Assembly & Measurement of a 3dB Splitter


3db Splitter

Lately I needed a 3dB splitter for a HF (1…30MHz) measurement project. Instead of buying me a splitter, I decided to build it on my own. After measuring the performance I am satisfied with the result…

Introduction

The circuit of this splitter is nothing new. It is known as a 0º – 180º Hybrid. The big advantage of this hybrid is that it can be used to force the equal current in two different loads. This means that the incoming signal will always be splitted equally, independent of the termination of it’s ports.

On the other hand, these hybrids can also be used to combine two signals without affecting the input impedance of the other port. In this case they are called “combiner”.

The three most popular HAM Radio applications are:

  • Distribution of a HF signal on two receivers (splitter)
  • Measuring intercept point of 3rd order (combiner)
  • Combining signals of receiving antennas, i.e. phased beverages (combiner)

3db splitter circuit diagramm

Preparation

For the ferrite core I choose Amidon 61 material which is generally a good choice for broadband transformers between 1 – 30MHz due to its low losses. The permeability of 63 material is specified with µ=125.

The following data has been taken out of Amidon’s datasheet:

Material

Permeability

Resonance

Wide Band

AL

61

125

0.2 – 10MHz

10-200MHz

75

The following ferrite cores could be used:

Ferrite core Outer Diameter Inner Diameter
FT-50-61 12.7mm 7.14mm
FT-82-61 21.0mm 13.1mm
FT-114-61 29.0mm 19.05mm

Since I had already a small case with 3 BNC plugs available I choose FT-82-61.

Two ferrite cores are necessary:

  1. A trifilar wounded core (7-10 rounds)
  2. A bifilar wounded core (7-10 rounds)

For 10 rounds on a FT-82-61 19cm of wire are needed. Including a few spare centimeters, I prepared 5 x 25cm of magnet wire (don’t use blank wire) with 1mm diameter.

Since we are dealing with HF, a metal film resistor (100Ω) should be used across Port 2 and Port 3. Metal film resistors have better HF performances than standard carbon film resistors.

Assembly

The wound cores should more or less look like in the picture below. The inductance was calculated on the paper with a value of 6,9 µH. The measured value was 6,53µH and therefore considered as ok.

AADE Measuring 3dB Splitter Transformer

The two ferrite cores were mounted in a tinplate housing with three BNC plugs. The solder joint between the windings of T1 was covered with a heat shrink tube in order to avoid ground closures. The 100Ω metal film resistor was soldered across the plugs of port 2 and port 3. The picture below shows the assembled splitter with already connected coaxial cables.

3dB Splitter with open Case

Measurement setup

The following measurements were made with a Vector Network Analyser (TAPR VNA) and should give a good impression about the performance of the hybrid. Please note that my “poor mans VNA” only is capable of measuring S11 and S21 at the same time.

A fixture calibration was made to exclude the cables and connectors of the measurements. The picture below shows the setup.

Measurement Setup with TAPR VNA for 3dB Splitter

Measuring Signal Distribution port 2

In this measurement the signal distribution between Port 1 and Port 2 was be analysed. Please note that Port 3 was terminated with 50 Ohm.

Schematic 3dB Splitter Measurement 1

TAPR VNA Result of first measurement on 3dB Splitter

It can be seen that the attenuation (S21 green) is almost continuously 3db. At 50 MHz it drops down to 4,2dB. The impedance matching (S11 red) is acceptable up to almost 50MHz. The phase (S21 blue) is constant up to 25MHz with about 10° and ending up at 34° at 50MHz.

Download the measurement result in high quality: 3dB Splitter – Measurement1

Measuring Signal Distribution port 3

In this measurement the signal distribution and the impedance matching of port1 were analysed.

Comparing the plot and the values of the markers with the ones from ‘Measuring Signal Distribution port 2 it can be seen that both ports behave symmetrically. Since also the phase is symmetrical, the relative phase difference between port 2 and port 3 is 0º.

Schematic 3dB Splitter Measurement 2

TAPR VNA Result of second measurement

Download the measurement result in high quality: 3dB Splitter – Measurement2

Measuring Impedance Matching of port 3

In this measurement the impedance matching of port 3 was analysed.

Schematic 3dB Splitter Measurement 3

TAPR VNA Result of third measurementDownload the measurement result in high quality: 3dB Splitter – Measurement3

Measuring Signal Distribution with mismatch

In this measurement the signal distribution was analysed with a mismatch on port 2 (open).

Schematic 3dB Splitter Measurement 4

TAPR VNA Result of forth measurement

It can be seen that the attenuation (S21 green) stays constantly at 3db. This proofs that the signal is equally shared between port 2 and port 3, independent of their impedance. Note that the mismatch is transformed into the impedance of port 1 (S11 red). The phase (S21 blue) isn’t affected.

Download the measurement result in high quality: 3dB Splitter – Measurement4

Measuring Port Isolation

With this setup the isolation between port 2 and port 3 was measured in order to see the dependencies between these two ports.

Schematic 3dB Splitter Measurement 5

TAPR VNA Result of fifth measurementIt can be seen that the isolation between Port 2 and Port 3 (S21 green) is very good up to 30MHz and acceptable up to 50MHz.

Download the measurement result in high quality: 3dB Splitter – Measurement5

Conclusion

In this paper a 0º – 180º hybrid, known as ‘splitter’ or ‘combiner’ was build and analysed. The performed measurements show a very good performance in the desired frequency range between 1…30MHz. The hybrid could be used on frequencies up to 50MHz with some minor degradation in performance.

The two ferrite cores were mounted in a tinplate housing with three BNC plugs. The solder joint between the windings of T1 was covered with a heat shrink tube in order to avoid ground closures. The 100Ω metal film resistor was soldered across the plugs of port 2 and port 3. The picture below shows the assembled splitter with already connected coaxial cables.
About Tobias (DH1TW)

Self-confessed Starbucks addict. Loves to travel around the globe. Enjoys the technical preparations of Amateur-Radio contests as much as the contests themselves. Engineer by nature. Entrepreneur. For more, follow him @DH1TW

Comments

  1. dan, w5xz says:

    very nice work.

    it is possible to use a single core. i used a ‘binocular’ core, 43 material, and it works
    very well to share the beverages between 80 and 160 m at W5WMU. for reception only, of
    course..amidon BN 43-202 i think…w8diz has them, very inexpensive.

    see:

    http://michaelgellis.tripod.com/magict.html

    the last item; with a 10 turn primary, and 7 turn secondaries, #30 wire. it works f.b. both on
    my miniVNA and in the ‘real world’ of multi-tx contesting. for 50 ohm systems, of course, the resistor should be 25 ohms. our bev. feedlines in La. are RG-6, 75ohms.

    73 from texas; keep up the good work. w5xz, dan

  2. werner, DK8PD says:

    Hallo Tobias,
    der Artikel ist gut. Leider enthält er auch einen Sprachfehler:
    wounded = verwundet. Richtig muss es heissen ‘wound’ (siehe http://dict.leo.org)

    While the splitter is adequate for most amateur radio applications up to 30 MHz there is room for improvement. Doug Smith, KF6DX gave some thoughts and measurements for high precision splitters in an appendix to his QEX article ‘In Search of New Receiver Performance Paradigms, Part 3′ (QEX March/April 2007, pages 28-34). He used binocular cores allowing small dimensions which is beneficial for the performance at the upper part of the frequency range.

    73 from Monheim. keep on experimenting, werner

    • Hey Werner,
      Thanks for the Information. Sounds good – I always enjoy reading articles from Doug. I’ll dig search for it in the QEX archive.
      Thanks for finding the typo. I already corrected it.
      73, Tobias

  3. Hi Tobias,

    thank you for sharing this nice article ! Already built one as described with two small FT-50-61 for my 2 JRC receiver, just great ! But I have few question:

    - Now I am doing some test measurement with another FT-50-61 toroid with same winding, from the tapping point I connect a 25ohm resistor to gnd and measure it with a network analyzer, wooww I get a nice 20dB return loss …

    - from where you start to calculate the input tri-filar winding ? I checked the toroid data and found matching the real measurement ( each singular winding is about 6,2 uH), and then when coupling them I measure = 6,2 – 24 – 57 uH.

    - but why those figure with the desired band ? 2 -30 MHz ?

    Thank you for any usefull info !

    • Hi Pietro, just measure one winding. So your 6,2 uH seems absolutely fine to me! The point of the measurements was to check the characteristics of the Splitter… just out of curiosity

Speak Your Mind

*