>Chuck,
>
>Frequently an amp will take off like that only with a particular combination
>of load and setting of plate/load capacitors.
? True. The SB220 typically get squirrely on 40m.
>That it seems to happen with
>an arbitrary setting of the caps tells me you more likely have something a
>bit more fundamental going on.
>
? To see what's up, measure the resonance at the DC blocking cap. My
guess is that it's c. 100MHz, +/- 20MHz. This will be the most likely
place to find the abode of the invisible squirrel.
- Measure the resonance of the grounded grid. If the freq is lower than
the dip freq at the DC blocking cap, the "grounded" grid isn't, and it
does not shield the input from the output at VHF so regeneration may take
place -- most likely during one's most fun radio contest of the year.
>Not a bad idea to check tubes for shorts, tho you say 2 different sets of
>tubes exhibit the same feature - sounds like low probability.
>
? Unlike 3-500Zs, 572Bs typically do not fil./grid short by bending the
fil.. The 572B's and 811A's filament tends to break during a glitch and
pieces of tungsten fall to the base.
>Not a bad idea to probe for VHF resonances with a GDO. Look in the
>50-150MHz area. While you're at it, see if you get a resonance on the band
>you have it set to.
>
? True, however, tuning the Tune-C capacitance setting has little effect
on the anode's VHF self resonance. This is mostly a function of
conductor L between the anode and the Tune-C and the 5pF or so of anode-C
per bottle.
>Have you tried powering on the amp with the tubes removed? All ok then, I
>suppose?
>
? good advice
>I still suspect you have an issue with the tank circuit - what the tubes are
>seeing looking outbound towards the load. If what you have is a short wire
>(ie the pi-network is not connected, or something close to that), you could
>be setting yourself up for a dandy VHF resonance.
? the ugly VHF resonance is always there - even when everything is
seemingly perfect.
>...
- R. L. Measures, a.k.a. Rich..., 805.386.3734,AG6K,
www.vcnet.com/measures.
end
|