Updated 20151231
End Connected Windom
This page describes a successful antenna designed during 2014. It covers 4 HF bands 40/20/15/10m. A developed version of the old "Windom" idea, where the feed point is off centre. Unlike the standard design it is fed from the end, giving it the name "End connected Windom" or more correctly swap the name Windom for off-centre-fed-dipole.
The mechanical layout makes it ideal for "portable" use. Two fibreglass poles can be used to set up in only 5 minutes. One pole can go in a car wheel base, so only one lashing point is needed. With shorter versions such as 20/10m which is only 10 m long, a single fibreglass pole can be used.

The secret of the design is using common mode current on the feed line in a controlled way to radiate. The heavy ferrite coaxial choke is at one end, and there is no feed line coming off the antenna near the middle. A second secret is the split point, which enables low SWR on the four main bands. Combining these ideas gives the block diagram as follows.
The antenna can be used with the coaxial choke (Guanella choke) in a convenient mechanical position. It can be near the bottom of a fibreglass pole if the wire is put into an inverted 'L' format. In my home installation the choke is mounted on a TV aerial mast. This is an ideal arrangement for properties with the garden directly behind the house. So on with describing the parts to make this antenna…

The 3.65m section is made of RG316 PTFE mini-coax, and combined with the small Ruthroff transformer does not weigh the wire down much. This is a big improvement over designs with the balun at the feed point. A key part of any modern Windom design is the balun/choke.

The Ruthroff transformer is wound on a ferrite rod. I used a rod bought from a radio rally for £2, but a less random part would be this one. I find using coax rather than the traditional twisted pair improves bandwidth. With a 200ohm resistor, the VSWR plot shows an excellent match.
Ruthroff RG316D

The short end/feeder is braid terminated onto the circuit board. A picture of a Ruthroff transformer built on the commercial standard PCB follows.
Ruthroff top
Here is a photo of the underside with wires connected. Ignore the surface mount parts.
Reducing the turns to 9 rather than 10 should move the optimum match to near 14MHz - ideal. Constructing two of these and measuring insertion loss in series gave the result of 0.36dB at 30MHz. So 0.18dB for one, and this is also worst case. At 14MHz the loss is a negligible <0.1dB.

Here is the circuit diagram of the board, which has a snap-off section at the end for anyone wanting to experiment with relay switching. It can of course be used at other split points, for example 17/15m with a switched 10m option.

Windom board

The board has a split off section to the right, which is for the optional relay switching. The relay switch components (R1,L1,L2) are entirely optional, again if you want to experiment with relay switching. The TE Connectivity IM06 relay is easily available and is the smallest SPDT relay which can switch 3A, and has a usefully high coil resistance.

Ruthroff PCB

This PCB is intended to be coated in
liquid tape. This removes the need for the transformer to be in a box, reducing weight and windage. The liquid tape sealing holds for years under British weather conditions (which are pretty horrible). If the relay option is used, a bias-T is needed to feed +12V up through the coax. A bias-T is just a high impedance choke to connect the power to the line, with a capacitor to decouple the transmitter output. Several designs for bias-T circuits are published.

Very high choking impedance is essential. To get it over a good bandwidth, a series choke with 2 types of ferrite is used. I wound 8 turns of the RG316 thru four stacked FT114-61 toroids. In series is 5 turns of the RG316 thru two
heavy type 43 beads. The cable is held in place with the top of a fibreglass pole fed thru the centre of the ferrites. An excellent mechanical arrangement:

The picture shows the coaxial choke after spraying with blue paint and some sealing varnish. With the antenna supported 6m high at one end and 8m at the other, I was rewarded with no detectable common mode current and a good VSWR:

Next steps:
1. Test the new relay board
2. Experiment with different coaxial chokes

Background history of End Connected Windom
The first serious home made antenna I put up was an end fed half-wave wire, similar to that described by AA5TB. Easy to make for single band QRP, but difficulties come with higher power and multiple bands. Wanting more bands led to a search for alternatives. Looking at HF antennas for small spaces (small is <20m long), then 40m is the lowest band that can be covered without using the ground. Using a ground I could fit an 80m ‘L’ antenna, and cover many bands with an auto-tuner. But I do not want to deal with putting down ground radials, and the ground loss becomes significant on the upper HF bands. I am less interested in 80m/160m because my VHF range is good for “inter G” contacts.

The type of antenna for a particular property depends on the layout. With the garden behind the house, as on most city plots, an end-fed is best. With the house in the middle of the plot a centre fed is best. In my opinion the two best wire antennas are the end-fed “windom” and the doublet with balanced tuner.

A doublet is a dipole not resonant at any particular frequency, and matched to the transmitter with a tuner. Just get up as much wire as possible, feed in the middle with ladder line, and use a tuner to match the radio. I have never setup such an antenna. Technically the best tuner will be a fully balanced one, with a coaxial choke at the input. The problem is no commercial fully balanced auto-tuners are available. Building one would be a big project in itself!

The rest of this page describes my development of this type of antenna, alternately called a windom and OCFD (off centre fed dipole).

Wanting more from the Windom!
The original Loren Windom magazine article of 1929 described a single wire fed 1/5 way along. This design has been superseded, but the name stuck. The modern way is to feed it with coaxial cable and control common mode current to make a reliable system. Given 20m of space the traditional design covers 40/20/10m, or 80/20/15/10m for the 40m long version. I wanted more bands and to keep it as light and simple as possible. For more bands these ideas were considered:
  1. “Stacking”, i.e. using extra elements at the feed point. This is called a “double Windom”, and is used in the commercially available RadioWorks antennas for instance. In future I might try stacking, but went a different way now.
  2. Bring a balanced twin line off the feed-point to an auto-tuner. Having twin line on an unbalanced antenna makes it unbalanced. So it would radiate. This idea is used in the “Carolina Windom”. Having an undefined vertical radiating section near the house wouldn’t be good for EMC. It would be parallel to the vertical house wiring, picking up noise. So radiating feeder antennas are rejected.
  3. Use traps (“Smithe’s Windom”) which is rejected absolutely! The nature of traps would make adjusting the antenna extremely difficult, introduce losses, and weigh down the wires.
  5. Use relays, and feed power up the coax cable to switch them. Effectively cutting off the ends to make a 30/15m OCF dipole. Feeding 12V DC through the coax, and through the balun. The DC voltage is fed into twin core inner sections. The two outer sections are then isolated by relays for 30/15m. The two core sections are twisted pairs, with capacitors on each end, shorting them together at RF. This provides a 40/30/20/15/10m antenna.

Relay switching is used for example in the
MacTenna dipole. In my opinion the MacTenna dipole is vastly overblown. Not only does it use unsightly 450Ω feeder, but its relay units are heavy and complicated. This means the MacTenna needs strong support at both ends. The OCF design, while covering 2 less bands, is lighter simpler and easier all round.

Relay switched OCF dipole for 40/30/20/15/10m
Can an OCF dipole cover 30m and 15m, when 10.15 x2 = 20.3, not 21MHz? I made the 30/15m inner section as a stand alone antenna with arms of 9.42 / 4.64m. It tuned to 10.5 and 20.8MHz. The mid-band VSWR on 30m is 1.2:1 and mid-band on 15m is 1.7:1. 30m is a narrow band of only 50kHz. 15m is much wider. Here is the VNWA plot of the 30m/15m antenna in practice:

OCF 30-15
The green plot is the VSWR, red is reactance, and blue is resistance. VSWR is below 1.5:1 across both the required amateur bands. Early on it became apparent the weight of the balun is a problem. A lot of tension is required to make the antenna reasonably “flat” because of the weight of the ferrite.

I built a bias-T circuit, from two 22uH inductors and some capacitors. The circuit allows feed through of 12V to the end of the twisted pair antenna elements. The near complete antenna has this VSWR (green) and r +/- jx plot:


The SWR is below 2.0:1 across the 40m and 20m bands, and part of 10m. Switching to the 30/15m response works fine. The SWR on those bands is similar to the first plot above. The capacitance through the relays made the length of the inner section rather shorter than expected. Several fiddly adjustments had to be made to the length of the inner section, which also meant lengthening the outer sections.
A sketch of the antenna design is given here.
Wire lengths are: 8.91m and 4.41m for 30/15m “inner” sections. 4.60m and 2.31m for “outer” sections.
Inductors are SC30 axial 22uH type (yellow). Relays are Panasonic TK1-12V, could use Omron G6E. Contacts must be suitable for dry switching, i.e. gold over palladium/silver.
Use 22SWG enamelled copper wire for inner twisted pair sections. Use PVC coated “flexweave” or similar thicker stranded wire for outer sections. The weight of the wire on the outer sections matters less than the inner sections.
Here’s a nice picture of the complete antenna at dusk against a crescent moon. The RG-316 coax is seen coming off the balun box, and the box has it’s LED lit indicating that 40/20/10m resonance is selected. The top of the fibreglass pole holding the far end is poking up in silhouette against the houses opposite.
OCF dusk
Five Band Half-Size OCF dipole
BUT... There are problems with the switched antenna:
  1. The DC coupling inductors skew the tuning
  2. Common mode current also skews the tuning
  3. The weight of the balun makes the wire and supports sag badly
The SWR results above are rather lucky, as my network analyser reading swings even from touching the cable, let alone touching the connector outer at the end to different metal surfaces. Even with 2x FT140-43, the common mode impedance is not good enough. The far end support is not very strong, so using larger heavier cores was impossible.

There's a design for a "Single-Core 4:1 Current Balun of Improved Performance" by Chris Trask (N7ZWY). It purports to be a current balun wound on a single core. It gives good bandwidth, but measuring common mode impedance with a network analyser is difficult. The N7ZWY balun does not have such good common mode impedance as the Guanella/Sevick design, but remember it only has one core. Going to 2 cores in series, the 2nd as 1:1, we can have CM impedance higher than the Guanella with the same amount of ferrite. A disadvantage of the Guanella is the self resonance of the 2 cores is at the same frequency. With 2 cores we can make self resonance different, so common mode bandwidth is increased similarly to having two different inductors in series.

The SWR and resistance plot of the final antenna follows:


The series balun approach tames common mode current, and a 23% split point allows 4 bands from a 20m long antenna. The improved common mode impedance of the new balun is the secret of making this design work. SWR is worse than the traditional 33% split version but an extra band to compensate! My 20W rig doesn’t mind a 2:1 SWR, also the feed point is closer to the house, making it less visible to the neighbours.

The 5-band mini OCF dipole is simple to construct, and the baluns are repeatable. Unfortunately 30m is not covered, but the whole of 40/20/15/10/6m is covered with a tuner on 10m. The 6m resonance is well defined, suggesting low balun and cable loss. The balun is not boxed, but covered in liquid tape, which saves a lot of weight. After 6 months the liquid tape coating is intact and SWR the same as when it went up.


This is an excellent design for anyone with a garden 20m long or slightly less. Taken Christmas Day 2011, the photo shows how unobtrusive this 5-band wire antenna is:

4band OCF-2

The SWR plot of the five band OCF dipole at another location is very similar to my own place. This shows the antenna is not dependant on special surroundings to get a good result. The second antenna has a worse SWR at 40m/15m, but better at 20/10m. It can be trimmed with more adjustment of the lengths. Ignore the red trace in the following plot.

chris hf
Vertical 20/10m Windom
An end-connected 20/10m version was wound onto a 10m fibreglass pole and erected at ground level. SWR is <2:1 across the whole of the two bands.
20-10 windom