The power of my WSPR beacon on 503875Hz has gone up by approximately 4dB.   Now running the Class E PA.

 

Thanks for tentative offes of thicker Litz wire, but on reflection, looking at the resulting coil dimensions needed, there isn't much scope for a larger coil diameter without a major chassis rebuild. 

 

The tank coil heating and resulting detune problem were mostly-solved by making up a plastic moulding to duct air from the fan round to the coil.  All a bit Heath-Robinson, but it can now run at the 500 Watt output level at room temp without undue runaway.   And is certainly OK for the low duty cycle WSPR transmissions its intended for.

 

Now outside in the shed, at lower ambient temperature so I'll probbaly be confident enough to leave it runniing without dashing outsuide too often to check.   

There is scope for another dB or even perhaps two by winding the Vdd up to 55 or 60V, but I won't tempt things too much.  The antenan impedance is a bit over 50 ohms, so power output is lower than into the test load, but still in excess of 400 watts.

 

And talking of test loads.    The 1kW 1GHz Bird attenuator increases its attenuation value as it gets warm.   I was wondering why power levels appeared to be falling after a period of testing at max power when DC and temperature were all pretty constant.    A recal using DC when the power-attenuator  was hot, showed more than 0.5dB additional attenuation over its cold value.    Not at all satisfactory.    I'm now using on a voltage / current monitoring bridge for power measurement - that is more reliable.

 

Andy