OK0MF – 6m duplexer development and construction

…This is a great band if you like having Mother Nature pull your chain…

For the construction of the six-meter repeater OK0MF I got the task to solve the antenna filter / duplexer. We had not a duplexer for the 50MHz band and its acquisition would be considerably more expensive than the whole repeater. In addition, the frequency offset for this particular six-meter band is only 600kHz and it was necessary an isolation of 80 dB which is quite cumbersome. On the Internet I found the construction of an amateur duplexer using an helical coaxial cable 1 “5/8 diameter. It had, however, two problems: get the proper coaxial and also the dimensions of the duplexer was not negligible. So this was not the right way to solve the task. After long consideration I decided to make the filtering using helical cavities. This desition was based on dimensions and parameters as well. In the initial consideration I thought even the use of a crystal filter as a band-pass in the Rx, but at the minimum required bandwidth of 30kHz needs the use of 3-5 sets of crystals. At the price per set about 70,-USD and the uncertainty of the result, we rejected this solution.

The basic requirements of the Repeater were:

  • The use of two antnnas, one for RX and another for TX.
  • Antennas will be ~ 23 meters apart. At this distance the free-space loss wil be around 34 db that will be taken in consideration for the design of the filters.
  • On the RX side we will include 1x band-pass filter (tunned at the RX frequency) and 2 notch filters (tunned at the TX frequency)
  • The TX side will contain 3 notch filters (tunned at the RX frequency)
  • It was also necessary to reduce the interference of one CB radio an FM radiostation (87,6MHz / 1kW), OK0MF.

As the material for making the cavities I chose a copper drainpipe with a length of 150 mm and a diameter of 80 mm. The wall thickness is 0.5 mm. For the coils I used a Cu wire ∅ 3.4 mm diameter (2.8 mm) .The trimmers to fine tuning were the Johanson 5202 type of 0,8-10pF. BNC connectors were used to interconnect the cavities. Caps are soldered to cavities and sealed with copper foil. The inner self-supporting coil have been fixed with a plastic melt from the gun, but it did prove not a good solution. Thermoplastic has a high dielectric constant and with an output RF power of approximately 20W it melted at the end of the coil. The coil is actually an auto-transformer, which has the power connected at 0,5 turn, so on the 10th turn the coil the voltage is big deal. In the final solution, I fixed the coils with bare PCB material and glued with epoxy resin.

RX section:

Because no silver plating is better than a bad silver plating, I decided to just polished the inner brass and cover it with lacquer painting (5 layers). At a frequency of 52MHz , the skin effect in Cu is ≈9μm, which is relatively low and silver is not even necessary.

The set of filters for the receiver has fulfill its task to sufficiently attenuate the unwanted frequencies. Following are the signals we needed to deal with and their levels:

  • CB repeater Faltus 27,255MHz (4W) at 52m away puts a level of approx + 0dBm at the Rx antenna connector
  • A commercial FM station (87.6 /1kW) at 44m away puts a level of approx + 16dBm at the Rx antenna connector
  • Our own TX OK0MF (10W to the ant.) at 23m away puts a level of approx + 10dBm at the Rx antenna connector

Own transmitter is the closest frequency to the RX frequency with an strong signal that would pass the input filters, clogging up the input preamplifier, which has its own gain ~ 20 db. The CB repeater transmitter and FM radio “Black Mountain” are already far enough. To suppress the CB frequency it was sufficient to use a band-pass cavity. The rest of the RX filter is composed of a two cavities as notch filters tunned at the TX frequency and a ¼ λ stub tunned at the FM station frequency.

“Path loss” ie attenuation between the two antennas is given by:

L = 20\ \log_{10}\left(\frac{4\pi d}{\lambda}\right)     [m, dB]

Several images of structures Band-Pass & Notch helical cavities:





vyroba dutin

rezani_dutin_2 rezani_dutin fixace_civky

The resulted parameters for the RX filter are as follows:

  • Average loss of the Band-pass cavity is 0.3 dB
  • Average loss of the whole RX filter at 51,370MHz is: 1.8 dB
  • Attenuation at 3.7 MHz ≈ -46 dB
  • Attenuation at 14 MHz ≈ -32 dB
  • Attenuation at 27.255 MHz ≈ -22 dB
  • Attenuation at 87.6 MHz ≈ -50 dB
  • Attenuation on OK0MF TX (51,970MHz) is -38 dB
  • RX sensitivity with the filter is 0.35 μ V / -116dBm (10dB SINAD)

See measurements:

reject_TX B-P_char Reject_FM

TX_prototyp1 RX_prototyp1 RTX_prototyp1

Chrakteristika_2m Chrakteristika_70cm Chrakteristika_HF

TX_mereni RX_mereni

What’s terms of mechanical strength and unwanted first test was the collapse of one of the Notch cavities from the table on the floors below it (80 cm). “Damages” are as follows:

Parameters changed only cavity that has suffered a drop, transfer the parameter change to take effect. Which is not to say the version with coaxial cables that are prone BYD only a small change in the position of a whole must be well fixed and usually tuned to the place of final use.


In this figure is clearly visible the change of parameters of the cavity Notch, that is shifted by ~ 220kHz. This chage affected the whole set in 13 db / 29.5% (from -44 db to -31 db)

  1. Band-pass frequency frequency
  2. Notch for TX of the repeater
  3. Detuned Notch cavities compared to item 2

TX section:

In the section of the transmitter it was used a triple Helical Notch filter tunned at the RX (51,370MHz). A maximum attenuation of -65 db was obtained. The filter is there to avoid from raising the noise floor of the RX because of the noise generated by the transmitter, produced by the oscillator and amplified by the PA.

The extarnal noise that “overwhelms” the receiver is called Desensitization noise

Desensitization noise should be about 10 db below the threshold of sensitivity of the receiver so that it does not interfere.

Theoretically the RX Noise Floor is given by :

[-174 + 10 * log B + F], where B is the bandwidth in Hz (16kHz) and F is the Noise Figure of the RX amplifier which is about 2 dB with the present Low Noise Transistor technology.

So -174+10log16000 +2=-130dBm.

In the filtering the cavities are joined by several kinds of jumpers with lengths of ¼ λ. They were made of coaxial cables of different impedances. The relation between frequency and wavelength is give by:

λ = c / f    (c is the speed of light)

Thus, we calculate the wavelength for ¼ λ and multiplying the shortening coefficient of coaxial cable.

RG-58U with shortening factor 0.659 and an impedance of 50Ω.

RG-62 AU with shortening factor of 0.84 and an impedance of 93Ω

H155 with shortening factor of 0.81 and an impedance of 50Ω

¼ λ jumpers RX (51.370 MHz)

  • RG-58U – 0.962 m
  • RG-62AU – 1.226 m
  • H155 – 1.182 m

¼ λ jumper TX (51.970 MHz)

  • RG-58U – 0.951 m
  • RG-62AU – 1.212 m
  • H155 – 1.169 m


The design and the construction of the Helical resonators and coupling them came to about 220 hours during 3 months.


I simply made a rig that really works, it is tough and it has good characteristics … and there is not a microprocessor 🙂


I really wish to thank my friend Mr. Carlos Hernandez who helped me providing the theoretical knowledge without the project would not be successful.

Leave a Reply

Your email address will not be published. Required fields are marked *


This site uses Akismet to reduce spam. Learn how your comment data is processed.