Improved Helical Antenna Design for 802.11b WLAN
October the 17th 2002
N.B: (Click here to jump to the improved PVC helix design)
The weekend of October
the 17th 2002 we tried to set up an 802.11b
WLAN radio link over just 200m.
The radio path
'Direct sight', seen from the roof. See pictures below.
As we had to bridge about 200 m, an external antenna was a must.
Making your own path calculation
Click here to calculate the distance you can bridge with a given antenna.
The connections for the external antennas
These pictures show how the external antenna connections were made. We used cheap surplus cables with SMA connectors. (about € 1 each).
Our experience: PTFE cables don't melt while soldering.
Detailed examination of the antenna design
Looking into detail, there IS a difference between de ordinary Helix antenna and the PVC-one: The classical helix is al 'on air' spiral, our spiral is 'on PVC'. Kraus' formulas, calculate air spirals. As our spiral is on PVC, it COULD be that the traveling velocity of the radio waves inside the antenna is being influenced because of the high εr of PVC. (similar to the cause of shortening factor of coax cables). In that case Klaus' formulas will give erroneous results.
The main question is, could our PVC plumber tube actually cause the working frequency being decreased? If the answer is 'yes', it must be possible to improve the design.
Note: for our antenna, the factor between design and reality is: 1650/2430 = 0.67
Study books say that the traveling velocity is reversed proportional with the square root of the dielectric constant.
some common values for εr:
appears that PVC has a fairly high dielectric constant.
'The proof of the pudding is in the eating'
So, we decided to design, build and test a similar downscaled Helix, designed for a 33% higher frequency.
2430 MHz is about 67% of 3620 MHz. So 3620 MHz will be the design frequency of our new downscaled test antenna
We will keep the reflector
according the original design of Remco, 14 cm square.
We converted the old
design to the new design with smaller dimensions. See and click the picture
In our living room, we tested again the 'ether transfer characteristic' at the distance of 5 meters.
The difference between the two antenna designs is obvious, despite of the upper limit of 2450 MHz of our equipment. The upper line is the plain directly connected coaxial antenna cable. The lower line is the antenna transfer characteristic, using the same coax, now connected to the antennas. The gain is multiplied by +30 dB. Note the markers' position and see the difference. (Both graph differ 20 dB because I forgot to make the scale of both graphs identical)
At the moment we do not have access to the right SWR measuring equipment, so proper SWR information is not available yet. If somebody took some SWR or reflected power measurements, please let me know.
Update: (June 2008): Mark Brasse describes the construction of a similar PVC - helical antenna, based on the design of Jason and using the ‘shortened’ design as per our experiences. Instead of Jason’s ‘wonder sub’ he uses a small 1/2 λ strip which can be tuned. The final results are amazing! See: ( http://www.safe-pc.net/helical.html ).
The radio link, that appeared not being possible with the original design, is working very well now, using the new downscaled antenna.
A final word: In literature, a ''dielectric loaded antenna' is well known phenomena. Actually, our helical a sort of 'partly dielectric loaded helix'. (Literature example)
Once we have finished our final antenna we will put some pictures on this web page. (We're making a neat 22-turns helix now)
The final 22 -turns helix antenna (added November 30, 2002)
Below you'll find the description of a
22-turns helix antenna according to the downscaled design.
Use a 32 mm end cap. (No 40 to 32 mm converter). We used just one M8-bolt and a large washer. To prevent for electrical contact between washer and vane, we isolated the vane with two layers of PTFE tape. See picture. The dimensions of the helix spiral are: wound on a 32 mm op PVC tube, windings are spaced 2,88 cm. We used ordinary 'home installation' PVC isolated copper wire of 2,5 mm2. You can use the print out of helixcalc, but inputs have to as described above (the 'false' 3520 MHz, 3,4 cm diameter).
The vane's dimensions are
identical to the original design of Jason. (71 mm) New:
Wim Pa0wkm made the vane slightly longer This should give much better results on
SWR. No further details, however.
We compared the antenna
transfer characteristics of the 11-turns and the 22-turns helical antenna
(two antenna pairs). The spectrogram shows a gain difference of about 8 dB. That
is for 2 antennas, so for one antenna the gain difference is 4 dB. We found that
the thin white PTFE cable that we used with the 11-turns antenna has a
loss of almost 1 dB. This implies that the real gain difference between the
11-turns and the 22-turns antennas about 3 dB, and that is exactly according to
theory. Our 'dielectric load' downscaling approach appeared to be reproducible.
Our final 22-turns helical has a N-connector mounted at the reflector.
Some points of interest
Using the antenna for a free sight
'point to point' connection:
Overview of frequency of Wlan channels.
How we connected the WLAN equipment to the antennas.
As our radio path was very short, there was no need to optimize for lowest RF losses at relative high cost.
We got from the surplus of a large project a length of IEEE 802.3 network cable of which we measured the RF losses. The losses were about 0,3 dB per meter at 2400 MHz. As a reference, good RF cable (e.g. aircomplus) is about 0,2 dB per meter. At a cable length of 5 meters, (that was all we needed), the extra losses were 0,5 dB. As we had to bridge just 200 meters, and our antenna was having a relative high gain, we decided to use this cheap (to us) network cable.
At one side we
used a SMA and N pigtail, the other side had a solid SMA to N converter. As a
general rule, prevent using pigtails, as they are expensive and generally
introduce relatively high losses. (As an example, our 30 cm pigtail had a loss of 1,5
dB, that is equivalent to 5 to 7 meter good quality RF cable). Use 'solid' converters
instead, if available.
If any remarks, you could contact me here.
73, pa0hoo (pa0hoo(deletethis)@qsl.net)
Bijgewerkt / Updated: 2005-01-23