Hi Paul , Nick and any others using that PA design,
(I 've being subscribed to the RSGB LF feed for 4-5 years ? and of
course seen reference to the blacksheep feed but interpreted that as
being a legacy rather than where I recently I subscribed and found to be
where the activity is.
I make it brief in-case I'm so wide off topic I'll fall of the edge of
the planet, and NOT to sound like I know what I'm talking about....
What MAY explain that circuits instability WRT to antenna loads as
opposed to a 50 Ohm load)
Tony's design
http://qrp-labs.com/ultimate3/u3mods/lfamp.html
was made around the era of the QRP Labs U3 using the AD9850 module with
a sine wave + associated DDS spectra through the BS170 PA
IF you are using the current U3S design that uses the Si5351 synthesiser
it is driving through the untuned PA a square wave with a very different
spectra.
I think Hans has described HIS PA as just being untuned broadband in
the sine wave fed U3
and 'Class D in the biased adjusted U3S design, not class E as there are
no frequency dependant components.
To further blur the situation on the referring page,
http://qrp-labs.com/ultimate3/u3mods.html#lf
Tony details coupling capacitor value changes, which are now used in the
U3S design?? But also a change to what will be your driver stage ferrite.
So his PA would probably be driven by this modified U3?
(Everybody has their own goals, favourite topology and most importantly
junk box....
If it were a clean sheet of paper and you wanted a class E PA using a
common switcher FET device, rather than an RF device.)
Work with the fact the Si5351 outputs square waves, use that to drive
the final stage rather than through an (RF) driver stage.
If you need a pulse shaper or just like the idea of a buffer stage, then
do it at logic switching levels in the PA Gate rather than RF.
Hopefully not being interpreted arrogant interloper. :-[
regards Alan
On 09/01/2018 17:36, N1BUG wrote:
Nick,
I was very interested to read your email. I am sorry that you are also
having problems with this PA but at least it isn't just me. A few
comments threaded in below...
I've been, sort of, following this as I have a similar problem with the
amp at <http://www.qrp-labs.com/ultimate3/u3mods/lfamp.html> I have not
used the 2n4 cap to ground on the gate (yet) but I have added a series
resistor of 15R to the gate prior to the DC blocking cap (or should I
have put it after the DC blocking?)
I put my 15R directly at the gate.
With out the 15R it was very fussy and would go into, what I presume,
was oscillation - the voltage trace on my scope match was 'fuzzy' -
looked like AM? with almost no provocations ... however lowering the
supply voltage cured this.
That is similar to my experience. There are two different scope
patterns I see when it starts acting up. Sometimes it will be one,
sometimes the other.
One is as you describe, the voltage trace goes fuzzy, more so at the
peaks than at the zero crossing. This will range from mild fuzz at the
peaks to extreme where it is fuzz from the peak to the zero line.
The other is different. I usually have my scope set to fit about two
full cycles (720 degrees) on the display. Sometimes the first and
second cycle will be of drastically different amplitude even though
both are crisp traces without any fuzz.
I have just re-worked the matching unit at the aerial and now have an
almost perfect 50R j0 match. so I know this is not a reactive issue
Have you tried it into a 50R dummy load? Mine behaves perfectly into a
dummy load but not into the antenna, even though the antenna is tuned
for 50R j0. I am thinking the antenna being reactive *off* the
operating frequency sets off the oscillation.
It still oscillated if I run it at 13.8V and drive of around 250mW (U3s
3 BS170s at 5V), drop the voltage to about 11V and all is well.
Yes, lowering the voltage makes mine more stable also.
Increase the drive to about 800-900mW (U3s 3 BC170s at 13.8V nominal)
and I can wind the supply up to a little over 12V before it goes wild.
The bias is set to give about 75mA at idle.
Mine is far more stable when I run the bias down into the class C
range (no idling current). For any given voltage what I usually do is
start with the bias at zero (wiper of the pot at ground end), apply
drive and turn the bias up until it goes into oscillation... then back
off a bit below the point where it stabilizes again.
If I reduce the bias & therefore idle current then the oscillation is
earlier & easier to provoke,
Unless I am misunderstanding, mine is the opposite. It is more stable
when the bias is set for class C (well below the point of having any
idling current).
if I go much over about 150mA the FET takes
off in thermal runaway with the current going skyward with no change in
bias setting - so a little under 100mA it is.
Mine does that too.
> However if I give the amp a capacitive load it is much
> happier ... and will take over 15V and no oscillation - that
> ought to tell me something but I'm not sure what
My amp is no longer like yours as I have 4 x 10 turns on the
transformer now, but yesterday I noticed it is happier operating
slightly below the antenna resonant frequency than dead on or above.
So I believe I am seeing the same thing.
I think the output transformer is at least part of the problem.
Changing it in some way seems to have more affect on the instability
issue than any other change I have made in circuit. I've also been
told by those with the ability to model them that the low pass filters
for this amp are terrible designs. The specifics went over my head.
Paul N1BUG
|