[CQ-Contest] Stack Switching
halken at comcast.net
Thu Jun 5 22:49:15 EDT 2008
I mentioned earlier today that there were one or more errors in the
Nov/Dec 2004 NCJ article: "Improved Stack Switching with Invisible
Stubs." Before I mention two small errors, and expand on the subject a
little, I'd like to compliment Terry, N4TZ, for a first-class article
that most of us can learn from.
1. The text just below the dashed box in Figure 2 says
"Non-shorting..." It should say "Shorting."
2. With respect to Figure 1, the statement in the last paragraph of the
first column: "When antenna 1 is selected the unselected positions
"both" and "2" reflect the open to a short one-quarter wavelength away
which is then transformed by the remaining phasing line to an open...."
is simply in error. When antenna 1 is selected, the open on the "both"
position is transformed by the 50 ohm section to a short at the tee, but
the 75 ohm section connected to the switch and antenna 2 does not see an
open - it sees antenna 2, which is nominally 50 ohms. The 75 ohm
section transforms the 50 ohm antenna load to 112 ohms at the tee, which
is in parallel with the 'short' provided the 50 ohm stub. The practical
effect of this is negligible - the circuit will work fine. However,
real-world stubs provide about 30 dB of isolation, not infinity, so a
small amount of coupling between antennas 1 and 2 will occur.
Two other considerations:
1. Configurations that depend on a short via a switch or relay at one
end of a quarter-wave section to provide a stub that is open at the
other end (figures 2 and 4) have a design issue with switch or relay
heating and potential catastrophic failure. A little math and the
knowledge that a quarter-wave section is a voltage-to-current converter,
elucidates the problem. Consider switch or relay contacts with a
contact resistance of 0.1 ohms on one end of a perfect 50 ohm
quarter-wave section. The "open" end impedance is Zo squared divided by
Zsource or 25,000 ohms. The voltage at the open end of the stub at
1500W and 1:1 SWR is 274 Vrms. 274 volts imposed on 25,000 ohms draws 3
watts. Where is the 3 watts dissipated? At the 0.1 ohm switch
contacts. This is enough to seriously heat contacts. At 1.0 ohms the
power dissipated is 30 watts - more than enough to destroy any of the
contacts in common amateur use. Making matters worse - metal contacts
all have positive temperature coefficients of resistance, so as they
heat their resistance goes up - sometimes resulting in thermal runaway,
i.e. catastrophic failure. In my experience most folks started
switching two-stacks using commonly available shorting switches per
Figure 2. Ever notice these switches get hot? I have personally seen
them fail in a shower of sparks at 3AM. I don't know why it's always at
3 AM. High quality switches and relays with clean contacts, used in the
shack, will last years used this way - but it's a good idea to feel them
for heating now and then. Relay contacts on the tower typically see
less than a full 1500W, but tend to get degraded by their environment.
I'm convinced a lot of on-the-tower stack switch failures that get
blamed on lightening are actually due to overheating in the shorted-stub
application. Do the math for 2:1 SWR - it's not pretty. For this
reason, I strongly prefer Figure 1 to Figure 2.
2. Beyond the scope of the article and this email is what actually
happens to undriven antennas in the stack. They are always
parasitically coupled to the driven antenna(s) and always radiate. They
are never truly "out of the stack." I've questioned lots of owners of
stacks about whether their undriven antennas are a quarter or half
wavelength away from the switching, and whether the switching is
providing a short or an open. The usual answer is: "Don't know - the
coaxes are all equal length," and, "Does it matter?" It can matter a
lot and modeling is often done in the ideal, which is to say improperly.
Most stacks in "top" or "bottom" are not up to par, in that significant
radiation at some uncontrolled phase angle is coming off the undriven
antennas.... and, this can be minimized with proper selection of
transmission line lengths between the antennas and the switches.
Sorry for the bandwidth.
More information about the CQ-Contest