A key component of a Doppler
RDF is the antenna array.
The picture below shows a big, professional array, used for finding
bearings of air traffic.
Such a big array will yield a very high accuracy, however overall
accuracy will be determined by the environment.
Maximum accuracy will only be achieved after excessive measurements and
calibrations, compensating for all possible distortions.
Although a radio amateur would steal the show in the eyes of his fellow
amateurs with an impressive array like this on his roof,
most of us will have to follow a much more modest approach...
A 4-element array will do the job in most amateur situations.
Therefore, we'll stick to the 4-element, and describe optimized
versions for base (fixed) use and mobile (car) use.
The fixed homebase array consists of 4 halve wave vertical dipoles in a
The spacing between two adjacent elements will be around 0.23
wavelength in my current array.
This is a good compromise giving a fair amount of doppler signal in
most FM receivers.
Wider spacings will yield a louder doppler signal but this may go
beyond the maximum deviation that some narrow band FM receivers can
Probably the best compromise spacing is around 0,25 wavelength.
I've been modelling my array using MMANA, and doing that I found a few
things that obviously were missed in other amateur designs.
For best performance, only the active element may be accepting current
from the incoming signal.
All other elements need to be virtually non-existent to prevent
parasitic behaviour and therefore severe distortion of the pattern.
This can not be achieved by the simple top leg switching of existing
The bottom leg of the element connected to the attached coax shield
will be picking up current heavily, as MMANA showed.
So, a third PIN diode proved to be necessary:
The below picture shows the current distribution of this new approach:
Note that in this simulation there are ferrite beads modelled half way
of the coax length, which turned out to be not necessary at all.
On the contrary: leaving them out will give some further improvement.
However, it is obvious that the switched off elements are really dead
now. A big improvement compared to the simple switching method I
This is also clear if we look at the pattern:
Only 0,33 dB of directivity is still remaining, which is superb.
Also, a simpler version was modelled, with a 1k terminating resistor
instead of the 2uH inductor between the dipole elements:
It has even less front to back ratio but there's a tiny reduction of
approx. 0,7dB to the sides.
However, still doing a very good job, and it's suitable for a wide
I also modelled the array versions of pretty elaborate designs, using
preamps on each element:
The preamps outputs are switched on and off to the output coax, so the
preamps are always loading the non-active elements.
MMANA showed massive interaction between the elements and directivity
was over 6dB in two directions.
On other frequencies it gets far worse still!
One more case proving that a higher component count does not
necessarily mean better performance...
Furthermore, the coax and control cable should be taped flat to the
metal tubing to avoid them picking up HF.
Additionally I have a few clamp-on ferrites in the first meter running
down from the central switch box.
One every 40cm would work fine:
Good Array Dimensions for a 145MHz version, suitable
from 118MHz to 172MHz:
Element length: 1m, (2x 50cm)
Element distance to centre: 32cm.
Element to element spacing: 45cm.
Make sure that all 4 antennas are absolutely identical, and use equal
lengths of coax to feed them.
I recently re-designed the construction of the array, to make it more
sturdy and to improve the looks:
Above the new housing for the summer, out of 125mm PVC tubing. The arms
are made out of 32mm PVC tubing, one running clean through,
the other two arms are cut in a way that they fit the first arm, so
they can be glued firmly together using PVC glue.
A metal top plate is glued on and bolted down. On the other side, the
bolt runs through the metal L-profile as well.
In this way a simple but very solid frame can be constructed. The four
arms still need holes to be drilled for the coax pieces to the
dipoles at the arm ends.
Here's the same PCB now
looking at the copper side, just to show the way the whips are soldered
The two 50cm whips are running into the tube through tight fit holes
and soldered to the small PCB.
The whips are made from 2 mm steel, and covered in heat shrink tube,
so they are well protected against the the environment.
Of course, the array may be scaled up or down for other frequency bands.
With the above technology and dimensions, good results will be achieved
not only on the designing frequency, but also over a considerable
frequency span up to 20% above and 20% below the designing
The electronics of the PA8W Doppler RDF will support your experiments
from around 30MHz up to 450MHz without alterations.
A 2m array as well as a 70cm array would be interesting of course, but
also a 6m or 4m version could be very rewarding.
I'll just leave it up to you...