Introduction
There is a lot of information on the internet about making 49:1 matching units for end fed "half wave" wire antennas. While some of it is quite useful, much of the written material ignores the basic difficulty of making a wide band unun that will handle more than a few Watts, especially more than 100 Watts of FT8, and give a low SWR across 3.5 to 30 MHz.

Most end fed half wave matching units use a single FT240-43 core. Using one, you can obtain a reasonable match across 7 to 28 MHz which will handle up to a peak power of around 400 Watts. A single FT240-43 core will overheat with more than 100 Watts of carrier. The above image shows a typical single core unun and the winding arrangement. A similar one is available as a kit from the ARRL. The above unun with a 67 foot end fed "inverted V" horizontal wire produced the SWR at the shack end of the feeder as shown below.

Going QRO (up to 500W CW/SSB, 250W data)
While the results with a single core are generally good enough, if you want to run higher power a single 240 size core is not sufficient. Stacking cores helps but you can run into problems trying to obtain a good match across the 3.5 to 30 MHz range. Using 2 x FT240 size cores helps with power handling. I have only experimented with 2 x FT240 size cores as my power requirements are modest.

The image above left is a completed matching unit with 2 x 240-52 cores wound with 2 + 14 turns, the result of a test with 2 x 1200 Ohm resistors in series across the output is shown above right.

Winding techniques
These are critical, tightly squeeze together the first half of the winding, and space out the second half. I didn't find it necessary to twist the wires together, heat shrink sleeving helps keep the wires close together while winding, also using a Ty-rap plastic cable tie keeps the first half tightly coupled. Some experimentation is often needed, having an antenna analyser that can display several bands simultaneously is useful.

There is a 100pF 3KV capacitor across the 50 Ohm input. Note, when testing/building one of these ununs, do not place the bare transfomer directly on an anti-static bench mat as it can alter the characteristics. Place the transformer on a plastic box for test purposes, and when mounting it in an enclosure, mount the transformer away from the base of the box using cable tie mounts. If placed on the ground (as may be the case for end fed vertical use), close proximity to the ground may alter the matching.

The above images show construction using a tapping on the winding, as opposed to two wires in parallel for the first 2 turns. The results are broadly similar. Using a type 61 core can produce good results, but over a smaller frequency span than a 52 core. Again the output is connected to 2 x 1200 Ohm resistors in series. Note again the bunched winding, followed by wider spacing.

From my experiments using a variety of cores and winding arrangements, it seems that squeezing together the first few turns produces better results than spacing all the turns. There is much discussion concerning "crossing over" the winding, or not. I don't think it makes much difference.

Core types
Type 43 ferrite is often used for 3.5 to 30 MHz, a single core will handle up to 400W peak and around 100W of FT8. Two FT240-43 cores can probably handle double that power. If you don't need coverage down to 3.5 MHz, type 52 ferrite is OK for 7 to 30 MHz and has slightly more stable characteristics as the temperature increases, if you are only interested in 14 to 30 MHz type 61, or possibly 67, may be more suitable. Type 52 is my favourite as I can obtain a better match over 7 to 30 MHz than with type 61, although type 61 ferrite has good permeability across a wide temperature range.

Testing
Several users place a lot of emphasis on making two identical transformers and testing them "back to back" with a VNA, producing a table of dB loss against frequency. Those who buy a ready made transformer probably won't want to buy another one for testing purposes, although if building your own, making another one isn't hard.

Moving the wires around the core by even a small amount can make a significant difference to SWR when terminating the output with a resistive load. Note that using a resistive load is not the same as an antenna wire and may not give as good a match as you think it should... remember, a 49:1 transformer is not an ATU.

Installation and use
Whether you like it or not, a horizontal end fed wire will use the outer of the coax as a counterpoise, this can give rise to noise pickup on the coax and RF radiating from it. You can use a common mode choke close to the transformer and a counterpoise wire cut for 0.05 of a wavelength on the lowest frequency (6 foot 8 inches for 40m), or move the choke so it is 0.05 wavelength from the transformer and not use a counterpoise wire. The choice of that particular wavelength is to try and obtain an optimum match.

A 49:1 transformer works well feeding a horizontal half wave of wire for 80 or 40 metres (128' or 66') and also works on the 20/15/10 metre bands. The addition of a coil of 6 turns of the antenna wire around a plastic former of between 1" and 1.5" in diameter, placed 78" from the transformer, helps with matching on the higher bands.

Limitations
A 20m half wave vertical fed against ground radials may give the best match much lower in frequency than expected, this is unfortunate as for example, a 1/4 wave vertical for 40m, can also double as a half wave for 20m and also a 3/4 wave for 15m. The reason is the secondary of the transformer acts as a base loading coil. The image above left (the one with type 61 cores) clearly shows how the loading coil effect occurs, there is the tightly coupled primary and lower half of the secondary, followed by several spaced out turns which creates a loading coil. The effect is more noticeable on the higher bands, such as 20m or higher.

A better alternative for single band use?
An L match is a good choice to feed the base of a 1/2 wave vertical, there is a article on making these by Martin K1FQL, link at the bottom of the page, plus my page on making QRO L-Matching units. The downside is that you could end up needing several matching units if you want to change bands, which with a simple 49:1 transformer can be achieved by swapping wires attached to a fibreglass fishing pole.

Heating/loss
Loss can be a problem as it results in the ferrite heating up, the single FT240-43 transformer shown at the top of the page overheats very quickly with a 120 Watt carrier on 7 MHz, yet the same transformer can handle the same power without the SWR changing on 14, 21 and 28 MHz. On 7 MHz the SWR starts off low before increasing dramatically after a minute or so. Tests using 2 x 240-61 cores showed no sign of heating up after several minutes of FT8 running 275 Watts on 17m, the box remained cold to the touch. Loss can be measured using a VNA by making two similar ununs and wiring them "back to back", however this test is done "cold" and characteristics will change as the core(s) heat up, this is more of an issue with type 43 and 52 ferrite, than with type 61 or 67.

The change of ferrite characteristics with temperature is called the Curie effect. In the case of type 43 ferrite, the Curie point is 130 degrees C. Type 52 is 250 degrees C, type 61 300 degrees C and type 67 450 degrees C. Type 43 cores are prone to SWR change as power levels increase, the images below show how the permeability changes significantly as the temperature rises for type 43 , type 61 and 67 have a virtually flat permeability vs temp change.

The above images are courtesy of Fair-Rite Products Corp.

Links to the data sheets for various ferrite types:

Type 43 Ferrite
Type 52 ferrite
Type 61 ferrite
Type 67 ferrite

Making an L match for end fed wires

Identifying core types
All FT240 size cores look identical. It is important to mark them when first unpacking as it is easy to mix them up. Type 31 ferrite, which is mostly used in common mode chokes, shows resistance if you touch the core with your meter probes. Type 43/52/61 and 67 appear as infinite resistance. Using an LCR meter, set to 200 KHz, shows the following inductance with a 3 turn coil on a 240 size core. Remember each time a wire passes through the core it counts as one turn.

FT240-43, 3 turns = 10uH
FT240-52, 3 turns = 3.5uH
FT240-61, 3 turns = 2uH
FT240-67, 3 turns = 1uH

The most likely used cores are type 43 and 52, which show noticeably different inductance with an LCR meter.

Conclusion
49:1 transformers are often sold as an easy solution, and while for the most part they perform well, the grounding/choking/counterpoise used can vary the results significantly. Best results require a certain amount of experimentation and in the case of a vertical for the higher HF bands, be ready for some surprises. Overall for many users, installing a half wave end wire is more convenient than either a doublet or a trap dipole.