Topband: Parallelogram echelon array of beverages

[Bill Conwell]

In looking at different receiving antennas to combat rearward QRM, I
considered a pair of beverages in an echelon (staggered) arrangement
connected as a parallelogram, as described in the (ARRL) ON4UN Low Band
DXing book (listed in the index under W8JI/Parallelogram in my 4th edition
of the book).  This antenna system was attractive due to its reported
endfire directivity.  Endfire directivity offers theoretically perfect
nulls at certain azimuths.  (The book repeatedly references "perfect
nulls.")  I share my musings about this antenna, in case the topic is of
potential interest to others.



When I modeled the described antenna system with EZNEC7/NEC5, my results
didn’t match those published.  In particular, the pattern produced by my
model did not include the deep nulls characteristic of an endfire array.  And
the f/b ratio on 160m was less than 10 db (as compared with a f/b ratio of
more than 16 db in a pattern published in the book).  So I downloaded a
model of one such antenna from
https://www.w8ji.com/echelon-log_beverages.htm.



The downloaded model shows an attractive pattern with deep nulls, at
azimuths where the phasing causes signals received by the two antenna
elements to cancel.  Such phasing is achieved by (a) the physical stagger
distance between the antenna elements, and (b) the electrical phase
difference between the elements introduced by the parallelogram feed system.
(Together, these two factors are said to cooperate, crossfire-fashion, over
an octave or more frequency span to yield endfire cancellation behavior.)



I then turned back to my original model to find where I’d gone wrong.  I
even went so far as to copy the wire table of the downloaded model into my
from-scratch model.  Yet my results continued to differ, with mine not
showing deep nulls.



I then noticed that the downloaded model used MININEC ground.  Changing my
model to use MININEC ground made the difference – my pattern now showed
deep nulls, like the published results.



So why the difference?  In reading the EZNEC manual, I found this warning
about use of MININEC ground, in bold print:



           “Do not use this ground type if the model contains horizontal
wires lower than about 0.2 wavelength.”



The wires in the parallelogram beverage array, of course, are very low -
lower than 0.02 wavelength.  The difference between results with the
MININEC ground model and results with the Real/High Accuracy ground model
is stark.  Based on the warning in the EZNEC documentation, it appears the
results published in the Low Band DXing book, with MININEC ground, do not
accurately reflect real world performance.



So why, if the Real/High Accuracy ground model is taken to be
representative of antenna performance, does the antenna not seem to perform
as advertised (i.e. endfire).  The logic in the antenna design seems sound
– the phase difference due to the physical stagger distance is augmented by
the phase difference due to the different feed wire lengths and should
produce far field cancellation.  So why didn’t it?



I think the answer lies in the single-wire feedlines, of differing lengths,
to the two beverages.  To generate deep nulls, the two antenna elements
must have currents of equal magnitudes - enabling cancellation.  For the
current magnitudes at the two antenna element feedpoints to be equal, the
impedances of the feedpoints must precisely match the impedances of the
feed wires.  That is, there must be no standing waves on the feed wires
that would cause the feed current to depend on the feed wire length.  This
condition evidently is not met with the parallelogram feed system.


In particular, on 160m, EZNEC (using the Real/Extended Accuracy ground)
indicates the feed point current of the first antenna element (wire 1) is
1.20A, while the feed point current of the second antenna element (wire 3)
is 0.93A.  The desired endfire nulling can't occur with such an amplitude
mismatch.  Likewise on 80m.  The feedpoint current of the first antenna
element is 1.54A, while the feed point current of the second element is
0.86A.  (In contrast, in the model with MININEC ground, the currents are
matched to within 7% and 12% on 160m/80m respectively.)


Due to the unequal element currents, the 160m RDF and DMF performance of
the published echelon array of two 150m beverages driven by the
parallelogram feed system (with a 1000 ohm termination resistance) seems
comparable to that of a conventional, single wire 150m beverage with a 350
ohm termination resistance.


The parallelogram system is not without its merits.  It requires no
grounding, so there is no vertical wire component to receive signals
omnidirectionally.  And phasing of two beverage elements, even with unequal
currents, sometimes yields modest improvements in RDF and DMF.


The echelon configuration, itself, is not a problem.  With more
conventional phasing arrangements, a pair of beverages in such a staggered
arrangement can yield attractive endfire behavior.  The problem here seems
to be reliance on single wire feeds of different lengths that are operated
without the required impedance match, which causes the antenna elements to
operate with unequal currents.

I may have made a mistake somewhere in this exercise, in which case I
expect I'll be corrected promptly.  And if the parallelogram antenna system
is to be used on just a single band, it appears the termination resistance
can be optimized to yield better results.  But my impression is that the
arrangement described in the Low Band DXing book does not operate in the
manner expected of an endfire array.  DXers may wish to proceed somewhat
skeptically in considering such a system.




73,

/Bill, K2PO
Portland, OR