[Lowfer] LF: PA problem

[N1BUG]

I wonder if anyone can help me with this.

I built one of these:

http://qrp-labs.com/ultimate3/u3mods/lfamp.html

It's not what I want but it's what I can afford. In all cases, I was 
driving it with 250 milliwatts at 137 kHz. It was running into a 
perfect 50 ohm resistive load (dummy load good through 2 GHz).

At first I had a DMM between the 13.8V power supply and the PA to 
measure current. I had preset bias to about 25 mA current with no RF 
drive. Output was about 19 watts and efficiency was only 30%. By 
changing the bias I was able to get 22 watts out. The bias point for 
maximum output was below the point where it draws any idling current 
without RF drive. Efficiency increased to 50%.

Next I removed the DMM. Power output dropped off. Re-adjusting the 
bias I was able to get 15 watts. Bias was still below where it 
starts to draw idling current. Efficiency seemed better as the amp 
was running much cooler.

Then I shortened the cable from power supply to PA. Power dropped 
again. I re-adjusted bias but this time had to make it quite a lot 
higher to peak the output. I didn't have an ammeter in line but from 
earlier tests I know it was now well into the region where current 
is drawn without RF drive. I was able to get 20 watts. At the 
instant I removed RF drive the FET died! I'm guessing oscillation.

In all cases, the output looked clean on an oscilloscope. I was 
using a scopematch to measure power.

I am only guessing but it seems like maybe the PA is not properly 
decoupled from the power supply? Or perhaps it is just unstable. Any 
suggestions? I need to make this work if I am going to be on 2200 
meters this winter. I have one more FET and will order a couple more 
but I would like to have some idea what to try with the next one.

73,
Paul N1BUG