So as a final test today I put a 50 uH choke and then a bypass
capacitor on the power supply side of it.
With 100 uF I (oddly) found 2V pk-pk RF on the power supply. So it
is worse with the choke plus bypass capacitor than with just the
bypass capacitor alone???
With 330 uF it was better. With 1000 uF better still.
Finally with 2200uF there is only about 100 mV pk-pk RF there. That
is the best I have seen so far. The amp seems happy. It is producing
25W RF output which is the most I have seen. Efficiency comes out to
about 70% with bias well below the point of any idling current. The
drain waveform is hard to describe... looks a square wave with
rounded edges and some other irregularities... but at least there
are no big spikes.
Now there is no change when I insert / remove the DMM from the power
supply lead.
By the way I am driving it with a sine wave. Maybe tomorrow I will
try a square wave to see what that does.
73,
Paul N1BUG
On 11/10/2017 04:53 PM, N1BUG wrote:
OK. I understand the problems with having RF reach the meter and
power supply.
To be honest I don't fully understand what you are suggesting with a
very small coil (few uH at best) and capacitor...
On my own lack of knowledge I put a 1.2 mH choke in the DC supply
lead at the amplifier. This made very little reduction in RF I was
seeing on the DC side of the choke!
So I put a big (1000 uF) capacitor to ground from the DC side of the
choke. That removed all the RF there but make the amplifier output
dirty. Seemed maybe it went into some lower frequency oscillation. I
tried 330 uF, which seemed to make it dirty for a couple of seconds,
then it cleaned up. With 100 uF it is always clean and there is very
little RF component on the DC supply lead.
It seems almost as good to have the 100 uF capacitor but remove the
big choke. Maybe this is what you were saying, that a big choke is
not needed. I will try replacing it with just a few uH and see what
happens.
It's supposed to be a class AB2 amplifier I guess, but when set for
any amount of idling current, efficiency is extremely poor. About
30%. If I set the bias lower, well below the point where it draws
any idling current, efficiency becomes about 80%. As I understand
it, the circuit is not correct for class E but that seems to be how
it wants to run.
73,
Paul
|
