4-way Splitter EME
The below described Power splitter/combiner is build after consulting the website of PA5MS/PE1OGF. There more details can be found, but I am not too sure on the given dimensions in the found drawing. The distance between the center N-chassis part and the two N-chassis parts at the end of the splitter/combiner are 525 mm, but in the drawing something unclear like 620mm is shown. Be careful before start drilling and think twice is my advice!
The original text of PA5MS/PE1OGF:
Power splitters are used to combine 2 or more antennas on 1 feedline. The impedance of each antenna will be transformed to the required impedance at the feed point. The impedance of the transformer can be calculated with the following formula : Z0=VzoutxZin Where Z0 = impedance of the transformer, Zout = impedance of the feedline and Zin = Impedance of the antennas. For example if you want to combine two 50 ohm antennas using a 1/4 wave power splitter you will have 50:2 = 25 ohm at Zin, the impedance of the transformer will be the square of 50x25 = 35.35 ohm. If you want to use a 1/2 wave power splitter you will need 100 ohm from each antenna at the feed point to get 50 ohm when you put them together. To get this 100 ohm you will need a transformer with the following impedance : Z0 = square of 100 x 50 = 70.71 ohm. Below there is a drawing of how to construct a 1/2 wave power splitter for 144MHz.

525 mm
550 mm
1075 mm
1100 mm
25 mm
 The dimensions in this table are according the mechanical construction I did (PA3EXV). These do not match the original description, partly because I used 25 mm to the ends instead of the original 30 mm, but also some unclarities are taken out from the original.

PE1OGF continues as follows:

This above drawing is not to scale. The outer tube is a 30x30mm aluminum square tube 2mm thick. This gives an inner diameter of 26x26mm. The impedance of the transformer depends on the ratio between the inside diameter of the outer conductor (D) and the outer diameter of the inner conductor (d). Below is a table of often used impedance's and corresponding D/d values. The first colom is the number of antennas you want to transform to 1 feed line. For example 4:1 means that you use 4 antennas each 50ohm and you want to transform the total impedance on the feed point back to 50ohm. This can be done by a 1/4 wave splitter with the feedpoint at the bottom side and 4 connectors at the top or a 1/2 wave splitter with the feed point at the center and 2 connectors on each side (1/4 wave) from the feed point. One important last note is that the inner conductor has to have a length of 530mm for a 1/4 wave and 1050mm for a 1/2 wave splitter. This inner conductor has to be exactly in the center of the square outer conductor to be sure of the right impedance. For inner conductor I used copper water pipe from the plumber. This water pipe is available in different diameters and rather cheap. For connecting the N-Chassis to the pipe I used 2mm silver plated copper wire.

PA5MS/PE1OGF also gives information on how to build a splitter/combiner for other than 4-way 1/2 lambda. In the below table you can find different solutions on how to transform the impedances back to the 50 Ohm input:

Number of
1/4 wave
1/4 wave
1/4 wave
1/4 wave
1/2 wave
1/2 wave
1/2 wave
1/2 wave
Z ohm
Z ohm

This is the start of the mechanics for the 4-way splitter/combiner. As in the technical description already mentioned, the outer conductor is made from square aluminum profile 30mm outside, 26mm inside. This is in combination with the 12mm copper inner conductor the needed impedance.
I started by fixing two of the N-chassis parts at one of the ends of the 1/2 lambda long pipe at 25mm distance from the end.

After the N-chassis parts have been fixed by stainless steel self-tapering screws, the two inner conductors are connected together by means of a piece of copper wire. The stainless steel screws are reduced in length to maintain the impedance as much as possible. This is reduction in length is of coarse not needed when you buy the correct length of screws with 3.5mm diameter...

In the picture to the left, you can see the inner conductor is prepared partly. The center N-chassis part is also connected to the inner by means of a short piece of copper wire. This copper wire is soldered in a hole in the inner conductor and pushed inside the drilled hole until it touches the opposite inner wall. This is done to prevent movement while soldering the center pin of the N-chassis part later, because the copper wire is then fixed between the center pin and the inner conductor of the splitter.

In the picture above you can also see the opening that is made in one of the side walls of the outer conductor. This is done to be able to solder the inner conductor to the center N-chassis part after the inner conductor is placed.

Now, after making small cuts at the very end of the inner conductor for alignment purposes, the splitter/combiner can be assembled.
The 12mm copper tube is inserted and soldered from the inside out to the prior mounted N-chassis part. Bare in mind that only these two N-chassis parts are mounted at this stage.
After this the center N-chassis part can be installed. Before this job can be done, carefully cut the copper wire to the correct length so it slides in the center pin, but does not force the inner conductor of the splitter of center. That might disturb the impedance and give later no good VSWR.

The picture here explains may be a bit more what is meant by cutting the small copper wire to the correct length. Usually N-chassis parts have an opening in the back of the center pin. The copper wire that connects the inner conductor to the N-chassis part slides in this opening while mounting the N-chassis part. When using the correct length of copper wire to the inner conductor of the splitter, this does not give mechanical movement to it and soldering is easy through the opening.

Last job is to mount and connect the remaining two N-chassis part at the other end. Before soldering to the inner conductor of the splitter, the cross connection to the two center pins is made by copper wire. This copper wire is also guided to two small half round openings at the end of the inner conductor of the splitter.
After that, the cross copper-wire is soldered as the first two N-chassis parts from the inner of the 12mm pipe.

Now the splitter is ready and can be closed by two standard plastic caps. These are hammered in place.

The here described splitter/combiner is measured and found performing very well. The 4 outputs have
- 6.1 dB as expected and also the 144MHz band is fully covered. All 4 outputs match within 0.1dB. After these measurements shown good performance, the splitter is painted to resist weather influences.

Below are some screen shots of the tracking-generator in combination to a Spectrum-Analyzer that is used during the measurements.

Reference level prior to measurement is
- 0.11 dB.

Here one of the 4 outputs is measured on insertion loss.
The measured value is - 6.29 dB. This is a difference of 6.18 dB compared to the reference level. 6 dB is expected because this is 1/4 of the input power. The rest (0.18dB) is losses in the splitter and partly inaccuracy of the measurement.

This picture shows the return loss when all four outputs are terminated by good 50 Ohm dummyloads.
The Spectrum-Analyser screen shows a frequency of 144.370 MHz were the marker is at that moment. The span during the measurement was 50MHz, which means that every division represents 5MHz.
When more narrow span is used, the resonance frequency is found at 144.200 MHz. Also coverage of the total 144MHz band is found to better than 49dB return loss.

The below picture shows the measurement setup. Here the splitter is already pianted to protect it against weather influences. During this paintjob the N-chassis parts were carefully covered in tape and this prevents paint to come into it. A few layers of paint were applied, especially were the N-chassis parts are mounted to the square outer tube and the cover of the opening through were the middle N-chassis is soldered.

In the picture the VSWR bridge connected to the Spectrum-Analyser (with tracking option) is visible, as well as 2 of dummy loads at the bottom of the splitter. The VSWR-bridge is connected to the middle N-chassis part. From this setup the above picture is taken.