My first DF9IC 23cm 250W PA board has been running for a month and it is working well, although I think the output capacitor positions still need a little optimisation. I am using the 9160s from utsource.net and the professional PCBs direct from Henning.
At about 8W drive, my Bird 4410A with a 1GHz slug reports 230W out for 7W in. I think the 4410A 1kW slug reads 9% low at 1296MHz, so that is about 250W in reality. That is using 30V and about 22A.
The 2p2 caps on the output lines are 12,76mm and 11,55mm from the FET body. I will build two more PAs and try to tune those more carefully, but for now, this one is working fine.
I soldered the PCBs to the heatspreader using the same 138C low-temperature solder as I used to fix the FETs.
I am using a water-cooling plate for the PAs. The heatspreader is a copper block 100mm x 71mm x 12.7mm. After 3 minutes of full power, the heat spreader temperature close to the FETs is 23.5C. I think that is cool enough! It works really well on JT modes and CW at full power.
I find that running 1.5A standing current for each FET gives best IMD performance on SSB.
Looking at the two-tone spectrum, with 165W actual output, taking 12.4A at 30V, the output spectrum from a pair of pure tones 1.2kHz apart looks like this:
At higher power, the IMDs start to rise, with the 3rd order -20dB at about 220W actual power, and the 9th and higher start to come up much above that. With speech it looks nice and clean, very straight sides to the spectrum at 150W and still OK at 200.
Next stage is to combine two PAs.
Henning’s website is at www.df9ic.de