Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.naic.edu/~phil/hardware/xmiterhf/dana/TX_wisdom_150517.txt Дата изменения: Mon May 18 14:59:01 2015 Дата индексирования: Tue Apr 12 05:24:30 2016 Кодировка:

NOTES ON TELECON WITH STEVE FLOYD, JIM BREAKALL, DUAN YEAGLEY, & ANGEL VAZQUEZ

(Dana at home, all the others gathered around Jim Breakall's cellular
"speakerphone" at the Dayton Hamvention, 16 May 2015).

This was basically a conversation between Steve and myself, with the others
at Dayton kibitzing.

NOW HERE'S THE BEEF:

The VSWR tolerance of a transmitter like ours depends on a number of
potential problem areas, such as plate voltage swing, voltages and
currents in the elements of the matching network and filter, voltage
and current tolerance at nodes in the Heliax line, etc.

One item in particular that Steve emphasized relates to PA output tube
survival. The idea is, if the negative excursions of the plate voltage
bring the plate down to near the screen grid voltage (or heaven forbid,
below the screen voltage), then the screen grid begins to pull dramatically
higher currents. Note that the screen grid is held at a fixed voltage,
with negligible voltage excursions due to its being heavily bypassed to
GND at RF.

The screen grid has a finite power dissipation limit (before it starts
melting!), and this limit is given on the tube datasheet. Transmitter
metering should show both screen voltage and screen current. We need to
monitor both and be sure to operate the tube so that the actual screen
dissipation is held to no more than 50% of its maximum rating (< 25%
is preferred) at all times. Steve said that the worst case tends to be
when the load impedance presented to the plate is on the high side,
thus leading to a larger than design value of the plate voltage swing.

From the perspective of this particular risk, then, we need to establish
the maximum output power level (and corresponding reflected power) that
keeps screen dissipation within safe limits when operating into the
actual antenna. This should be pretty easy for a single transmitter,
but could be more of a problem for multiple transmitters feeding the
cross-coupled antennas in the array.

But first we should establish that all the tubes are being operated
at reasonable and proper bias conditions per the transmitter manual.

Steve also recommended that we do a sweep test across a few hundred
kHz (including our operating frequency) to check transmitter flatness.
This should be done with dummy loads.
If we see more than about 2 dBpp unflatness, we should consider that
the output network is probably somehow mistuned, and take steps to
correct this (if we can figure out how to do it). This test should
be run primarily on a dummy load, but doing the same on the antenna
and keeping the results might not be a bad idea either. Note that
the transmitters' AGC systems must be switched OFF (as they probably
should be for all our operations, in fact).

Here's my thought for an all-tansmitter VSWR trip point test:

At severely-reduced power on all transmitters with all antennas in
place, tune up the feedpoint phases and amplitudes for proper array
operation. Then slowly advance the power levels on all transmitters
together, while watching PA screen current indications on all the
transmitters at once. When the first one approaches the danger point,
we have arrived at the highest safe power. Stop there and record all
the operating parameters on each transmitter, especially including
the reflected power. Set the transmitters' VSWR trip points
accordingly, then the tubes should be pretty safe. This procedure
has a caveat, which is variations in the phase monitor indications
with feedline temperature. It is likely to be necessary to re-tweak
the phase settings as we approach the maximum safe operating power,
with time allowed for the feedlines to reach their new equilibrium
temperatures.

Stevee suggests that we try to obtain the impedance of the load
presented to the tube with the antenna connected, although what we
do with this info is not necessarily clear (it would be nice if the
TX manual were to indicate what this was meant to be. Perhaps we
can work this out from knowledge of the DC plate voltage, plate
standing current, and the tube characteristic curves from the tube
datasheet.

One way to do this is to repeat the early-2012 measurements with a VNA
clipleaded into place, but with the tube installed (it wan't then) and
the antenna connected (instead of doing a 2-port measurement of the
internal networks as in 2012). That is, we'd be doing a 1-port (S11)
test with the VNA connected from the tube's plate to GND. Actually
we should do this for both the dummy load and the antenna as loads.
If the impedance is higher with the antenna, I think we need to be
especially vigilant for dangerous screen current levels when testing.
Unfortunately, this test may not mean much for the case of all
transmitters operating, becaue of cross-coupling between antennas.

We might also be able to use the 2-port S-parameters recorded in 2012,
by terminating the black-box networks in the antenna impedances and
adding the tube's nominal plate capacitance across the input of each.

If we also added the tube's parallel-equivalent plate resistance at
the input of each network, along with an adjustable-phase current
source, we would seem to have a complete simulation tool at hand for
assessing the whole shootin' match, including plate voltage swings.

The discussion turned also to the passive components of the output
networks, and their voltage and current handling limitations. Here
we are hurting a bit because we have no specifications on the current
handling capabilities of the inductors, and probably not much info
on the voltage breakdown and current handling capabilitied of the
capacitors.

I mentioned to Steve that some component values in the matching net-
work and filter circuits are adjustable, and I opined that these
might be adjusted to improve the situation vis-a-vis plate voltage
swing. He expressed discomfort with this idea, pointing out that
we could unwittingly disturb (redistribute) expected current and
voltage stresses on the network components, as well as degrade
the filter's stopband rejection. He says that we should be working
by trimming feedline lengths and/or adding matching networks at the
antennas. But doing this would completely scotch the present
feedpoint phase monitoring scheme- we would be forced to abandon
our "economy" system and do it "right" (actual current sensing
devices at the antennas and separate coax lines running back to the
TX building.