>>Here's how I am testing:
>>Key up amp, no RF, Ctune at min, Cload at max, open circuit on output
>>(ie: no antenna or dummy load - no coax attached). I figured this
>>be worst case.
>Sure is and not a reliable test of the amps operational stability. Try it
>with real world tuning values and then remove the load. A good layout
>should then be stable.
>In a perfect world what you are trying should work but the 4-1000A at
>that voltage is just asking to take off, why encourage it? There are
>even some commercial ham amps that would not pass that test.
OK. Well, I guess I am just of the school that wants an amplifier to be
bullet proof. Under "reasonable" real world values, I have no
oscillation. It's just when my turns counter on my UCS-300 gets very low
(about 5 with a minimum of 0) that I have the problem. It's probably
barely a crank from its absolute minimum. Realistically, I'd never take
it that low even on 10m.
So are you saying that I don't really need to worry here?
>>What happens is I seems to have a housing based resonance problem with
>>one of my tubes.
>Just one tube or all ?
>What bandswitch position?
Just one of my tubes. Perhaps it's a problem with the tube and not the
amp. I've a suspicion that this tube is possibly flaky. Bandswitch
position is 10m.
>> When the cover is off, every thing is fine.
>>when I put the cover on and bring Ctune close to its minimum value,
>>plate current starts to rise. I must have a resonance in the tank
>>I tried adding a 10 pF vacuum cap in parallel with Ctune as some have
>>suggested to improve stability in the past. On paper you would think
>Yuk ! All you are doing is increasing the chance for problems. On top of
>that the 10pf added to the approx 7pf of the Tune cap and the approx 8pf
>of the tube puts you well out of range for a reasonable Q on 10M and is
>even marginal on 15M. Throw in a few more pf for strays.
OK. Help me out here then. When I do tune up on 10 M, the turns counter
is well above its value from where it starts to oscillate. No, I don't
have an idea of what value of C. I figure it must be at least higher
than 20 pF. The minimum value of the UCS-300 is 10 pF. My guess is that
it starts to oscillate somewhere between 10 and 20 pF. So I thought
adding another 10 pF in parallel should keep the minimum capacitance
above where oscillation occurs (if oscillation starts at say, 15pf, then
an addition of 10 pf will always ensure that the minimum Ctune is 20 pF
which is above the oscillation value). I understand what you are saying
about Q and that dovetails with some of our other correspondence on the
pi-net values. Perhaps my 10M circuitry does not have good Q and that's
why my power out is poor on 10.
>If you have a GDO what is the oscillating freq in diode mode?
Yes, I do. I haven't measured that though. I will try that this
evening. How would one do this? Just start the amp oscillating and then
tune through the GDO to find the oscillating frequency? Do I hold the
GDO just above the PA cabinet?
>Is the tube base adequately grounded with the required clips? The base is
>a shield and MUST be grounded.
Yes, the base has 4 clips all attached directly to the floor of the amp.
And all 4 clips are in very good contact with the base.
>>Should I add some "walls"
>>the housing to try to break things up?
>Some EchoSorb on the walls or cover may help but that is a brute force
>fix. Also try a different plate choke on the chance that it has a common
>series resonance with the enclosure. The GDO may be of some help.
Resonance of the plate choke in the circuit is about 22.5 MHz with
another resonance around 35 MHz.
I thought about the absorber stuff. Wish I had some! However, I am not
sure how effective microwave absorber will be at HF frequencies. We used
to use this sort of trick in our 6 GHz amplifiers back at Motorola. Yes,
it's brute force. I even thought of the idea of using a plexiglass lid.
However, that isn't a real solution either.
"A life lived in fear is a life half lived."
FAQ on WWW: http://www.contesting.com/ampfaq.html
Administrative requests: amps-REQUEST@contesting.com