Call Search

New to Ham Radio?
My Profile

Friends Remembered
Survey Question

DX Cluster Spots

Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement

donate to eham
   Home   Help Search  
Pages: [1]   Go Down
Author Topic: Phasing multiple dipoles  (Read 3686 times)

Posts: 585


« on: March 06, 2013, 09:54:37 PM »

Its long been an idea of mine to recreate (albeit at a higher frequency so its smaller!) the antenna system of a German WW2 Knickebein station. These used phased dipoles (actually dipole plus reflector) to create a very narrow beam. For anyone not familiar with this system, it was a blind bombing system, two sets of phased antenna arrays were used, side by side, with the RF switched between them, the result being dots on one side of the beam, dashes to the other, and a steady tone where the two beams met, this tone area being incredibly narrow (enough so that it was within a mile wide over the city of Coventry in England when transmitted from Holland)

However, I dont yet fully understand how to phase dipoles together to get this effect. I can work it out for a pair, but these antennas used  lots of dipoles.

Can anyone direct me to a suitable practical explanation of how to phase a number of dipoles together in this way? I understand the principles for phasing, say, four vertical dipoles in a vertical array (one above the other) but this system used a horizontal array of vertical dipoles. Does the same phasing method apply in this case?

I can find lots of information on colinear arrays of vertical dipoles in a vertical array, but no info for a hroizontal array

I'll trial the system on 70cm first, as this is nice and small. Ultimately i'd like to recreate the system on a frequency close to the original, which would mean going quite low as they were on about 25MHz if i remember correctly
« Last Edit: March 06, 2013, 10:24:59 PM by G7MRV » Logged


Posts: 412

« Reply #1 on: March 07, 2013, 01:35:29 AM »

Its called a vertical broadside array I do believe. Vertical dipoles spaced horizontally approx half wave apart and connected by crossed over twin feeder. Fed in the centre.

I have emailed you a diagram from an old RSGB handbook. Just add more dipoles left and right with the twin feed crossed between them.

For fascinating read a about Knickebein, its detection and other WW" radio devices get "Most Secret War" by RV Jones from the library
Best of luck.


Posts: 4310


« Reply #2 on: March 07, 2013, 05:40:57 AM »

Its called a vertical broadside array I do believe.

A quick 20m EZNEC simulation says that three collinear 1/2WL dipoles spaced 1/2WL apart and fed in phase will have a gain of 13 dBi and a beamwidth of 20 degrees. Total antenna system length is 165 feet for 14.2 MHz.

73, Cecil,

Posts: 17476

« Reply #3 on: March 07, 2013, 01:16:09 PM »

You don't actually phase the antennas to get that effect.  Well,
not directly in the manner we are used to thinking of.

There are two separate components involved in the system (which
is - or was - similar to the standard Instrument Landing System,
and, in fact, the system was originally discovered by the British only
because the ILS receiver from a downed aircraft had much greater
sensitivity than would have been required for a normal landing.)

First, the detector method can be explained using just a pair of
vertical dipoles arranged as a 2-element yagi:  if you use a
switch with the proper line lengths, each dipole plus the length
of feedline to the switch will operate as a reflector when the
other is driven.  So as you switch back and forth between the
two the pattern fires in opposite directions.

Now, imagine the pattern on receive.  In directions where the
left-pointing beam is dominant, you would hear a stronger signal
when it is selected.  If the signal comes from the other direction
it will be stronger when the right-pointing beam is switched in.
If you draw the relative patterns of the two antenna, you find
that they are equal at right angles (broadside) to the antenna.
Switch between them fast enough and you can then use a phase
detector (driven by the same switching pulse) to read out the relative
signal strength on a zero-center meter:  when the meter needle
is centered, the source is at a right angle to the antenna.  This
approach is used in the L-Per DF receivers from L-Tronics, and gives
a very good bearing accuracy with a simple antenna, but the
same principle applies when the transmitter switches the antennas
and the receiver listens to the signal.

Now in typical aircraft usage the antennas were not switched with
constant pulse lengths, but with a pattern of 2 short + 2 long
intervals.  Depending on which side of the pattern was stronger, the
operator would hear a Morse "A" or "N", which blurred into a steady
tone when the two were the same strength.  Instruments were
developed to provide a meter readout to replace the aural method,
but the principle remains the same.

Such a system with a single pair of antennas is probably good to
2 degrees or so, depending on the accuracy with which the antenna
is built, and the sensitivity of the detector.

To improve the angular resolution, you need to use a sharper antenna
beam, so that the difference in signal per degree of angular variation
is greater.  The ILS systems that I drive by use an array of broadside
antennas, perhaps 8 or 16 separate LP beams.  You can either have
two such arrays that are aimed in slightly different directions and
switch between them, or switch the relative phase of the various
elements to electrically steer the beams in different directions.  For
example, with 8 beams, if you feed the one on the right at 0 degrees
and each successive one with 5 degrees more than the one before
it, the beam will shift to the right.  If you arranged for each such
beam to be able to switch between two phasing line lengths, you
could make the beam switch between, say, +/- 10 degrees on either
side of center.  (Because the dB per angular degree increases as you
get away from the center of the beam, there is a trade-off between
aiming the beams wider apart for a more distinct corss-over point and
aiming them closer to each other so the signal is stronger at that point.)
What is really happening here is that you switch the phasing of
the antennas to change the beam direction, rather than using two
separate beams (which also works.)

In this case, the aircraft could fly outwards on one beam (using the
same system to know if they had deviated to either side) and then
a separate receiver showed they had encountered the second
beam.  When the signals in the two lobes of the second beam were
the same strength, they were over their target.

This system proved susceptible to British jamming once it was understood,
and it was later replaced by one called Wotan, an old German god.
When intelligence first reported the name of the new project, an astute
officer observed that Wotan had only one eye, so must be a single beam
system, and successful Countermeasures were developed based on just
that one piece of information.  But that is more difficult to explain...


Posts: 412

« Reply #4 on: March 08, 2013, 05:04:39 AM »

There are some artists impressions of the antenna on the net, and in my book are some pictures with "aerial array tentatively drawn in". Here's a big Knickebein

EXACTLY how the dipoles received the RF is not at all clear, indeed its no wonder you asked, G7MRV! Some pictures looked like there were centre fed dipoles and reflectors. (Vertical broadside array) However most pictures look as if both front and rear elements are being connected to the feeder, like a W8JK array. There are knickbein antennas 4 to 8 elements wide plus some antennas with the same array vertically above.

A very interesting phasing problem. The frequencies were typically 30.5 to 33.5 MHz. Please let us know how you get on. Is it possible to test a theory using eznec?

Posts: 17476

« Reply #5 on: March 08, 2013, 09:07:53 AM »

The exact feed system isn't important to how the system actually works:
any narrow beamwidth antenna can be used.  Or, more accurately, TWO
such antennas with beams pointing in slightly different directions, though
it is possible to create those two beams by changing the phasing on a
single set of elements.

I modeled an array of 16 dipoles with reflectors, which require a length of
about 15m (50') on 2m.  The half power beamwidth of the resulting antenna
was 6.4 degrees (and a gain of 18.73dBi.)

This in itself isn't good enough for the required accuracy.  Even at 1dB down,
the beamwidth is still about +/- 2 degrees.  But at 5 degrees from the peak
of the beam the signal is down about 8dB, and drops at a rate of 3.3dB /
degree of rotation.

If you use two such beams aimed 10 degrees apart, then the differential
sensitivity is about 6.6dB / degree, and a detector with a 1dB sensitivity
could resolve equal-signal angle to less than 0.2 degrees.

You can get close to 0.1 degree sensitivity with slightly wider beam
aiming, but the desired point is then 13dB down from the main beam,
which might make it too weak to find reliably at a long distance.  The
other problem (at least with my model, which used 40" spacing between
bays) was a sidelobe at +/- 10 degrees, which might be a problem.)

If we take the distance from Holland to Coventry as about 400km,
a target accuracy of 1km requires a bearing resolution of
arctan( 1 / 400 ) = 0.14 degrees, with is right in our calculated
range of 0.1 to 0.2 degrees of bearing accuracy with 10 degrees
between beams and a detector sensitivity of 1dB.

With a more sensitive detector you could use a single array of 8
elements for each beam, putting one on each side of the center
and switching between them.  Then the whole framework could be
rotated to the desired angle.

I used a simple 2-element yagi that I had on hand in EZNEC to
start with.  Looking carefully at the photo it appears that each
dipole is actually two half waves in phase (because the length of
each half is about equal to the bay spacing.)  I'd guess that they
were fed using 1/2 wave phasing lines in the normal manner.  The
rear elements could be two insulated reflector wires, or (again, more
typical of the period) a phasing line to the rear element.  The old
antenna books up into the 1970s used to have good descriptions
of broadside, end-fire and colinear phasing and various ways to
accomplish them:  in this case, each pair of elements would be an
end-fire array, with all the arrays connected in broadside.  That
makes it likely that there were two separate antennas, rather than
one with switched phasing, to generate the two beams.

Posts: 412

« Reply #6 on: March 09, 2013, 12:09:13 PM »

G7MRV, I think I have worked out how the Knickebein dipoles were interconnected and will email you a sketch if you wish. (Or to others interested who might like to try it on eznec.)  I did email you via your QRZ address. It looks an interesting and practical project to actually build a real working model.

What you were after was all about connecting dipoles, not how Knickebein was used. I noticed error in Wikipedia and published pictures. Simply, the array produced two diverging blunt beams, which had a narrow equi-signal between them.


Posts: 585


« Reply #7 on: March 11, 2013, 10:00:12 AM »


yes, please i would be interested in what you have. Sorry its taken me so long to get back on but work has taken me over for the past week or so!

Yes the idea is to create a replica of the specific array design as used by Knickebein. Incidentally, I have RV Jones 'Most Secret War' and it is from this book that my interest stems


Posts: 412

« Reply #8 on: March 12, 2013, 03:15:16 AM »

Martin, I have emailed a pdf of my Knickebein drawings using your QRZ email address. If you dont get it reply to me via my QRZ email. Should anyone else be interested do the same. It would be an interesting eznec project.
Pages: [1]   Go Up
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!