I got the errant amplifier (a copy of the GJ4ICD design) working - what it
needed was more L in the pi network. As built, the anode load could be set
in the range 600-2000 ohms, which was a bit too low. Increasing the inductor
by 1 turn took it to 800-2800, and the amp then did a comfortable 1.6k out
at 50% efficiency.
The original inductor measured 270nH, the replacement was 390nH (measured at
50MHz with a network analyser). The effect of the inductor is changed
considerably when the lid is put in place.
Another point of interest that came out of measuring it:
Measuring C with the analyser, there is a ground and live connection, not a
true floating system. Looking at the 8877 in isolation, ground on the anode
and live on the grid reads 12pF, close to the data sheet value. Ground on
the grid and live on the anode reads 19pF, even when the valve is suspended
on threads, well clear of any objects. In the socket in the box but no
connection to the anode, it's up to 23pF, and with the anode feed in place
and connections up to the blocking cap (but not the tune cap) and the lid
on, it's 29pF. A lot more than might be imagined from the data sheet.
I see the same in a MLA2500 (my own conversion for 6m). The valve spec. is
7pF Ca - in the socket but unconnected it measures 15pF. Connect both
valves, hang on the anode feed, blocking caps, and tune cap on minimum C,
and it's up to 60pF! Ok, my layout is poor and the tuning cap is not the
ideal one for the job, but again there's a lot more C than might be
predicted from datasheet + strays.
If you can't measure what's going on at the frequency in question, it's
going to be hit and miss getting such amplifiers working.
Steve
--
FAQ on WWW: http://www.contesting.com/FAQ/amps
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|