Here are some characteristics of off-center fed dipoles which I've
observed in doing careful measurements:

1. The feepoint balun is commonly a voltage balun, which may have the
nominal claimed impedance transformation ratio over most of the HF range
when terminated with 50 + j0 ohms times the tranformation ratio.
However, the antenna doesn't have this impedance at hardly any
frequency, and can be very different at some frequencies. When presented
with load impedances typical of the antenna, the transformation ratio is
way off and becomes complex, and the balun adds considerable shunt and
series reactance.

2. Whenever a voltage balun is used to feed an asymmetrical antenna, it
creates an imbalance current in its attempt to equalize the voltages at
the two halves relative to the "cold" side of the input. This imbalance
current flows down the feedline as a common mode current.

3. Additional common mode current results from the unequal mutual
coupling between the feedline and unequal antenna halves.

4. It takes very concentrated efforts to reduce the common mode current
to a low level on all bands. Multiple current baluns (probably what the
CW calls and "isolator") are required, and even then it might also
require feedline length adjustments to get low common mode current on
all bands.

5. Without being able to quantify what the feedpoint balun will do in
terms of transformation, reactance, and common mode current generation,
it's impossible to build a model of one of these antennas with any
confidence, even if the feedline is included in the model. The best
efforts I made to measure a real antenna and its balun and build a model
based on the measurements led to generally poor agreement between the
measured and model impedance. Consequently I'm extremely skeptical of
any model that purports to predict anything about OCF dipole performance.

Roy Lewallen, W7EL

Owen Duffy wrote:
Roy Lewallen wrote in
:

The guy wire requirements would be about the same as for a "current
balun" (common mode choke) -- somewhere around 500 - 1000 ohms is
typically necessary. At that impedance level, it makes no difference
whether the impedance is reactive or resistive from the standpoint of
effectiveness in choking current or in terms of dB loss. But there can
still be enough power dissipated to overheat the cores if they're
resistive and the power level is very high. Then you're stuck with
using ferrites which are more reactive and less resistive (e.g.,
Fair-Rite 60 series), but they also give you a lot less impedance per
core so you need more cores yet. That makes the ferrite solution even
less attractive.

Roy,

I have attempted to model the famous Carolina Windom. It takes some
guessing since it contains proprietary (ie secret) components, namely the
"balun" and the "isolator". My flat top is at 10m height.

Since they argue that the vertical feedline is an effective radiator, I
make the assumption that the balun is a "voltage" balun, and that it is
transparent to common mode current, so I have modelled the feedline as a
wire attached to (as it happens) the inner end of the short leg of the
flat top.

The "isolator" is argued to prevent the current flowing on the line above
itself from flowing on the line below itself. It is a naive notion, since
there is mutual coupling... but lets guess that it is a bunch of
supression sleeves on coax.

At 7.2MHz, the isolator is subject to appreciable current, and it does
not effectively force a current minimum until it is well over 1000+j1000.
(BTW, about half the power is dissipated in a 1000+j1000 isolator.) 2000
+j2000 is becoming reasonably effective.

I know there is a proposition that chokes such as the W2DU balun need
only have choke impedance about 10 times the nominal Zo to be effective.
I think that design constraint is effective in limiting the extent to
which such a balun unbalances the load by its own shunt impedance dropped
accross one load leg for a load~=Zo, but I think the criteria has nothing
to do with the choke's influence in forcing a current minimum in the
region of itself. It is the difference between a bench test criteria for
insertion VSWR, and what is needed when plugged into an NEC model where
the intention is to force a current minimum.

Thoughts?

Owen