[ARC5] Hello Smart People: Output Network "L"

[Ian Wilson]

This thread prompted me to dig out the 813 chassis that has always given me
problems.
With no voltages of any kind on the 813, small 5k resistor from plate end
of pi network to
ground, looking into the output with an antenna analyzer:

a) disconnected the 813 plate connector. This tunes up OK; I can get a SWR
of less than
    1.1 with one of the available inductor taps. The inductance and tuning
capacitance agree
    with a pi network calculator. The loading cap value seems rather high,
though.

b) added the 813 plate connector. Not able to get SWR much less than 2.5.
Tried routing
    additional ground paths to the 813 screen grid temporarily - no change.

c) remove the plate connector, tried a 15pF mica capacitor in place of the
tube. Depending
    on where I ground this, the tuning behavior ranges from OK to dreadful.

So, I think I have a ground loop (i.e. an inductive path) somewhere in the
output layout.
My layout looks something like this:
Top of the chassis has the tapped inductor, plate tuning variable, and
tube, in a sort of
L shape (inductor and tuning cap adjacent to the front panel, tube set back
about 7"
from the panel. The loading capacitor (a multisection receiver tuning cap)
is under the
chassis, directly under the inductor.

The tube socket is recessed under the chassis. There are heavy connections
from g3
and the g2 decoupling caps to the frame of the loading cap.

Any suggestions about how to improve things? The operational effect is that
the output
doesn't dip, and efficiency is poor (grid current/drive are OK).

73, ian K3IMW