true for switching MOSFETS.
The RF devices are also much more expensive, but they are more or less guaranteed to work at HF.
There are a few, like the 2N3555, which were in use very early. They are sort of bastard devices with sometimes
good and sometimes bad characteristics. Note that the 2N3555 is really a MOS clone of the 2N3055, a NPN
ordinary transistor. The 2N3055 gained fame as an output transistor in linear power supplies. This was designed
as a MOS replacement, which in fact does run cooler than the original. Doug Demaw was an early investigator of these
devices, and he noted very early on their "odd" characteristics. When the IRF devices first came out, it was
discovered that they could be used as RF devices (much to the design engineers' surprise), but with all the
caveats noted above.
Incidentally, that Hackaday post is a copy of the one he put on his blog. It, and more, are available in full there.