Antenna Builds

I build all my antenna and have build quite a few 70cm and 2m Antenna including a 4 bay 10 element 2m yagi with rotatable polarisation. I tend to use the powabeam designs from G4CQM whose designs are very repeatable. I've also used some G0KSC designs that have been good but require quite a bit of fiddling to get to work well.


 My current project is to build a low cost 2.4m dish for 23cm and up.











The design parameter for this are:


  • it has to be lightweight.
  • it has to be as low profile as possible.
  • It doesnt need to be too rugged as the dish lowers down beside the shack when not in use
  • It has to be relatively inexpensive
  • it  should need no special tools.

DIY 2.4m dish

I based the design on the RFHamdesign 2.4m mesh kit. But changed the rear plate from circular to a piece of 300mmx150mm by 6mm thick sheet to allow two arms to be fitted to join to the rotator.


6mm backplate to support the 12 arms M6 holes

The first issue was bending the aluminium frame into a parabola. This turns out to be a lot harder than expected.


My first attempt was a home made tube bender made of some scrap aluminium and some bearings. I then routed out the profile onto a sheet of 18mm ply and slowly bent the 5/8ths 16mm square tube till it fitted in the slot of the plywood.


This actually proved to be very time consuming to get a good consistent bend. As it's a parabolic shape you really need to have several large radiuses to bend it round to get the parabola accurate to a mm.


The second issue was that the rear support frame needed to be coupled to the front surface to give it some strength. This looked like it was going to involve a lot of cutting of aluminium, drilling and riveting.


I experimented with aluminium brazing - but frankly it still seemed to take ages to build. Since like all my antenna builds - these are meant to be experimental and to try new construction methods.



Eventually I settled on each of the 12 arms of the dish to consist of a 3mm thick 20mm wide strip of aluminium to form the front face that the mesh would attach to. The rear support would consist of 16mm square tube with 1.6mm wall.


The front strip is easy to form to the right shape, the rear support frame would only need two small radius bends.


To join the support and the front strip , I desided to use polyurethane moulding materials with blind rivet nuts embedded in it to make a fast to build lightweight support.


Each moulding was placed at the bend points of the support to reinforce the bends, The coupling to the front strip helps add to the strength of the structure.


A quick test with a 2Kg weight at the end of the 'arm' showed a distortion of only one milimeter.



On my dish I chose a small distance between the front strip and support frame. If I was building a stronger dish I would have probably started the frame 4" behind the front face of the strip to give it more rigidity - but for my use I started at 1"


The moulding.


My first idea was to use the base board where I had routed out the profiles for the 3mm strip and the 16mm support to fit into as the bottom of the mould. I would just rout out the the areas to fill with the resin and coat them with varnish.


As I had run out of plywood, for this test I used MDF which is easier to work and get a smooth finish on. However it turns out even with varnishing and smearing the MDF with vaseline, it still stuck to the MDF when taking it out of the mould.


So I built small aluminium chambers as per the picture above which released much more easily. However it was very difficult to get the moulds to seal - so I ended up using my granddaughters play-doh to fill in the cracks to stop leaks - not much else available early on a sunday morning!



I used polycraft c-3680 resin as it had a good combination of strength without becoming too brittle like some resins. I added about 120% filler to it to save a bit on the cost.


For each arm I mixed up 25g +25g of the resin and 50-60g of filler. It took some experimentation - too little filler and the mixture tended to leak from any small gaps - too much filler and it was difficult to get a solid fill with a good top surface.


A 1 Kg pack of resin mix  and 1Kg of filler is enough to build one dish.

Mesh supports


I opted for two 5mm rods bent as circles to support the mesh between the arms. However as the mesh is mounted onto a 3mm thick bar - and the 5mm rods go behind the bar, I needed to use a stand off mount.

I looked at using tie wraps and fitting sections of rubber fuel hose around the bars - but that seemed messy. The other option was terry clips - but those looked hard to attach to the mesh. In the end I found on ebay 3/16th brake pipe mounts which were cheap and easy to use.





This is working out suprisingly inexpensive. A complete 2.4m mesh dish including the rotator mounts etc would cost me about £900+ delivery. I'm aiming at a quarter of that at most!


I designed the dish to use metal lengths of 1250mm as  the standard sheet length is 2500mm and the length of square tube are most economical in 5000mm lengths.


This created some compromises as the linear length of a 2.4m parabola is 2.53m - so by fitting a 100mm dish in the centre meant the strips could be kept to 2.5m. Also the support frame would ideally be made of 2.58m lengths - so with this design the last 1cm of the parabola is unsupported - but is bent through a right angle to attach a flat plate strip round the circumference of the dish which gives it a little more strength.


Rather than buying flat bar for the front strips , I found a company that would sell sheet aluminium cut to any size. So the parabolic strip was purchased as 20mm x2500mmx3mm sheets and the mesh cover was 20mmx2500mmx1.6mm sheets. This reduced the cost of the strips by 70%


To finish the outer ring of the dish I bought 3 more strips of 2mm sheet. . I chose the material for all  the strips to be cut from  harder than standard  aluminium sheets to be more like flat bar.


*The mesh is sandwiched between the 3mm 20mm wide strip that forms the parabola and a 1.6mm 20mm wide strip riveted on top.


The rear frame consisted of 6 of 16mmx2500mmx1.6mm square tube, cut into halves and each part then bent in two places.


The support plate that the arms mount onto is 300mmx150mmx6mm aluminium plate.


The circular front was from ebay - a 100mmx4mm circle for a couple of pounds.


The mesh I chose was an expanded aluminium mesh sandwiched between the strips and also supported by two 5mm round rods formed as circles for extra support.


The mesh has apertures of 3mmx1.8mm which should allow me to work higher frequencies.


The total cost for the dish including nuts bolts, polyeurathane resin, aluminium, rotator mounts and amplifier mounting plates has come to £220


The parabola


The parabola was chosen to be a f/d of 0.42 to suit the feed I have in mind. Calculating the parabola was really easy using a freeware parabola calculator that can easily be found on the web.


The hardest bit was marking out the parabola on to a piece of MDF and trying to keep the accuracy to within a milimeter.



Marking up and routing out the template board was a slow process to get good accuracy - about 3 hours in total.


Cutting the aluminium and bending, drilling and moulding the arms took a total of two days - mainly due to needing to leave each arm to set for an hour - but in the drying time I could drill the front plate and the centre circle etc and mount the previous arm etc.


In total  I think this project will be about two busy weekends ( or in my case several weeks as I never get more than half a day free at a time) - but could be done in a weekend at a push. It was slower as I was experimenting a lot of the time.

2m Yagi Polarization Rotation

Whilst the obvious thought entering your head at the moment is why bother - just buy  x-pol yagis. In my case that isn't an option.

This antenna needs to fold down and dissapear so it has to fold flat. Imagine something out of thunderbirds......

So my only option is to build an antenna where each of the 4 yagi's on the H-Frame can be 'spun' along the long axis to give me continuously variable polarisation, but can rotate the elements into a position where some mechanical origami means it can fold up and dissapear into the roof of the shack.

The Issues this raises are

  • Loss caused by metal in the plane of the elements
  • Weight of the rotation drives and inertia effect on the az/el rotator.
  • Rotation matching of the 4 antenna
  • Cost.

However on the plus side it should give me an average of 1dB improvement over a switchable X-POL assuming I rotate it to within a few degrees...

Oh and I only need one four port coupler and one less relay before the preamp.

But being honest - it's a pain in the backside , but another challenge to take on. Now it's up and running - I think it is probably more effort than it was worth!

Basic Specs/Requirements:

DC motor Driven

64 position encoder feedback  (changed 16/1/2017 to analogue - see updates at bottom of page)

Low noise

Low weight

Low Cost

Rotates 9 element yagi on 25mm dia 3mm wall tube

** UPDATES*** ( older updates at bottom of page)

Update 27/5/2018

Sadly it is time for the 2m setup to come down, I knew I wouldnt be able to get away with it for ever - so the new challenge is building a 70cm system. Initially this is going to be fixed horizontal polarisation - but I suspect at some point I will move the rotation system over to the 70cm system.

Update 20/3/2018

The folding up process meant that the 3/16 elements were getting bent - so I have just finished building four new Yagi's using the G0KSC 9 element owl design. These have a 1" boom split at 2m from the rear with a 20mm round core to fit the bearings on and nice chunky 1/2" elements.

Initial tests show good gain and pattern and a very nice flat 1.1:1 SWR range extending to nearly 2MHz and they seem totally unaffected by rain and last weeks snow.

I'm just about to add an Alfa Spid RAS  on to replace my kenwood/solenoid elevation combination so I don't have to keep remembering to reposition the array.

Update 26/1/2018

First night back on air. For the first couple of hours it worked well and it was fascinating to see the effect of rotating the polarisation, it certainly does make quite a difference. The difficulty was that the natural fading made it quite difficult to peak  a signal well. Also several contacts mentioned the fading on the signal - mainly me getting the transmit polarisation wrong and adjusting it while transmitting.  Sadly after a while the performance seemed to drop and after shining a torch on the array it was running 3H1V most of the time as one  of the potentiometers had broken. Of course it was one of the top Yagis. Luckily I had decided to order some spares on the 23rd and I'm hoping for a delivery today to replace it for tonight.

No video of it rotating on the roof yet as it moves to position (+/- 90 degrees) faster than I can set the new position and get out into the garden to film it) -

To do :

1) Build a remote control unit so I can operate the system from on the roof.

2) Rewire the elevator so pressing the up switch makes it go up and not vice versa.......

Back to the design and build

Obstruction Loss

The obvious place for the drive is to drive it from behind the reflector - however this meant that the antenna could not fold and I was sticking the biggest weight at the end of the boom, greatly increasing the inertia felt by the main rotator.

So Decision 1 was to put the motor in the centre of the boom.

Whilst mechanically it seems sensible to put the motor drive on the H frame - this means unless the drive is really small it is going to interfere at some position of the elements. Since the elelments have to rotate 180 degrees to get the most benefit this makes it rather difficult.

So the second was to put the drive on the boom itself.

This means the drive can be mounted under the elements at right angles to the elements.

The drive system was designed with the smallest possible profile and the cables then feed along the underside of the boom and out the back with the coax feed to the antenna.

Luckily I have a clear unobstructed view of about a mile so I have tested the effect of the motor assembly on the pattern and gain of the antenna - and I can't measure any difference - any effect is certainly <0.5dB which is about my limit of measurement accuracy.

So as you can see from the sketch the motor drive rotates round the pole of the H frame meaning it is always out of the way of the elements.

The only loss is that the pulley on the boom has a diameter of ~50mm which means it extends 12.5mm outside the boom.

However I use Paraclips on my aerials to raise the elements above the boom - which raise the elements up by about 20mm so this shouldn't be an issue.

The system has sufficient drive to actually fit the support arm of a yagi through the bearings instead of the boom going through the bearings and still rotate a decent sized antenna - allowing through boom elements to be used. However this can lead to vibration so would need some experimentation to make it work reliably.

This Sketch gives a better view of the arrangement. The antenna boom is supported by two bearings which are fitted to the GRP tube. This has two advantages - one of which is that it spread the loads

and the second of which is the GRP is quite cheap (it was left over from the rest of the H frame) and has useful parts like the long Tee fittings which simplify the construction.

Obviously it spreads the load across the boom as well.

My antenna are about 4.5m so will self support with 25mm Aluminium tube with a 3mm wall.

The Drive

Initially I wanted to use gears - but there was a lack of good aluminium gears suitable for the task.

Also using a belt drive can simplify the design a little.

The biggest issue on this was how to build this without access to a machine shop - but relying on the tools I have lying around.

This meant some 'interesting' choices were made.

The first choice was stepper motor or DC motor. I chose DC as it offered me a physically smaller motor and substantially cheaper - and easier to drive. Stepper motors need a good reduction gearbox to be useable and missing steps can be an issue.

The gear ratio of the pulleys changed many times - In the end I selected a 32 tooth T5 pulley (10mm belt width) mainly because its boss was 38mm - which could be slid into a 40mm tube with 1mm wall thickness that would then slide into the grp of the support arm.

I also found some 25mm bearings with 37mm O/D that could similarly be placed in the 40mm tube.

One thing to remember with the motor is to fit  3 capacitors  across it M+to M-  M+ to case, M_ to case.

The motor will be running during RX so we need to reduce the noise as much as possible. I can't hear any noise during rotation - but that may be due to the high noise floor here rather than a comment on how good my screening has worked! * After replacing all the PSU's in the house with analogue ones and replacing the old plasma I can hear the difference.

Bearing details for internal mounting on 50mm grp tube and for external mounting 32mm tube.. This required a fair amount of hammering and filing!

Position Detection & Motor Control


This was my favourite part. My initial design relied on a potentiometer. But This either meant a multi turn pot or needing quite a large pulley to match the rotation of the boom to the 260 degree travel of the pot shaft - either way this would enlarge the metal area and could affect the antenna performance.


I thought about rotary encoders - but they have the same issue if driven off a pulley  and so I was looking at making my own reflective encoder stuck to the back of one of the existing pulleys as an alternative.


Quite by chance I stumbled on a 16 pules (64 positions per rev) encoder on RS for under £2. (RS Stock No.729-5927)  And it had an 18mm diameter bore - That sounded familiar - that happens to be the boss size on a 16 tooth gear.

That meant I could mount it on the motor shaft engaging with the boss of the pulley and keep the design to 2 pulleys.


This gives me relative position - but not abolute - so by adding a single reflective optocoupler on the motor mounting plate I could detect a mark on the boom pulley to give me a starting point for rotation.


Since these components require a pcb, I took the opportunity to add a PCB with pads and a hole so that the cabling could be soldered to the front side of the board on pads and then fed through the mount and out to the back of the antenna.


As the motor current is intermittant and <400mA at Stall I found some 6 core screened cable 20SWG for about £0.80 a metre.


The wiring is

+5V to reflective coupler LED (red)

0V (screen)

Encoder A (yellow)

Encoder B (white)

Zero datum (black)

Motor + (blue)

Motor - (purple)


Each of the drives then feed back to a central controller on the H frame next to the 4 port coupler.


A suitable controller was the sparkfun MKII motor controller shield which can drive four DC motors - stick it on the back of a clone Arduino MEGA 2560 in a box and hey presto -  one motor control system.


I was going to use the Leonardo as it has 4 interrupt pins and each encoder needs at least one interrupt to work correctly, However Idecided to use serial comms between the shack controller and the motor controller so I had to upgrade to the MEGA so I could have 4 interrupts for the encoders and still have interrupt driven serial communication between the two units. This only became necessary as I decided to control the elevation from the same controller.


I use the Encoder library for the teensyduino as it supports 4 encoders.


Also I added some debouncing on the encoder outputs














The old motor controller mast unit on LHS and new pcb on RHS. The relays are for controlling the elevation jack and the feedback is from a potentiometer on the A0 input which is belt driven from the main cross piece. This had to be modified to add two independent power supplies as the screw jack was pulling down the arduino supply and the relay noise was causing issues. As per update 31/12/17 a new arduino shield has been designed to go between the motor pcb and the arduino mega - this has all the interconnections brought out to 36 connection points and individual 8V and 5V power supplies as well as the jack relays.

So whilst the above picture looks tidy - this was when it was on the bench and before the extra boards were put in and extra wiring was added for the jack and pot etc.


Sadly I threw all of the above out - the PWM noise and background noise from the processor on the boom was detectable so I rebuilt it with potentiometers and  moved the processor and control into a case in the shack where it could be screened filtered and provide no detectable noise when moving during RX.




Shack control and Polarization rotation in use


For the shack controller I had originally thought of just bringing down pushbuttons to control the motor controller. I decided instead to build a handheld controller with an LCD screen left over from the linear build and some controls.


Since I would need to use the controller on the roof whilst supervising the folding, I fitted waterproof  RJ45 connectors so I could plug the handset directly into the motor controller - which would also provide 5V DC to the handheld unit.


I fitted a joystick to set the polarisation rotation target and the elevation jack target. After a couple of nights of experimentation I found that I needed to quickly change the polarisation in the RX><TX changeover time so I could reverse the effect of the Faraday rotation.


 I added a potentiometer to set the expected geometric rotation change of the remote station. Then when I found the best RX signal I could immediatly( in the TX/RX changeover period)  press the switch on the joystick to rotate the polarisation of the antennas the opposite amount of rotation to the geometric rotation.


So lets say the geometric rotation between me and the states at a particular time is 60 degrees.

On RX I get the strongest signal at 90 degrees.


So to present the best TX signal to them I press the button and the polarisation changes to 60-(90-60) degrees or 30 degrees.


If the geometric was 70 and measured RX 80 then the TX changes to 70-(80-70) or 60 degrees.


I also exposed the usb on the handheld so I could commant the elevation rotator from the PC and also put in a P.T.T. RCA connector. The idea being that the RCA connector can supress the encoder inputs during TX so that if I suffer interference when I'm running high power, the antenna won't wobble around! I havent had that problem yet - but I'm only running 400W (UK legal limit) at the moment.




My current line of thought is to stick the motor body in a length of 40mm waste pipe and seal it - I don't think heat will be a great issue.

As for the PCB/Encoders/wiring I think I am just going to pot the lot in resin as one assembly. They sare only five pounds in parts cost so if they break I will just build a new one!


The weight of the drive,including  angle bracket, 2 pulleys, bearings, both bearing housings, u bolts, motor and encoder mechanism is 512g.


In the table below I have detailed the relevant parts cost.

As you can see it is not too bad - around £175 all in. You can save more by shipping around - I certainly did so it has cost me less than that so far, Having said that I've only got one prototype and a pile of bits to make the rest with - so It may end up costing more.

As well as the parts for ther drive system I also had to buy parts for a new GRP H frame and new 25mm booms for the antennas

So in addition to the drive parts I needed

1 off 6m 44mm GRP tube

4 off  GRP long  tee

4 off 25mm 3mm wall aluminium tube 5.1m

As an approximate guide for the drives etc:

Figures in Red are the parts added for the analogue V2 version





Motor 5rpm 12v








Belt 250mmx10mm T5




32T pulley T5 10mm




16T PulleyT5 10mm




25mm Bearing












40mm tube






£0.80 /m










Leonardo Clone




Motor controller




Aluminium Angle 3x2"




ubolts 25mm




5 Turn 470R Pots




Idler Pully




Prices inc VAT & Delivery *=ebay!                                                                            total £195


One option is to use a smaller set of pulleys - say 12 teeth and 27 teeth. This would mean the outside diameter of the boom pulley is reduced to 43mm - or if you are really handy with a lathe a 25 tooth pulley could be used and the outside diameter falls to 39mm which keeps the total diameter of metal parts on the top side of the boom to be only 8 mm above the boom surface.

Another option is to go the stepper driver route - it does not need any rotational feedback as with a sufficiently sized motor you can count steps - so only the index opto is required.

You would however need a second motor driver to drive the extra coils - but luckily the adafruit shields are stackable.

However using potentiometer feedback and a DC motor for the Version 2 ( analogue) model seems to now give the best results.

Build Photos (these are with a 100mm bearing sleeve rather than 2x50mm bearing sleeves to be used in the final version)

Video of unsealed unit

Update 23/1/2018

Well it took a long day - but the new controller was built and wired and the old encoders were removed and the new potentiometers and extra pulleys fitted and the extra wiring to the shack got made. So it is now an analogue system. The noise while moving is pretty much undetectable which makes it easier to spot a weak signal on the waterfall and rotate to suit so I'm very happy.

BIG Update 16/1/2018

After some fun testing the system and pleased with parts of it - I have come to the conclusion it has to change! Quite a shame since the new PCB (see image below) did a good job of making it a neat job)

Basically sticking 1500W into the aerial does cause miscounts on the encoders - Not many but enough to irritate me. I was going to add some more caps and shielding but then..

After having spent months removing all the switch modes from the shack and the house and the neighbours - the noise (albeit small) from the computer and motor controller on the H frame means I've decided to go analogue instead. I'm blessed with a pretty low noise floor here -( I can hear when my neighbours half a mile away switch on their main LED's in the lounge)

the extra 1 - 1.5dB increase in the noise floor is bugging me.

So the encoders are being replaced with 5 turn pots driven by an idler pully which just about fits on the brackets. The motor control is now in the shack and  also passed through relays to disconnect the motor lines when the aerials aren't rotating ( not strictly necessary but I had a relay board spare).

The new controller is half built - the metalwork is done and the code from the 2 arduinos have been merged onto one arduino and modified for analogue input instead.

So as long as the extra pulleys arrive in time then I should be back on air just in time for the end of the month when  degradation falls to a minimum.

So when the world is going digital - I'm going back to analogue!

Update 31/12/2017

Well the system works and has withstood 5 months of sun and rain. During that time though I have been too busy to actually do any EME work - so hopefully I will be back on air in the next week or so. It has been a bit of a slog and the system has been modified a few times over the summer and autumn. - the verticals of the H have been rebuilt again using 50mm grp instead of the 44mm grp as it just flexed too much in high winds for my liking.

The old controller box has been binned and i have designed a new pcb for all the connections and power supplies and relays. It was just too cramped in the box and this made it a bit unreliable.

Update 07/07/2017

I have built and tested the handheld controller and had a play with a single yagi and experimenting with adjusting rotation during reception on a couple of nights.

Currently I am rebuilding the motor controller box into a bigger box as the original didnt have enough room for the elevation screw jack relays. I am also fitting the handheld controller into a box rather than have the bare PCB lying on the bench

All 4 antennas and rotators are up - but I'm now fitting inline connectors to the antenna for the motor feed as when I take the antennas down to experiment with, the leads mean I cant move the antenna away from the mast to work on them.

I'm a bit busy to update the page, but I now have the new collapsible H frame completed - using 44mm glass fibre tubing. This meant fashioning some adapters between the 50mm plastics tees and the 44mm tube - but it saved a lot of weight. The next job is to move the whole assembly from the garden to on top of the shack. At the moment the system is handcranked from flat to raised - I think fitting the electric winch is a job for next year.

(update 21/06/17 I have now moved the whole mast/frame/support onto the roof of the shack)

I have also had the pcb's back - with one irritating mistake where the encoder connectors were in completely the wrong place- but they work well enough and I have finished the first antenna and rotator assembly which works exactly as it should. I'm torn between waiting for new pcb's for the final version or just going with what I have.

It needed some hacking as the 25mm test boom I used was indeed 25mm - however the actual booms measure 25.3mm and I could not get the 25mm bearings to fit - however much I heated them. I then had to change to bike headstock bearings which are 27mm diameter - so I added adhesive lined heatshrink to pack the boom out.

On the plus side I got some 41mm O/D tube with an I/D of 38mm - this cut out the need for mucking about with the lathe as the pulley and bearing just slide in with the aid of a hammer.

The only tools you need other than a basic toolkit are a vice, 30mm+ cone cuter or 30mm drill and a typical dremel tool.

So I can sit in the shack and dial up the desired polarisation angle and the antenna rotates to it. So just 3 more to build......


Aluminium Tube ( long Lengths)        

50mm GRP Tube and Tees:   (V Cheap!)

32mm GRP tube:                                   

Pulleys  (Although cheaper from China!) :