Topbanders...
I became intrigued with W8JI's design for a short four square receiving
array during his presentation last year at Dayton. This year, I lost my
usual space for my 2-wire beverages. I was disappointed as they had always
worked very well. With 30 acres, I'm not really short of space but much of
the property is dense timber with some pretty good ravines.
As they say, necessity is the mother of invention. I decided to follow
Tom's design on his web page and build up the array. I started out
building one single element. I used 20' of 1/2" conduit guyed four ways
with 20' of the same military surplus telephone wire I used for the
beverages. I put on four 66' radials laying on top of the ground. The
antenna resonated at about 4.3 Mhz. I began experimenting with L to
determine how much was needed to get resonance at my target frequency of
1825. It took about 36 microhenries. At resonance, it showed a radiation
resistance of about 22 ohms. I put 29 ohms in series and had an antenna
bringing the composite resistance up to 50 ohms, and, at 1825, had a 2:1
bandwidth of about +/- 20 or 25 Khz (if I remember correctly). With only
four 1/8 wavelength radials, I was a little concerned about ground
resistance variation. As it turns out, going from wet to dry, I saw about a
7 ohm variation. I experimented with the series resistance and got the
antenna "centered" on 50 ohms with soil conditions with medium moisture
(between extremes of wet and dry). Obviously, the climatological variation
in ground resistance would go away with more and longer radials (Tom
recommends six 1/4 wave radials I believe). I was trying to minimize
ground radials, at least temporarily, due to the high deer population.
After erecting one element, I decided that 1/2" conduit was just a bit
flimsy for my Iowa climate. I then put up three additional elements using
3/4" conduit which were definitely more stiff. I made the four square 70'
on a side (99' diagonal). I experimented with each element to determine
the correct amount of L and R. They all resonated within about 15 Khz of
each other with 36 microhenries on each one. The series resistance varied
by two or three ohms. I fed my antennas with 50 ohm coax cut for an
electrical 1/4 wavelength. I used 50 ohms (RG58)as I had the coax on hand.
I bought a 160 meter ferrite bead current balun kits for the antenna
feedlines, but, unfortunately discovered that the beads would not fit over
RG58. I did not use the kits. If I did it over again, I would follow
Tom's recommendation and use 75 ohm coax to feed the individual antennas.
This will result in increased bandwidth and lessen the affect of ground
resistance variances. It will also result in more consistent and even
current distribution, which as it turns out, is very important with this
antenna (more below). Incidentally, the composite resonance of the array
comes in about 20 Khz higher than the individual elements. I'm not quite
sure why that is, as the mutual coupling between the antennas seems to be
minimal. I can tune and detune three antennas while measuring one and not
see any apparent change of impedance or resonance of the element being
measured.
While I was putting up the four square, I began building the phasing
switching box as prescribed in Tom's website (w8ji.com). I incorporated the
phasing and switching all in one compact box. I used a 12vdc DPDT and 4PDT
relay from Radio Shack. I used RCA phono connectors for the delay lines
and coax runs to the elements (saved a lot of space). I called the core
supplier (FairRite Products) and attempted to place an order for the cores
Tom specifies. They said the order was too small to process for money but
they would be happy to send me free "samples" which arrived in a couple of
days (I think Tom can supply limited amounts of cores). I wound the cores
with 30 gauge insulated wire from Radio Shack.
After getting the switching/phasing box all assembled I took it out to the
antenna to test it out. I was very disappointed to get an odd pattern with
very poor f/b which varied from direction to direction. I was bummed. I
brought the unit back in to the bench and checked out the wiring and found
that I had omitted a wire. Thinking that this probably fixed the problem, I
took it back out for a test. Nuts...a lousy, yet different pattern.
I checked the transformers by terminating them with the proper values. They
looked fine. I rechecked my wiring. It was fine. I put 51 ohm resistors
on each antenna port and looked at VSWR with the analyzer at various places
internally and at the external antenna port going to the shack. Everything
looked fine...but the pattern was still terrible.
I finally dropped Tom an email. He looked at his original design and
notified me back quickly that, after re-examine it, it contained an error.
He got back to me the next day with the corrected information. I
incorporated the changes in the box. The pattern was much better, although
the f/b was not yet reaching the values which the model showed (20+db). Tom
sent me up his current meter so I could check for current balance between
the four elements. I used a small amount of RF (about 2 watts) to transmit
into the short four square. The four values seemed pretty close but it was
hard to tell exactly how close as the position of the meter clamp on the L/R
wires feeding the antenna would yield somewhat different current levels.
I called Tom. He said to send down the box and delay lines and he'd check
it out. Tom made some minor adjustments to his design, incorporated them
into the unit, made extensive measurements with a vector voltmeter, and
returned the unit. Bingo, it worked very well!! F/B on low angle signals
is slightly more than 20 db. As a result, it is nice and quiet. The
modeling shows the pattern to be fairly close to a full wave beverage. I
do not currently have my 2-wire beverages up so I cannot do A/B comparisons.
However, the F/B and F/S are on a par with my previous 580' beverages. As
with my beverages, I needed some preamp gain to make up for the loss of the
antenna. I use an ARR preamp with about 15 db of gain which yields good
performance. I know it's not really necessary, but I did place the preamp
out at the antenna rather than in the shack (300' of RG8X to the short four
square). The four square array is about 250' from the TX antenna (an
elevated feed 1/4 wave vertical). Port-to-port isolation between the TX
and the short four square (worst case with the four square pointing toward
the TX antenna) is about 52 db. So the four square is nice and quiet and
does not seem to pick up re-radiated noise from the TX antenna. F/B
performance is essentially the same in all directions.
Some lessons learned:
1. Feed the individual antennas with 75 ohm coax.
2. Pay attention to (balance) your radiation
resistance values...this
design will not yield it's full capability if the
current distributions
are not close to being equal. 3. Each individual element radiation
resistance must be measured at the switching/phasing
box end of the
feedline.
4. Make sure your layout is really square.
5. If you see excessive variation in the
radiation resistance between
wet and dry, you need more ground radials.
6. Follow the design carefully...it's easy to make an
error,
particularly in the switching section.
7. The pattern degrades fairly quickly as you get beyond
a 2:1 SWR
(measured at the elements).
8. Make sure each antenna resonates within 15 Khz of each
other.
9. Remember to make your element feedlines an
electrical quarter wave.
As a result, when you're wanting to increase radiation
resistance
(measured at the switching/phasing box)
you'll have to decrease the
resistance.
I know this not the first time someone has used short elements in a four
square configuration for a receiving antenna. However, this design is easy
to duplicate and can be used on multiple bands with a FET follower to switch
the inductance. My compliments to Tom for a good design and his excellent
response and support.
Now, it would be great if we can just get some decent Topband conditions
here in Iowa so that I'll have some more signals to listen to...there hasn't
been much excitement so far this season.
73...Dave
W0FLS
|