Wi-Fi point to point helical 2.4GHz antenna you can build!

A while ago I thought I'd try my hand at building a 2.4GHz Helical. It was a sobering experience. The first design I tried to duplicate was Jason Heckers. I had read mixed reviews of his design. There was much conjecture regarding the dielectric constant of PVC and it's impact on tuning. With this in mind I used a thin wall (about 1.5mm) under-sink plastic (not PVC) drain pipe. The outside diameter was 38mm. Coil spacing was 30mm. One end of the pipe had a molded flange for the threaded nut that slips over the pipe to join it to under-sink plumbing. Grinding the outside edge of the flange I was able to make it fit tightly into a 1-1/2" PVC end cap and secure it with cement. The cement did not stick well to the pipe but it did to the cap so it never slipped out. A nice little gap was also formed between the O.D of the pipe and I.D. of the cap, perfect for tinkering with various sizes of the matching strip. The pipe inside diameter was also a perfect for sliding in a piece of 1" ID PVC to test the effects on tuning in real time. The first prototype used the specified dimension for the matching strip of 17mm x 71mm. The helix was 11 turns of 14ga. Kapton insulated copper magnet wire.

Note in this chart that 14db return loss = a minimum target SWR of 1.5:1, 18db = an acceptable SWR of 1.29:1. My return loss test set was the Anritsu S331B, calibrated prior to each set of measurements to cancel any standing waves within the test equipment and cabling. This is especially critical when sweeping a wide range of frequencies in microwave bands. Looking at the skewed data others have posted on such projects I definitely get the impression they did not adhere very closely to mandatory calibration procedures greatly affecting the accuracy of their results.

The actual return loss measurements I got were less than satisfying and, despite claims by others to the contrary, my creation actually resonated higher than the design target, an interesting twist I didn't expect. To lower resonance close to 2.45G the strip dimensions had to be enlarged to 23mm x 95mm. Even then, return loss measurements barely made 18dB - within tolerance, just not extraordinary. Nowhere did I get anywhere near a perfect resonance and certainly not the smooth curves I was looking for indicating residual standing waves elsewhere in the antenna system.

Forging ahead I performed my gain tests. The signal source was a wireless router with a single half wave vertical dipole. Measurements were made against a half wave vertical dipole (2.14dB over isotropic) with a SpeedStream PCMCIA card modified with a SMA connector and Net-Stumbler for dB measurements. The best gain I was able to achieve with the helix pointed at the source was only 4 dB over the best half wave signal (4dBd) or 6dBi. Add back the 3dB cancellation between straight and circular polarization and the actual gain can be estimated at about 9dBi. Not bad but not even close to the 16dBi I was looking for. Interestingly, sliding the 1" PVC pipe down the center as shown at left did not significantly detune the resonance point. What it DID do is swamp the helix reducing the forward gain by another 5 dB. Placing a larger protective PVC pipe snug over the helical had the same negative impact on gain. The introduction of high dielectric material definitely does affect efficiency of the helix requiring an adjustment in diameter and spacing. I can only imagine how poor the gain would be with a full size PVC form. But by how much should it be scaled down to compensate for the velocity factor of the dielectric? x0.67 as postulated by PA0HOO?

You'll see in the photos of my finished design using 1" PVC that I also tried the Hecker "Wonder Stub" on the 1" PVC pipe form. On the smaller form I couldn't get this combination to resonate at all so it was abandoned.

Armed with this information my next prototype was based on the W0OQC design. (Note: The original drawing has orientation errors. My revised diagram to the right and photos below should clarify.) It used the same ground disk as the Hecker design and was wound on 1" PVC per the PA0HOO design, schedule 40. The helix was made from wire coat hangers. The stiff wire was perfect. I started by close winding it on a ratchet socket slightly smaller than the OD of the pipe. It's spring made it just the right size to slip tightly over the pipe. Then I stretched it by hand, made sure it was even and screwed it down over the PVC pipe lining it up with the spacing marks. One hanger was enough to make only 8 turns. To make 11 turns I just added another section of 3 more turns and soldered them end to end on the form. The matching section is a half wave strip tapped roughly 40% from ground. Mine was crudely cut with tin snips from a piece of galvanized steel roof flashing I had laying around. The W0OQC measurements are very good but fine tuning with proper test equipment is still necessary. The PVC end cap holding the pipe is mounted to the ground disk with a single screw through the center from behind allowing the assembly to rotate on axis inside the matching section for fine tuning adjustment. Another huge benefit of this style matching network (as pointed out to me by W0OQC) is that the feedpoint connector is held to DC ground rather than floating making it virtually impervious to static discharge effects. Forget nearby lightning strikes. Even a steady breeze on a dry day blowing across the PVC could be enough to accumulate a static charge sufficient to fry sensitive preamplifiers. Not here.

This design resonated exactly around the target frequency of 2.45GHz. Return loss was fine tuned by adjusting the helix feedpoint position and height of the matching strip above the ground disk. Return loss was greater than 30dB from 2.348-2.488GHz! For those who must think in terms of SWR that's 1.05:1 or better. (Note the difference in scale from the trace above.) This is as perfect a match as you will ever see in any antenna system at any frequency. The 18dB BW was .273GHz (2.273-2.546GHz)!!! When I saw these results I couldn't wait to run the gain and off axis tests.

Forward gain improvement was nothing less than spectacular! About 8dBi improvement, roughly 17dBi overall gain. Yes, it really was that good. Clearly, the efficiency and tuning of the helical is heavily dependent upon the form material. Open air = close to theoretical dimensions but difficult to build. High dielectric form = lower velocity factor, smaller than theoretical dimensions and much easier to build. A good matching network certainly helps too. The design of this ground disk and matching network around a PVC end cap also facilitates the easy swapping of prewound coil forms should I decide to experiment further. It figures - I spent extra time making provision for failure and wound up not needing it.

Two things are for certain though. I will definitely use this same matching network for all future versions of the helical and I would never recommend attempting to build any GHz helical without the aid of a network analyzer. Reproducing any design without one will almost certainly be an exercise in futility. Many thanks to Marcus Communications for loaning me an Anritsu S331B for this project.

Mark Brasche
Safe-PC.net

   
Additional photos of the finished project. Now what on earth
could he be doing with that mounted to the roof of his car? ;-)

Call Safe-PC toll free today at 877-7SAFEPC
(877-772-3372) or contact us by email.

Safe-PC is more than just our name. It's our mission.

Terms and Conditions