[TowerTalk] Vertical Dipoles

Eric Gustafson n7cl@mmsi.com
Sat, 2 Jan 1999 12:24:04 -0700 (MST)



Hi Tom,

Yes, I realize that _all_ of my comments apply specifically to
the antenna I was describing and _only_ that antenna.  They are
not intended to be interpereted as "J-poles are generally
wonderful".  But rather as "This feed method should not
necessarily be rejected out of hand".

>Date: Sat, 02 Jan 1999 00:04:53 -0500
>From: "w8ji.tom" <w8ji.tom@MCIONE.com>
>
>Hi Eric,
>
>This is gone so deep into your specific antenna, my points to
>the people considering an HF J-pole are probably missed. My
>points are..
>
>1.) It is easy to "mess up" a J-pole, because it is a *very*
>poor or problematic antenna design (just as end-fed Zepps are).

Problematic - yes.

Poor - not necessarily.



>2.) The unbalanced radiator, fed by a balanced stub, will ALWAYS
>set up some common mode excitation of the stub. That common mode
>current ALWAYS causes loss of FS at zero degrees elevation.

As I review all of this and the antenna we built in my mind, I'm
starting to see a possible reason why ours "played" and didn't
give us some of the comon mode feedline current problems you
describe.  More on this later.

As to the null at zero elevation, this can be a good thing
depending on the actual configuration of the system.  It occurs
at zero degrees relative to the horizon only for antennas that
are actually installed (both stub and radiator) as a true
vertical.  In our case, this effect, to whatever degree it may
have been operating, was probably showing up at quite high
elevation angles.  I can go back to the model and force this
condition to see where the cancellation should happen for our
sloping catenary design.



>3.) How much of a problem that creates is a matter of impedance
>ratios at the top of the stub and the bottom of the 1/2 wl
>radiator.
>
>The J-pole (and Zepp) violates one rule of good design. Never
>make a balanced to unbalanced transition in a system without
>proper consideration of that transition. The J-pole makes TWO
>transitions from balanced to unbalanced. One at the junction of
>the stub and radiator, the other at the junction of the coax and
>stub. Often people ignore both areas, and the result can be MUCH
>less than optimum performance. Some might not know it, and might
>not care, but I suspect many antenna builders want every free dB
>they can get!

Yes, but we are not _required_ to ignore these considerations...



>To maximize performance of a J-pole:
>
>1.) Do NOT ground the bottom of the stub or connect it to any
>large metallic structure or feedline (without a choke balun).

Very good advice.


>2.) Do NOT use a thick radiator, especially when compared to the
>stub thickness.

Also very good advice.


>It's always better to install an antenna knowing what to watch
>out for, if you want to get maximum performance!

Possibly, the best advice of all!


>One way to un-optimize a J-pole would be to build the 1/2 wl
>element out of thick tubing, and the stub from a small open wire
>line. Another would be to ground the bottom end of the stub
>(considered to be an "electrical neutral point") to a tower or
>even a feedline-without-a-balun.
>
>73 Tom
>

I think we may have accidentally mitigated the problem with the
balanced to unbalanced connection at the stub to radiator
interface.

Our radiator was made from the same diameter wire as the stub.
But we were supporting both sides of the stub from the radiator.
This support took the form of splitting the radiator into a "Y"
configuration and mechanically connecting both sides of the stub
to the radiator.  One leg was connected using an insulator (3
inch long ceramic strain insulator) and the other was directly
connected and soldered.  We were pleased to find such a simple
elegant way to support the feeder without putting undue stress on
our spreader to wire connections.

This seemed to work pretty well until we applied high power to
the system.  The amplifier seemed pretty happy until we let the
peak power get up near the maximum.  Then, we started getting
grid current spikes.  Since the time this happened was near the
beginning of the test and we were all running around like
headless chickens trying to get all the last minute details
squared away, we solved the problem by reducing the output power
slightly so this didn't happen.  We'd analyze it later.

Things went pretty well until part way into the first night of the
test.  Then almost simultaneously, Larry started hollering about
weird SWR excursions and one of the ground crew who happend to be
out on a nature break noticed that there seemed to be some "fire
in the sky" near the top end of the stub.  We got out the
binoculars and shined a light up to the junction and were shocked
to discover that the support on the open side of the stub had
come apart.  So we shut down and dropped the antenna to fix the
mechanical connection.

It turned out that the failure had not been a mechanical fault
but that the wire had vaporized where it was hooked through the
insulator and broken.  There were little copper globules on the
open wire ends and sprayed on the insulator surface.  Apparently
the HV at the open end of the stub was enough to cause corona
sufficient to melt the wire in two.  So we doubled up on
insulators and tried again.  This seemed much better and made the
amp happy.  But, now that it was dark, and we were looking, we
could still see a bit of light at the end of the stub.

So we let the antenna down again.  This time we increased the
length of the last several spreaders and put a corona sphere at
the end of the open wire.  Bingo.  No more corona.  No amount of
power caused anything to change.

Now, I'm wondering if the sphere caused the feeder to think it
was seeing a more nearly balanced (current wise) load.  The
sphere diameter was _much_ larger than the wire diameter.  If
this effect was significant (no claims, just speculation), might
it not have forced the "electrically neutral" point on the stub
back closer to where it is supposed to be at the shorted end?  Of
course, it would probably appear almost completely reactive but
would that be a terribly important consideration?  I'll have to
spend some effort to figure out how to calculate the optimum
sphere size for this...

At the other end of the stub, we may have also gotten a bit lucky
due to the specifics of how the thing was actually implemented.
But this is already _way_ long so I won't go into any more
details here.

All I can say for sure is that the same group of beads that
remained at ambient temperature while isolating this feedpoint
from the coax shield got noticeably warm when used at the
feedpoint for a "squashed" delta loop which we used in subsequent
efforts.  When the delta loop system was tried without the beads,
there were _very_ noticeable "RF in the shack" prblems even
though the feedline was laying on the ground for a considerable
distance.

73, Eric  N7CL

--
FAQ on WWW:               http://www.contesting.com/towertalkfaq.html
Submissions:              towertalk@contesting.com
Administrative requests:  towertalk-REQUEST@contesting.com
Problems:                 owner-towertalk@contesting.com
Search:                   http://www.contesting.com/km9p/search.htm