[Top] [All Lists]

Re: [TowerTalk] balanced line loss on a mismatched antenna

To: "Gary Slagel" <>, <>
Subject: Re: [TowerTalk] balanced line loss on a mismatched antenna
From: "Paul Christensen" <>
Date: Tue, 28 Feb 2012 11:18:01 -0500
List-post: <">>

I imported the same model into 4Nec2 and scaled the frequency to 7.0 MHz.  
Feedpoint Z is 4.6-j17.7.  Your Z values look quite extreme (also see notes 
below about feed Z for the 8JK).  

Using Zo=450-ohm line, two points result in minimum SWR(50) at the line input.  
The first occurs at 65 ft where SWR(50) is 7:1 and a total loss of 1.8 dB.  The 
second point will occur a half-wave more at 130 ft. where SWR(50) is 5:1 and 
total loss is 3.0dB.  VSWR is lower, but it's the result of additional loss due 
to SWR.  So, loss is higher, but SWR is lower due to added loss.

Me, I would add an L network at the feedpoint.  First, look at transforming the 
low feedpoint Z to 450+j0.  What L network values?  A high-pass can be made 
with 800 pF C in series at the feed, followed by a shunt of 1 uH of L.  L would 
be a small turn or hair-pin of wire.  That combination results in a perfect 
match from 4.6-j17.7 into 450+j0.  Total L network loss is only 0.23 dB with 
reasonable component Q. Finally, let's see what 150 feet of loss is with the L 
in place.  Well, SWR(450) is 1:1, and line loss is now only 0.1 dB but SWR(50) 
is 9:1 at the line input.  So, a tuner must be used between the Tx and line 

How about a deliberate mismatch at the feedpoint to some intermediate Z value 
to allow for line length trimming?  We can still use an L network, but now 
convert 4.6-j17.7 into 50+j0. That's 0.37 uH across the feed, followed by 700 
pF of series C.   At 127 feet of line, SWR(50) is now 1.1:1 and line loss is 
respectable at 0.4 dB (+0.23 dB of network loss = 0.63 dB total) and -- no 
tuner required!

FYI - Here's the comment section for the 4Nec2 file concerning the 8JK:

"Originally designed by John Kraus, W8JK in about 1940, this antenna has some 
interesting properties. It by two closely spaced elements driven out of phase.
Although the fields from the elements don in any direction, gain is nonetheless 
achieved because of lowering of the radiation resistance due to mutual
coupling. And lower it is -- note the feedpoint impedance of only 4.74 - j19 
ohms. Compare this to a single
element. The lower resistance results in heavier current, hence greater field 
strength, for a given power input. The
difficulty is that system losses can quickly eat up the gain. Making this 
antenna from #12 copper wire (try it --
and include wire loss) drops the gain about 0.65 dB, not too bad. But great 
attention must be paid to losses in
matching networks. And losses rapidly increase in significance as the spacing 
is made closer than the 0.1
wavelength of the example. When mounted low (0.25 wavelength for the example), 
W8JK-type antennas have a
lower radiation angle than many other horizontal antennas due to the inherent 
lack of high-angle radiation. As an
interesting exercise, save the pattern for later comparison. Then delete the 
second source, making the
antenna into a Yagi. Note the increased gain. Even though the "takeoff angle" 
is higher, the Yagi gain is as good or
better even at lower angles. In addition, the feedpoint impedance has increased 
to a much more manageable value. On
the other hand, the W8JK will retain its performance over a much greater 
frequency range than the Yagi."

Paul, W9AC

  ----- Original Message ----- 
  From: Gary Slagel 
  To: Paul Christensen ; TowerTalk 
  Sent: Tuesday, February 28, 2012 10:11 AM
  Subject: Re: [TowerTalk] balanced line loss on a mismatched antenna

  Thanks Paul,

  I've attached the Source Data output below.  This, I believe, is showing me 
SWR and impedance at the feedpoint of each element.  I'm not sure if EZNEC 
gives me a way to see the SWR at the single feedpoint that I'll use to feed 
these two feedpoints but I assume it will be equally as bad.  

  I'm using the standard ARRL W8JK model that comes with the demo version of 
EZNEC in the ARRL antenna book.  I modify that model to look like what I want 
to build and as long as I don't save it the software performs like the full 
version of EZNEC.  I'm over my head with antenna modeling as soon as I move 
past modeling a dipole but I forge ahead and try and get as much useful 
information as I can from it.

  The question still is, if I get the antenna working properly how much loss 
will I see on the feedline.  Thanks much for your help.

                      EZNEC ARRL ver. 4.0
                2/28/2012     7:53:59 AM
           --------------- SOURCE DATA ---------------
  Frequency = 7.05 MHz
  Source 1      Voltage = 397.6 V. at -89.63 deg.
                Current = 1 A. at 0.0 deg.
                Impedance = 2.597 - J 397.6 ohms
                Power = 2.597 watts
                SWR (50 ohm system) > 100  (75 ohm system) > 100
  Source 2      Voltage = 397.6 V. at 90.37 deg.
                Current = 1 A. at 180.0 deg.
                Impedance = 2.597 - J 397.6 ohms
                Power = 2.597 watts
                SWR (50 ohm system) > 100  (75 ohm system) > 100
                Total applied power = 5.193 watts

  Gary Slagel


  From: Paul Christensen <>
  To: Gary Slagel <>; TowerTalk <> 
  Sent: Monday, February 27, 2012 4:48 PM
  Subject: Re: [TowerTalk] balanced line loss on a mismatched antenna

  > "I think I can tune the antenna with the feedline and get it down to a more 
reasonable level, maybe 10 to 1, at the tuner end. Then the tuner will tune it 
to resonance so the xmttr can put its full load into.
  > Question is, since the swr is 10 to 1 at the radio end of the feedline but 
100 to 1 at the antenna end of the feedline, am I going to see the 3.5 db loss 
from a 100 to 1 swr or the .5 dbi that a 10 to 1 swr would give me."

  Gary,  you can vary the impedance at the input end of the line, but not the 
SWR.  The SWR on the 450-ohm section will remain nearly constant over your 150 
ft run, even with length trimming.

  In EZNEC, what is the Z of the antenna you're modeling at the operating 
frequency in an R+j format?

  Paul, W9AC 


TowerTalk mailing list

<Prev in Thread] Current Thread [Next in Thread>