The Science of Successful antenna design



As promised, James, Brad and Hughes, and anyone else interested, some info on turnstiles and the methods of antenna matching.

James made the following statement:

I came across this link  which explains how to build what they call a turn stile antenna.  Could anyone try and explain why this antenna would work as I had really poor performance with it.  I ended up using a simple cross dipole on the RX side with much better results.

I gave the start of an answer in a long reply which can be found in his post : 433 UHF LRS Antenna "Turn Stile"

Some further questions where asked regarding the methods of match the antenna, and the antenna tuning, so I will try to elaborate. 

First some fundamentals.

When I speak of antenna radiation it also includes the reverse, that is, the antenna 'collecting' radiated energy from the ether. 

Any piece of wire will radiate energy when connected to a source of RF ( Your transmitter or receiver). How WELL it radiates that applied energy  is dependent on only two factors-

  - That the antenna is resonant at the same frequency as the radio signal applied and

  - That the feedpoint of the antenna is matched to the impedance of the attached transmitter energy source.


Maximum energy tranfer between source and load ( the antenna) occurs ONLY when the load and source impedance's are equal.

How well the antenna radiates that energy in any or all directions, if the above conditions are met, is then only dependent on the antenna design, shape, or style. For example, a simple vertical 1/4 wave radiator will radiate a doughnut shaped pattern, in all direction of the compass, with low energy upwards and downwards, in the direction of the antenna element.

A yagi type antenna, on the other hand, ( such as your vhf or uhf TV antenna) focuses the energy in a single direction, as would a torch. Antenna do not have gain. They focus the energy to a greater or lesser extent, in a direction of design, but do this by robbing energy from other radiating directions - Your torch puts more light out the lens end, with nothing out the rear end. If you remove the reflector from behind the torch bulb, the light is radiated in all directions, omnidirectionally, but is much weaker at any distant point than the focused beam.


Antenna radiation is polarized; that is to say, the radiated electromagnetic wave has a net polarization plane. This is usually either linear or circular. Elliptical polarization is also found, but that is merely a mix of the two former types.

A 1/4wave vertical antenna will radiate linearly, with vertical polarization. Placed on its side it will radiate horizontal polarization.  A Helical antenna ( looks like a coil of wire wound in a screw fashion) wound clockwise when viewed from behind will radiate clockwise circular polarization, and vise versa.

For reception of maximum energy, the two antenna must be identically polarized. There is a massive loss of signal ( easily some 30dB, although the theoretical loss is infinite) if one attempts to receive a horizontally polarized signal with a vertically polarized antenna. Similarly, there are massive losses if trying to receive a circularly polarized signal with an antenna of the opposite circular sense.

The odd man out is that there is only a 3db loss between an antenna that is circularly polarized and one that is linear.

Why would one use circular polarization?

If the two antenna in question could not be made to maintain similar attitudes, such as one in a pitching , rolling aircraft, then there would be unacceptable signal losses as the aircraft banks and pitches. So you could use a vertical on the aircraft, and a helical or turnstile, or similar, on the ground segment. This way you would only ever experience a 3db maximum loss. ( all assuming good line of sight view). Or you could gain back the 3db loss by fitting a similar circularly polarized antenna on the aircraft, giving the best of both worlds. But you actually gain more than that with circular polarization at both ends.

Assume first that the two antenna are simple vertical monopoles, radiating vertically polarized signals. When you are flying, at the flying club, etc, you are probably near some metal structures, the 'hanger' , cars and other vehicles, etc. All these structures reflect the same energy you are trying to receive. In addition, when the aircraft is low and far, the RF transmitted by the A/C antenna follows two paths to your receiver - one directly, and one via a reflection from the ground, somewhat midway between you and the A/C.  What happens to the reflected wave is that the polarization is changed in unpredictable ways. Your receiver ( and antenna) does not know or care where the received energy comes from, so it receives this reflected energy as well. These multitude of received waves add constructively and destructively with the main received wave, causing large, short duration, signal drop-outs - a sort of 'flutter' in the signal.

If both antenna are circularly polarized, however, the picture is quite different. When the circular polarized waveform is reflected , it REVERSES its polarization. When this reversed polarized signal arrives at your receiving antenna it is largely rejected and hugely attenuated, so interfering minimally with the main received signal.



On to Issues of resonance and matching.

To repeat a little in my post to James:

Most simple linear antenna are either of the monopole or dipole form. A single monopole ( 1/4 wave vertical for example) or a single dipole will only radiate linear polarization.

Any antenna is only resonant when it is exactly the correct length AT the frequency of operation.( this does not apply to the class of broadband antenna, such as helical antenna, etc. The helical will easily cover an ocatve with good performance).

At resonance the antenna will exhibit its characteristic feedpoint impedance. Feedpoint impedance is expressed with two terms, the pure resistive part, and the reactive ( j operator) part.

Most transmitters and receivers terminal impedance are made to be 50ohms resistive, or very close to that. So it stands to reason the antenna must also be 50ohm resistive to have max energy transfer.  However, none of the antenna are that obliging, so we have to do some feedpoint matching to meet the criteria.

A 1/4wave vertical monopole over a ground plane has a resistive feed point of around 75ohms. A half wave dipole is around 72ohms. As with resistors, placing two dipole in parallel as in the IBcrazy turnstile, will result in a feedpoint impedance of 35ohms.

A 75ohm feed connected to a 50ohm coax and transmitter will exhibit a 1.5:1 SWR ( the ratio of power going out to power reflected). A 1.5:1 SWR means that approx 3% of your transmitter power is not being radiated. ( 30milliwatts for a 1watt transmitter). That is not so bad, and we can live with an SWR of 1.5:1 in most cases.

The turnstile antenna is a pair of crossed dipoles, fed 90deg out of phase with each other, thereby generating circular polarisation. You CANNOT simply connect the dipole in parallel at the coax feedpoint though. Apart from the halving of impedance ( which we decided we can live with) the radiation pattern and polarization of the antenna will be totally destroyed by unwanted radiation from the coax cable. The RF energy, at the dipole connection point, 'leaks' out and currents then flow down the outer shield of the coax. As mentioned previously, any piece of wire will radiate RF energy, and so the coax radiates this energy, and the radiation again adds constructively and destructively with the main antenna radiation, causes complete distortion and signal nulls in the pattern. This radiation from the coax MUST be prevented.

This is done by means of a Balun transformer. - which is is an acronym for 'Balanced to Unbalanced transformer'.

A dipole is a balanced device - it is electrical equal along each element, outwards from the feedpoint. It therefore requires that the feedpoint be fed in a balanced fashion. Coax cable is an an balanced feeder - the shield is at ground potential, while the inner core carries the energy. This effectively ( oversimplifying a little) connects the one dipole half to the 'live' core, and the other half to 'ground' unbalancing the dipole. This causes currents to flow on the coax outer shield, and distortion of the dipole radiation pattern.

Baluns can be constructed from coax cable, but the accuracy required in coax cable length ( they are normally length multiples of 1/4 wavelength) is very critical, especially in the GHz range - 0.5mm can have a great effect.

The turnstile is not new - it is some 50 to 60 years old, and is well researched and published. Up to the VHF and lower UHF region , the coax balun, with embedded impedance match transmission line transformer, is used, along these lines:


For the higher microwave frequencies, a plumbing type version is more appropriate. This is called the spilt tube or split sheath balun, and looks like this when used as a feed for a pair of crossed dipoles.



The balun and feed match consists of an outer and an inner tube. The ration of diameters D/d is chosen to give the desired impedance:

D/d = 1.86 for 75ohms, and 1.5 for 50 ohms.

Typically the outer tube would be around 8mm for use at 2.4GHz.

In order to obtain circular polarisation, I mentioned that the two dipole have to be fed 90deg apart ( phase quadrature).

This can be done as in the coax balun version above ( inserting an extra 1/4wave length of coax in the leg to one dipole gives an extra electrical wavelegnth of 90 degerees). 

Or, this can be achieved by slightly lengthening the one element ( becomes more inductive) and shortening the other( becomes more capacitive) - this also introduces the required phase difference between the elements.

This can be seen in the images above - the one element is typically around 0.21 wavelength per half, while the other is around 0.25 wavelength. One short and one long element penetrate the outer tube and are connected to the inner tube, while the opposite pair of elements are connected only to the outer tube. The outer tube is split or slotted ( 0.5mm width slot). The slot is approx 0.23 wavelength long.



The relationship in length between the two dipoles is critical, typically this would be measured on a network analyser and the feed impedance of each element set to say R+j45 ohms ( longer dipole) and the other to R-j45 ohms. This will give the correct phase relationship between elements. A half mm variation can have a great effect, turning a good antenna into a mediocre one..

The last image above shows a teflon tube - this is inserted in the tube from below, and fits snugly inside the outer tube, and over the inner tube. This is then slid up and down to adjust the 'R' part of R+-jX, till the match is a good 50ohms. This does not affect the antenna radiation pattern or characteristics. Obtaining a 50ohm impedance match can be done by trimming the element lengths as well, at the same time destroying the antenna radiation pattern and circularity.

And that is why it is not so simple to do at home, and why the 'Hobby King'  et al variants sold everywhere are mostly trash..You will probably achieve a few km range with those- remember, any old piece of wire will radiate -  I easily  achieve 15km with 500milliwats at 2.4GHz using two split sheath balun , properly matched and trimmed, crossed dipoles..

For those interested:

References are - RSGB VHF/UHF Manual - page 8.45

Modern Antenna Design - Page 255

Here are some images of my split sheath balun crossed dipoles..




The Nampilot.

E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • Fascinating (and frustrating) stuff.  I have a related question and it's more about real world than theoretical.  Most folks are doing FPV on a UAV.  What about a rover?  I've played around with Friis equation solvers that tell me I should be able to get several miles of range on a 125mW 2.4 GHz link.  I've also read that Fresnel effect will significantly reduce signal when the antennas are close to the ground even though you may have perfect line-of-sight.  Supposedly, the signal isn't laser-beam straight but rather is more like a loose string between the two antennas.  If the sag intersects the ground, you lose signal.  Care to comment?

  • @iskess'

    Have a look at my comments to Justin on TX/RX bandwidths and mutual interference. By default, the low cost of the modules used in this typical application means that many compromises have been reached in their design and implementation. The main compromises are always in the front end selectivity and the ability of the front end RF preamplifier and the first mixer to withstand out of band and/or strong signals. So , yes, it is always better to try and keep these modules as far from each other as possible. But any co-location issues on the aircraft are fairly easy to detect and test for - simply tape the antenna in place and test the system with all systems active. You do not need to move the aircraft far from the ground station, etc, since the problems will manifest on the aircraft only.

    The Inverted V, when employed as an inverted V, has an 'almost' omnidirectional pattern, which eliminates the signal nulls of the end of a conventional dipole. However, the inverted V also has less gain than a dipole, maybe half, but agonising over 1 or 2dB is a waste of energy - it will make no real difference at all.

    Not that in my diagram the inverted V is in fact a side-ways V, but the pattern will be relatively unaffected. It will work just fine. The compromise is required in order to get it to fit in your available space.

    Your question 2 is covered - the advantage is a more omnidirectional radiation pattern - also it is more easily matched.

    Here is a design for a 5/8 base loaded antenna at 915MHz. I used exactly this on our Hornbill aircraft ( if interested, search for Hornbill in my earlier blogs..) with a 915MHz, 1watt Digi modem, and on ground I used a 10 element Yagi - we got 12km range easily, while only achieving 4km with a 1/4 vertical on the plane, and 9km with a vertical diople.

    NOTE: there a number of 'designs' for the 5/8 vetical. The highest gain design is the one I show below, and if done and matched perfectly, will give around 4dBD ( that is 4db gain greater than a dipole!) . The design concepts are as follows:

    A 1/4 wave antenna radiates omnidirectionaly with a contiguous radiation pattern. Its gain is quite low, and its capture area, a function of its size, is low.  A 1/2 wave vertical has much higher gain, contiguous radiation pattern, a greater capture area, but its feedpoint is at a high voltage point, so has a very high impedance and is difficult to match. A 3/4 wave antenna also has a high gain, even greater capture area, and the feedpoint is a much lower impedance, and easy to match. However, the 3/4 wave antenna starts to exhibit a break up of the radiation pattern into a number of main lobes, with in-between nulls. We can now combine the trait of increasing gain and ease of matching for longer antenna, by making the main element 5/8 wavelength long, and fitting a coil at the bottom that is an electrical 1/8 wavelength long. This gives a 3/4 wave electrical length antenna, but only 5/8 log physically - the coil does not radiate appreciably. So we get a very good radiated pattern, high gain, and an easy match.

    Here are also some picture of the antenna as I fitted and used it on our Hornbill aircraft.

    Note that it was fitted upside down, not ideal as the shape of the radiated signal is not optimum this way.





  • Great Blog Post Joe

  • Your "crude" drawing is very nice.

    The Vertical Stab is 7 inches from the top of the Horizontal Stab, and 10 inches to the bottom of the tail.

    I like the your layout idea, but there are some complications I haven't shared with you. This plane has a removable horizontal stabilizer, which would make that 5/8 vertical more complicated. One solution is to embed the telemetry into the horizontal stab so I only have to detach a 4 pin connector to remove the stab. However this won't allow me to have a long vertical feed.

    The other issue with the 5/8 is that there is a big foam hump to fill the gap under the rudder/vert stab. This will make for a lumpy ground plane. Maybe I could slice off the hump, apply the copper tape, and glue the hump back on over the copper.

    I'm starting to think that this isn't the right plane for this layout. I'm still very intrigued by the 5/8 vertical and think the perfect installation would be a Cruciform style tail (non-removable).

    I also still have concerns about the Vtx since it probably has a pretty wide analog bandwidth that spews noise all over the place. FPV pilots always recommend keeping the VTx as far as possible from any receivers no matter the frequency. Maybe this is naivety, but it seems I have seen too many reports of people having issues when they have inadequate separation.

    So my guess is that the best antennas to have in close proximity to each other on the tail is the 433MHz RC control and the 915MHz Telemetry. They are both digital signals with frequency hopping, which should help them avoid each other, and the Tx power is only 100mW on the telemetry versus 800mW on the VTx.

    Based on your drawing, it would seem you would recommend a Vee and a dipole.

    Here are some questions:

    1) Did you choose the Inverted Vee simply because of the shape of the horizontal stabilizer, or does it also help to lessen the interaction with the nearby dipole? As you saw in one of my previous photos, one of my installations already has a 433MHz dipole installed. If I were to add a 900Mhz antenna here, could I just do another dipole on the other side of the vertical stab, or should I make a Vee like in your drawing? Future new builds can use a 433 Vee like your drawing.


    2) Most FPV pilots use the Inverted Vee for the VTx over a simple dipole (as shown in my earlier photo). Do you see any advantage to the Vee?


    3) I would love to see a tutorial on how to build a 5/8 antenna. I don't know how to make that coil, or what size the ground plane needs to be. Until I figure out how to install it on an airplane, it should still make an excellent ground station Omni for my diversity system.


    I don't have any equipment to measure SWR. I've seen some tutorials on how to DIY, and I think its time a take another look into that.

  • @Brad,

    Keen to see how you tests go - always difficult to get qualitative results in this though - each flight is different!

    Good luck.


  • @ Justin,

    At the risk of copping out on this one...

    This is almost impossible to predict Justin. As I harped on before, the problem is that the suppliers just to not provide sensible or useful specifications of the TX and RX modules to be able to predict or model this.

    Solid commercial Radio TX and RX specs normally include the receiver input filter characteristics, bandwidth, etc.

    There are so called two-tone transmission parameters, where in effect two RF carriers are generated,  close to each other, and injected into the receiver input. One of the frequencies is on the required channel, the other just adjacent, by 'an' amount. This is an 'interfering' signal. The main signal ( data, telemetry channel, etc) is then monitored and the signal quality measured. The interfering signal is then increased in strength, till the quality of the desired signal deteriorates by a given amount. This is used as a standard of measure for adjacent channel interference. Receivers with poor input filtering and poor front end mixers will exhibit interference at lower levels of interfering signals, and also interference from signal spaced much further away in frequency. And that's the problem..

    None of the DIY domain telemetry, RCS and video RX equipment and modules manufacturers provide any useful and relevant info on this. The only way to to try it, I fear.

    Some basics do apply - do not co-locate antenna that are directly harmonically related. Try to have their respective polarization opposing if possible. If co-located, try ( hardly ever possible!) to keep a separation distance equal to at least a 1/4wave plus 20%  at the highest antenna frequency.

    Co-locating telemetry and video is normally ok - the telemetry has a 'locking' protocol which in itself will reject the video signal somewhat. However, the video receivers generally have VERY poor front end selectivity since they have to cover many channels, and the video signal occupies some 15MHz bandwidth. These receivers do not tune the front end filters so are very wide band and let the world in. Luckily this is a problem at the ground station and there you have more space to place antenna better.

    However, in ALL cases, if the receiver in question has a really inadequate RF front end and is desensitized ( which was your question in the first place) , then all bets are off. Filtering may or may not work - the problem is that sharp effective filters will introduce losses, ( a good few dB) and that is normally unacceptable.

    You can only try the setup and see what works..


  • @travis;

    Sorry about the typo - after a while one tends towards the unbalanced a s well...

    The image posted on the crossed dipoles show a way of matching and providing the 90deg electrical phase shift between the two antenna elements. However, you may have spotted that that particular antenna is matched to a 75ohm coax feeder system ( the feeder that would connect to your tx/rx)  If the TX were a 50ohm system,the TX would 'see' a 1.5:1 SWR, which as I intimated before, is not critical.

    The images below shoe a 50ohm feeder system with the matching cables are how they are derived. If you work back, you will see the method holds true for the 75ohm feeder system as well.

    In essence:

    The dipole has a feedpoint impedance of approx 70 to 75 ohms at resonance. If both were simply connected together at the coax, the two would be in parallel, giving a feedpoint Z of 35ohms ( SWR of approx 1.4:1). Also, the now crossed dipole would still not radiate circular polarization since the elements are now fed in phase.

    Since for circular we need to have a 90deg phase feed lead or lag, we can achieve this by inserting a length of coax that is electrically 90deg long ( taking into account its velocity factor, etc). However, using 50ohm coax all round will still not match the antenna to the feed system and TX/RX. Transmission line theory allows us to compute the required length and impedance of the various pieces of coax, to be able to achieve the 90deg phase lag AND a better impedance match.

    So, what we need to do is to transform the dipole impedance up to 100 ohms, so that when connected in parallel they present 50ohms to the feed system.

    Since our feed coax is 50ohm, and we wish the dipole to exhibit 100ohms, we need to compute the impedance of the 90deg length of coax to use as the impedance transformer, from this:

    Z_unknown = SQRT( Z_required * Z_Terminal)      ( no idea how to do SQRT token in this post!!)

    Z_unknown is the Z of the coax section that will be used as he matching transformer.

    Z_Required =50ohms, the Z we want the system to be

    Z_Terminal = 100ohms, the Z we want to transform the dipole impedance to

    So Z_unknown = SQRT( 50*100) = 70ohms.

    Since 70ohm coax in not common, we use 75ohm; the resulting small mismatch is negligable.

    We now have a very well matched, parallel connected, crossed dipole antenna, but is still does not radiate circular polarization. We have to add a section of coax, 90deg long, to one leg of a dipole feed, to generate the 90 degree phase lag ( quadrature feed)

    We use an electrical 1/4 wavelength of 50ohm cable, (NOT 75ohm!) since the system is required to be 50 ohm, and is already matched with the two 75ohm coax lengths previously computed. This is fitted to one leg only. The one leg will give left hand circular, the other leg right hand circular.

    Coax lengths are 1/4wave times the velocity factor ( from manufacturers cable specs).

    Unfortunately it is not easy to cut these lengths accurately at the higher frequencies ( 2.4GHz and up esp) and an 1mm length difference in phase feed can reduce the circularity ( axial ration) easily by a dB.

    There are ways of measuring the length ( apart from fancy network analysers, etc) but still some instrumentation is required - a signal source the frequency of which can be swept above and below the desired frequency, and means of accurately measuring the sources frequency, etc.

    For 2.4GHz:

    Short circuit one end of the 1/4wave coax.

    place a single turn loop of wire across the other end of the coax. ( loop made from thin copper wire, 10mm long)

    Attach a similar loop of wire across the terminals of the signal source.

    Then you need a sensitive microamp meter, say around 50uamp. connect another loop of wire, one end to the -ve terminal of the meter, the other end of the loop to a small germanium diode, and the other end of the diode to the positive terminal of the meter. This is now an 'absorsion wavemeter' Bring this loop into close proximity with the signal source loop, not touching. Adjust the output level of the signal source for a solid indication on the meter - you are rectifying the energy from the source and the DC amps flow in the meter. Now bring the loop of the 1/4wave coax in the same proximity as the other loops. Sweep the frequency of the source up and down. At some point, the coax with represent a resonant circuit, and as the other end is shorted, will absorb energu from the the signal source. This will 'take away' energy from the meter and the meter will show a drop in reading. Read the frequency of the source - that is the frequency where the coax is an electrical 1/4wavelength. If to low in freq, cut some of the shorted end, replace the short, and try again, etc. Not that the short circuit must physically be very short, fold the braid over and solder to the center conductor - any excess length here negates the measurements.

    Pardon the dissertation..




  • Could you also speak about the co-location of antennas nearby each other with regards to desensitizing receivers when transmitters are operating nearby on different frequencies? 

    For instance if on my FPV tower I were to locate my 1258mhz video RX along with my 900mhz telemetry tx/rx on a horizontal bar of some length how would I go about determining the minimum acceptable spacing?

    I realize that vertical spacing is much preferred over horizontal, but if you had no other choice. 

  • Part of my reply was culled because of the html-like angle brackets I used.

    I said "I believe it's a half wavelength" before 'length' in my previous post.
  • Travis, I can't provide a reference right now, but using a length of 75 Ohm coax is a standard technique for transforming the impedance seen by the radio.
This reply was deleted.