Duplexer  for High Frequency Repeater

The building of an experimental  high frequency repeater made from two Codan transceivers for SSB required implementation of a much  improved isolation between the transmit and receive antenna .
I was using two hf antennas separated by 100 m  and both antennas being end on for maximum isolation but the interferance  from insufficent separation was still too much

I decided to build a set of narrow band filters ,one for transmit  and one for receive  to minimise the interaction . since the very low frequencies precluded the use of cavity resonators, they would be a quartter wavelength at 4.8 MHz,  a  lumped inductance helical system is the only option .  I have some shareware  ( helical.exe)  that will produce the dimensions for boxed helical resonators  so  after much reading up on the subject I went to work . I decided to use coils 160mm in diameter ( bigger would be better .). and the cheapest and easier was plastic biscuit barrels  under a brand "Click Clack"  They had a substantilly fitting sealed lid by which I could attach them inside the aluminium box cavity  and thus facilitate easy removal of the whole assembly . I decided it was easier to wind these on the small metal lathe I owned. I ran the shareware and arrived and dimensions .
             

I fabricated a handle to hand wind  this gave me a little more control  doing it solo   I used 1mm dia enamelled copper wire and some plastic coated hook up wire as a spacer to hold the turns

                            

The two fabricated  aluminium boxes were  tig welded together  forming 3 chambers into which helicals were mounted  by the lid of the biscuit barrel screwed to the floor of the cavity  .  Note  I ran my cavities  upside down i.e. the hot end of the coils were in the depth of the chamber  this was an attempt to minimize capacitive effects when playing with coupling  systems etc hand capaciance ..  see I started winding the  coils 1/3 way up the biscuit barrel ..the cold end  ( earthed  and coupled will be  at the top .

                            

 

                        

 

The helical  barrel was plugged into the secured biscuit barrel  lid, the cold end of the coil   sucurely attached by a good low impenance connection to the inside of the compartment , the helical was tapped to the outside world 2 turns up fron cold end and then connected to the tracking generator via 30 ohm coax . The filter was ttuned to the wanted HF  Frequncy  eg 5190 Khz

                                      

by chopping of turns and leaving a wire spike free it the hot end  this wire could be bent close or away from the walls of the cavity  for final tune  , dont go too close to the cavity wall or anything more than 25 watts is going to make this wire corona  or arc from the high rf voltage at resonance . once I had one helical coil  I began to experiment with notches  either  side of the passband  using series capacitence in the tap to the coupling link .

                      

This passbound was acheived with a fixed mica  capacitor  and a mica compression capacitor in parallel 

 

                     

you may notice there is only one photo that shows the component connection to the tap on the coil  a 1000 pf in parallel with a 300 pf trimmer
you can over whlem the inductance of the tap turns  less capacitance allows inductive notch to appear  and too much capacitane completely over rides the inductance to show a capacitance response .

                     

 

the depth of the notch is related to the "Q" of the resonator and the degree of coupling to the resonator hence it is a balance between great filter performance  and RF loss through the filter

The next experiment involved coupling  two resonators of the same passband shape  there was no attempt to improve notch depth by having quarter wave connections between filter as you can do with VHR and UHF  full  length duplexers

                     

                      
                                             This  is two filters in series with  quite good performance 

                              

 

                   

                  
                   
                    the next  step was to go to three coupled helicals with the same bandpass shape

                    this is a respectable    performance  given the depth of the notch 90.4 dB  and passband loss just over 1 dB   

                 

                                              the    3dB  bandwidth of the notch   is approximately 30 KHz

 

               I ended up by making a high pass filter  and a low pass filter  for the transmit and receive frequencies  and ran the transmitter at 20 w  as I figured that into a full size dipole would provide a good  signal .each antenna was connected to the appropriate filter

The system connected  did enable corrct operation  and my mobile signal did no longer distort and  whirl like I was talking from the dark side of the moon ,i suspect the real problem that would show  was the temperature stability   all in all a sucessful outcome 

                                                                   back to home page