Wednesday, July 20, 2011

Multiband Cubical Quad

Multiband Cubical Quad

" ...A thing of beauty and a joy forever..."

I wanted an antenna that ....

1. Would not be too expensive

2. Would cover as many bands as possible.

3. Would not compromise on performance

4. Would be at least equal to a 3 El trapped commercial yagi on all bands, eg Th6

5. Could be rotated by an inexpensive rotator

6. Would withstand severe weather conditions.

7. Could be built with readily available parts.

8. Would lend it self to future modification, experimentation etc

9. Could be fed simply with a single feedline.

10. Reasonable size.

So, taking these requirements what was available.....

Well first of all, virtually all commercial antennae were ruled out on cost and/or performance considerations. I looked seriously at the possibility of building a log periodic , but ruled it out on the grounds of size and complexity. So that left the Quad!!

The Quad meets all the 10 listed requirements with a few disadvantages, these being -

1. Three dimensional, therefore visually more noticeable (more beautiful??) than a Yagi.

2. Limits the use of your tower for wire antennae (Dipoles etc.) because the quad arms hang below the rotator.

3. Prone to damage from ice (we get no ice in EI).

4. Very difficult to install on a fixed tower. A tilt over tower is very desirable, as the quad can be built onto the tower when tilted over.

( Pic of tower tilted over, so that the quad is at the other side of the hedge on the lawn.)

OK.. We'll start with a specification of my present Quad (2 element).

It covers six bands, 20m-10m on HF, and also 6m. It is a Boomless (spider quad, gem quad)design. It uses Glassfibre arms (a must). It uses a single coax line to a homebrew antenna switch. From there, a seperate feedline goes to each of the Driven Element feedpoints. There is a homebrew Choke Balun at each feedpoint.


I have no accurate method to measure forward gain, but I reckon it is the text book 6-7 db.. F/B ratio is consistently 5-6 S points on my TS850 S meter, on all bands 20-10m. I don't know how many dbs per S Point for my TS 850, but it is surely at least 3db per S Point. So this translates to a minimum of 15 db, and arguably as high as 30db. As with all 2 element Quads, It has a wide beamwidth, about 60 degrees.

(Measured plot of polar diagram..15m. Using software by I0JX and TS850 to produce the plot)

Construction -

The spider was built for me by my friend Jim E18GS,(EI7M) who is an Engineer (He also built my crank up, tilt over tower). It is made of stainless steel, which of course, does not rust. The boom is about 18 inches long ( not critical), with the angle iron arms welded to the boom at an angle of 110degrees to the horizontal. The pipe, a foot long, is welded at right angles to the boom in the centre. Depending whether you weld the pipe to the flat side of the boom, or on the corner, gives you either a diamond or a square shape quad.

( Drawing of stainless steel spider)

( Pic of spider )

(Pic of spider mounted..First arm attached)

I use the square shape. This pipe slips into the 2 inch OD pipe which comes from my rotator, in my case the pipe to the rotator is about 5 feet long. The glassfibre arms that I use are about 16 1/2ft long, but 14 1/2ft should be long enough to accomodate the 20m loops. The arms are clamped to the angle iron arms (which are each about 1ft long) using jubilee clips (Hose clamps).

(Pic of element to spreader attachment)

It is a good idea to put circles of insulating tape around each arm at roughly the points where the elements will be attached (see photo above). Later, when you are tensioning the loops, you have a visible check , so you can have the attachment points at equal distances from the spider.

( Attaching the elements....Ger EI8HT strutting his stuff...Known in our QTH as SUPERMAN..!!)

There are many different ways of joining the element wires to the arms. I use two strong cable ties at each attachment point. One cable tie is closed to form a circle of about one inch diameter, and the other is threaded through this and around the arm. This allows the tie point to be moved up and down the arm to tension the wires. Do not drill the arms as this weakens them. Bob K2US, uses the best method that I have seen, of attaching the elements to the arms. He puts loops of nylon webbing around the element corners, and trap the ends of the webbing against the arm using stainless steel hose clamps. This is a stronger and more durable option, but is more expensive. When the Quad is built, and everything is tuned and working OK, cut lengths of strong wire about a foot long. Bind these around the elements at the corners where they attach to the arms. This helps to stop the wire from flexing and breaking at the corners.

( Pic looking up at one of the arms. Note the circles of insulating tape, and the spiral wound reinforcing wires at the corners. Also, note how the elements are being pulled away from the arms, which improves insulation.)

The elements-

Now for the controversial bit.....

The inherited wisdom for element size is 1005/f for the driven elements, and 4% to 5% larger for the reflectors.

The actual size of element that you end up with (usually after hours of adjusting , measuring and testing) ,will depend on a few different things, such as Type of wire, wire diameter, insulated or non insulated, etc.(In other words....The Velocity Factor)

If you use the following guidelines, and formulae, you can build this 6 band quad without tuning stubs, gamma matches or any other complexities, and it will work properly with little or no adjustment required......

The wire should be bare hard drawn copper, not less than 1mm diameter, and not more than 1.5mm diameter. (18-16swg). The formula for this type of wire is 996/f for the driven element, and 2.5% larger for the reflectors.

EI7BA Two Element Quad Dimensions(feet)
Driven Element Reflector
996.00/f 1021.00/f
20melementlength70.2970ft 3 1/2ins elementlength72.0572ft 1/2ins
17melementlength54.9454ft 11 1/2ins elementlength56.3256ft 4ins
15melementlength46.9846ft 11 3/4ins elementlength48.1648ft 2ins
12melementlength39.9439ft 11 1/2ins elementlength40.9440ft 11 1/2ins
10melementlength34.9535ft elementlength35.8235ft 10ins

A number of these quads have been built, using these formulae, with slight variations in wire diameter, but always using bare hard drawn copper wire. All have delivered the goods with no adjustment of element size, giving good gain, f/b, and bandwidth. I emphasize once more, that if you use insulated wire, or bare copper, different to what I have used, The above measurements will not be correct. For instance, If you use insulated wire, the loop size will be approximately 2% shorter, depending on the type of insulation. This 2% represents a 300kHz frequency shift in resonance on the 20m band. Other wire types and sizes will work just as well of course, but you will end up with different loop sizes.

( Almost There...Put your back into it Ger..! )

The elements should be mounted on the outside of the spreaders, because when they are tensioned, they pull away from the spreaders, and improve the insulation at the attachment points. When the elements have been attached very loosely to the spreaders, attach equal lengths of nylon twine about 12ft long (4 off) between the 20m driven element attachment points and the reflector 20m attachment points. Adjust the 20m element spacing with these twines. I use 11.5 ft spacing between 20m elements, which corresponds to about .18w/l spacing. Do the same at the 10m or 15m element attachment points with some more nylon twine, and then tension the elements on the arms. You now have a nice box like construction which is very strong.

So far, so good......

( pic showing feedlines and a choke baluns)


In my experience, when feeding a 5 band quad, it is not advisable to take too many short cuts. I found that if I joined all the feedpoints together to a single feedline, results were mediocre, with reasonable performance on 20m,17m and 15m, but very poor on 12m and 10m. There is a lot of interaction between elements, giving a bad pattern, poor f/b ratio and very little gain on 10 and 12m..

There are two major detrimental interactions..

10m is an harmonic of 20m, so the 20m loop will accept power when transmitting on 10m

10 and 12m are so close in frequency, that they tend to see each other's elements as parasitics within their own band..

If you absolutely refuse to use seperate feeds, then you should at least use two feedlines, and group 20, 15 and 12m on one line, and 17 and 10m on the other.. This should go some way to improving matters.

A Quad is a balanced antenna, and should be fed in a balanced manner. Although Gamma matching is an obvious solution, there is a simpler one... Choke baluns...

I use a single feed of 50ohm coax to the centre of the Quad spider. There I have mounted a stainless steel box. In this box I have a homebrew remote antenna switch. This consists of six medium sized 24v relays, with good quality contacts. These act as single pole, double throw switches. I run a multicore cable from the switch box in the shack to the relay box. This 8core cable allows me to switch bands by putting 24v on the appropriate relay. The unused relays are wired so as to ground the feedpoints of the unused driven elements.

( Pic and schematic of Relay box with the 17m loop selected. Note that all unused loops are DC grounded at the relays )

I used to feed my Quad at the bottom corner, i.e. half way between horizontal and vertical, or "cross polarized". I did this for three reasons. For curiosity, to limit the wind area, and to give me nice secure tie points for the feeders against the quad spreaders. It performed well, and was mechanically very good, as it held less wind with the feeds taped to the quad arms. However, I have now reverted to the traditional feedpoint half way along the bottom horizontal wire.

Shortly after I built this Quad, I had some interesting email converstaions with L B Cebik W4RNL . He Modelled my Quad, and pointed to the fact (which I had noticed on air) that the side nulls moved towards the back of the Quad with this corner feed arrangement. Having reverted to the traditional feedpoint, the side nulls are back where they should be.

( Pic showing the stainless box mounted on the spider, with the homebrew switchbox inside. The stainless steel box has a weatherproof lid, which was removed for the pic)

Up to Christmas 1997, I fed the loops using 75 ohm 1/4 w/l matching transformers which were connected back to the switch box using 50ohm coax. This was to match the theoretical 100 ohm to 120ohm feed impedance of the driven loops, at the wider spacing that I use. (.18 w/l, as against the more traditional .12 w/l). However, having had damage to the quad on Christmas Eve 1997 during 110mph gales, I decided to feed the loops directly from the switch box with 50ohm coax as I was in a hurry to get back on the air. I was amazed to find a nice swr curve on all bands (1:1 in most cases at resonance). I have'nt as yet fully figured out why this should be, but for now I have no plans to change back to the 1/4 w/l transformers. (If it ain't broke...don't fix it...)

As I stated earlier, LB W4RNL took an interest in my attempts at Quad building, and modelled my Quad. You can see the results at his 2element multiband Quad section . This was very interesting to me, as I was not into modelling at the time. He came up with an explanation as to why my SWR curves are so good without the 75ohm 1/4 w/l transformers. It seems that when you have umpteen elements all nested together, the feed impedance is changed from the single loop value of 100 to 120 ohm value, to anything from 40 to 80 ohm, depending on whether the loop is in the middle of the group, or on the outside.

The Choke Baluns....

Since I have moved back to the traditional feedpoint, I now use RG58 coax from the Remote Switch to each driven element. I wind as many turns of the RG58 as I can, on an Amidon T200-2 iron dust core, wound as in the following picture..

I cabletie all the feed lines together, and tie the chokes to the feedpoint insulators. This arrangment is fairly light and strong, and survives well.

The original chokes consisted of 5-10 ferrite cores slipped over the coax at the point where the coax is joined to the driven elements. I used six at each feedpoint. They create a high impedance path to any RF that may feel inclined to run on the outside screen of the coax, thereby spoiling all your hard work, giving misleading SWR readings, screwing up the pattern, and generally behaving in an anti social manner...

ADDENDUM.. An extra two bands..!

As I stated earlier, my quad arms are 16.5 ft long. This meant that they were longer than needed for 20m by about 1.5 ft. As it is very easy to chop off the surplus length, and not so easy to add it back on, I decided not to " look a gift horse in the mouth", and use the extra length to my advantage. I have added two more bands..i.e. 40m and 30m. So now, it has 7 HF bands.

I have added three elements for 6m and 2 elements for the European 4m (70mHz) band to the existing spider, and on a seperate 6ft boom which is clamped to the spider, I have 5 elements for 2m, and 9 elements for 70cms.. A grand total of 11 bands..

On 40m, I built a 40m dipole, bent in a square, and attached to the tips of the arms on the Reflector side of the Quad. Because it is bent, the feed impedance was very low. Originally, I added a simple rolled up "hairpin match" at the feedpoint to compensate. The matching coil consisted of an airwound coil of about 8 turns, with a 1.5 inch diameter connected across the feedpoint. Pull out, or compress the turns of the coil for best SWR.

I updated this feed method in anticipation of the extra 100kHz on 40m that we got in 2005. I needed to increase the SWR bandwidth to cover 7 to 7.2 mHz. I did this by converting the bent dipole to a bent folded dipole. In one fell swoop, this has increased the feed impedance to 50 Ohm, and has given me a 1.5:1 SWR bandwidth of about 200kHz

This approach is not new. Steve G3TPW has been selling his "Cobwebb" antenna for years. It's a nest of these bent folded dipoles on a single feed. The well known VHF "Halo" antenna, which was much favoured by 2m mobile operators back in the '60s, is also a version of the same antenna.

Mark M0MRR, has a very comprehensive page on his homebrew version of this antenna type on his excellent website. It's a particularly good antenna for people with limited space.

Andre GM3VLB had an article published in RadCom (the RSGB Magazine) on his three sided version which he calls his "Mini-Delta". Andre does a lot of "island hopping", and has been giving out rare IOTA contacts for years. He is a past master at creating effective portable antennas for little or no cost.

The diagram below should be self explanatory..

On 30m, it is a full wave loop, attached to the arm tips on the driven side of the quad. The two linear loading wires (as per G3FPQ) are spaced about 6 inches in from the loop sides, and are connected to the loop at their mid points. These wires are adjusted for resonance.

Again, the drawing should be self-explanatory

The performance on 40m is as you would expect....That of a dipole at a reasonably good height, i.e. simple but effective..

On 30m, it is excellent. The loop is fed with RG58 via a choke balun.The SWR bandwidth is very good

There is NO f/b on these two bands of course, but there are side nulls.

Rob ZL3RG went a different route, and added a second element for 30m, instead of the 40m folded dipole.He added 3 1/2 ft to each arm of his LightningBolt 2 element Quad, and added the capacity hat loaded 30m elements. He has had great success, and sees 4 to five S points f/b, and deep nulls off the side. he seems to have a pileup to Europe every time he shows on 30m CW or digimodes.

The ZL3RG Modified LightningBolt antenna, with an added two capacity hatted elements for 30m

Glen WB4KTF's homebrew 5 band Quad

No comments: