Topband: K3NA Loop Array

[Eric Scace K3NA]

Hi Luis --

    I tried to answer your questions inter-linea.

-- Eric

on 2010 Jun 20 16:04 Luis Mansutti IV3PRK said the following:
> Hi Eric,
>
> many thanks for the kind reply and congratulations for your fine paper.
> I'm very surprised that , since the publication, nobody on the 
> Reflector did mention your loop array.
> Actually it looks like not an easy task, but I would like to 
> understand it.
> Unfortunately I am not an engineer or a technician, just an amateur 
> with financial and commercial background, so I need some help.
>
> Many years ago I built the 4-square Rx vertical array by 
> W7EL-K9UWA-KD9SV design (ARRL Ant.Compendium, Vol.3).
> The verticals were L loaded and phased with 135 degrees; it has been 
> my best Rx antenna for many years, but too much critical and WX 
> dependent. So the next one was with the W8JI resistance loading idea 
> and quadrature feeding, but I'm still getting best results with Flags 
> and Pennants (see http://www.iv3prk.it/rx-antennas.htm )
>
> As you see I have been trying every kind of Rx antenna, including the 
> last TX3A DHDL (quite simple within the flags family), but now I have 
> been interested by your design and I'm trying to fully understand the 
> proceeding. At first I'm going to model it on EZNEC 5 to compare with 
> all the others, but I have these questions:
>
> 1) how are calculated the matching networks? are they like L network 
> stages? could you give me an example with the values you got for  R1, 
> C1, C2, L3,C3? I am interested on 160 m. ONLY.
>
   I dug back into my lab notebooks for 2007 Jun 26.  For 160m I built 
and measured seven units that day on my board design.

R1: zero -- this resistor was only needed on 80m.

C1: 539 to 562.2 pF.  550 would be a good compromise.
C2: 1715 to 1775 pF.  1745 is probably a good compromise.
C3: 250 to 260.3 pG.  255 is a good starting point.
I used 1 kV caps.

L3:  31.2 to 31.9 µH.  This was 45 close-spaced turns on a Magnetics 
Corp part #C0-55122-A2 MPP core.  The core material was selected because 
its properties are relatively insensitive to outdoor temperature changes 
in our area.  "C0" grading has a 2% tolerance.  I used a 90cm length of 
teflon-coate #26 solid wire from Alpha wire.  As it turned out the core 
diameters were just a bit on the small side.

A small 3-terminal gas tube surge arrestor on the loop input side adds a 
couple of pF to the system.

A typical output impedance (toward the coax feedline) was:
    1800: 70.9-j1.1
    1820: 76.2-j0.4
    1840: 78.0+j0.2
    1860: 83.2-j1.4
    1880: 82.9-j0.4
    1900: 79.9-j0.8
    1920: 75.0+j0.0
    1930: 71.2+j0.7
All the units were within 1-2 ohms of these values across the band.

If one has access to a network analyzer, one can fine-tune the 
capacitors while looking a Smith chart plot.  The correct plot is in the 
presentation: a little loop (in the shape of a mirror image of a lower 
case Greek alpha) around the 1.0:1 SWR point in the center of the 
chart.  (Off the top of my head I think all the points plotted within 
1.05:1 SWR circle.)
    C1: controls how quickly the curve folds back on itself in 
mid-band.  Too high a value means the curve looks open, like the mirror 
image of the letter C.  Too low a value means the loop disappears and 
the plot starts to resemble a ">" shape.  10 pF is a big change here; 
getting within 5 pF of the correct value is needed.
    C2: shifts the curve to the left or right.  Too low moves it to the 
left.  Too high puts it to the right.  50 pF deviation is too much.  
Within 25 pF will be OK.
    C3: shifts the curve up or down on the Smith chart.  Too low moves 
the curve too high.  This needs to be within a couple pF of the correct 
value.

The L3 coils all measured quite close to one another as wound.  One 
could push a few windings together/apart to zero in on the design value 
if desired.

The best adjustment procedure was:
    L3
    C2
    C1 (balance the band edge points to be symmetric around j0.0)
    C3 to resonate
Repeat tweak of C1 and C3 to finalize.

Rather than use trimmers, which are not as stable as fixed capacitors 
during outdoor temperature changes, I just padded with fixed value 
capacitor.  I built the units without cutting the capacitor leads, but 
found it necessary to add 1 pF to each capacitor (on 160m) at the end to 
make up for the capacitance lost when the leads were clipped.  OK... 
maybe I was being a bit fussy just to explore how close I could make things!

The balun on the board is a variation of W8JI's design that had better 
results: two binocular BN-73-292 core, glued one on top of the other.  
10 passes of wire (about 20cm) of wire went through the holes in the 
following winding pattern: in bottom-left... out bottom right... in top 
right... out top left... repeat.  The two windings were NOT twisted 
together.

    The coax shield current choke was build from a pair of Amidon 
BN-43-7051 binocular cores (same as FairRite 2843009902, I think).  75 
ohm cable of about 2.5mm diameter was used.  I ran 5 passes through one 
core... then about 10 cm of free cable... and 5 passes through the 
second core.  The two cores were mounted in a plastic box, separated, 
and at right angles to one another.  Twelve were built with the followed 
measured performance:
    Loss: about 0.1 dB at 30 MHz.
    Phase delay: 30-32° at 30 MHz.
    Shield current rejection:  minimum -18 to -21 dB at 1.8 MHz... a 
broad, flat curve (upside-down U shape) of -32 to -33 dB, typically 
centered around 16.3 MHz... and then dropping down to -23 to -25 dB at 
30 MHz.


> 2) how is the required current ratio to the elements 0.54 : 2.00 : 
> 1.00 related to the turns ratio of 13 : 24 : 12 ?

    The combiner is a FT-82-77 core with four interlaced flat windings.  
The output winding (to the receiver) is 13 turns.  The antenna windinds 
are 13, 24 and 12 turns.  I used colored Teflon #26 wire for the 
windings: red white green and black.  For this core and wire size, the 
following lengths were used:
    12 turns: about 27 cm
    13 turns: about 30 cm
    24 turns: about 54 cm

    The step-up matching transformer also used the FT-82-77 core with 7 
turns on the combiner side and 15 turns on the receiver side.  This was 
about 17cm and 35cm of wire, respectively.  A little bit of residual 
inductance remained, which was tuned out with a 247 pF capacitor on the 
receiver side of this transformer.  (I evaluated some other designs but 
the two-winding design had the best performance.)

    With 75 ohm resistors on the antenna ports, the combiner system 
provided these impedances toward the receiver for the first two units 
over the range 1800-1925 kHz:
    78.0+j1.9 at 1800 kHz to 78.3+j1.7 at 1925 kHz
    77.4+j1.6 ... to ...  77.8+j1.2
Essentially identical.

    Driving the system from the RX side, the measured ratios of the 
antenna ports were: 0.542 : 1.993 : 1.000 across the entire band.  There 
was a slight difference of phase between the three outputs: +0.12°: 
+1.0° : +0.7° at the bottom of the band... to 0.00° : +1.2° : +0.5° at 
1925 kHz.  The phasing lines could be adjusted slightly to compensate.

>
>
> Thanks again Eric.
>
> 73
> Luis IV3PRK
>