My WSPR beacon is back running around 30 Watts, which is just short of midway between the previous two power levels.  (It would be more if the antenna were matched to 50 ohms rahther than 100 ohms, and don't have a suitable 1:SQR(2) transformer at the moment)   ERP is therefore around 4dBm, so the WSPR message is still a lie, but a smaller one now.

 

The PA is a cobbled together esign  made in anticipation of running data modes needing linear amplification some time in the future.  It uses pair of ancient RF MOSFETS is push-pull class AB.  Running from 26V  a 1:2 transformer at the output presents a push-pull load of 12.5 R which gives a max Pout of around 60 - 70 Watts when driven hard.  At the 40 - 50 Watt level the output is absolutely clean, with harmonics < -30dBc.   The devices need about 100mW drive for this output, 200mW for saturation,  so a BFY90 in class A does the job perfectly from an input of +3dBm

 

The FETS (type RFPP53 - your challenge is to find a full data sheet for them !)  were designed for 50V supply and are non-optimum at a 26V supply.   Hence the maximum power is not the 100 Watts it chould be at this voltage and load R.   They came from an old industrial plasma welder at 13.56MHz, and three boards containing 8 devices, generating 1kW each were combined for 3kW total.  I have one of the boards operational, and managed 930 Watts at 14MHz out of it, at which point my 50V PSU current-limited.  But there was clearly more in hand if the PSU could have delivered.    The ferrite transformers  won't let it go much below 5MHz, and the devices roll off badly above 20MHz, so not much use as it stands as an HF PA.  

 

If I can find an SMPSU core spare, will make up a matching transformer to get the antenna down to 50 ohms.

Andy
www.g4jnt.com