a directional coupler for HF and lowband VHF
by ralph klimek  VK3ZZC
April 2016
the motivation for yet another device for measuring impedance was dissatisfaction with my homemade SWR detector that was producing completely misleading results  above 20Mhz and was proving useless at 50Mhz.  This SWR monitor was made  with a line sampler of made from ferrite cores. Bleah.   The coupling was a very strong function of frequency and planetary alignment.  I wanted something that was determined by fixed GEOMETRY , first principals and nothing else.  This requires a true multi-axial line sampler and decent metal work.

The construction is somewhat reproducible because it uses commercial off the shelf  (COTS) metal extrusions and standard components.  By some wonderful coincidence the theoretical impedance of the main line would be within 5% of 50 OHMS. However, that came from a table of standard line section impedances that cannot take into account the disturbing effect of the sampling line and the slot. The sampling line also requires accurate termination before meaningful measurements can be acquired.  Determination of the sampling line impedance from tables is fraught with danger due to the disturbing effect of the mainline. This sampling line  would have to be calibrated somehow.  I made the deliberate choice of COPPER for the line elements.  This was chosen soley for the ability to solder the electrodes to it.  My original slotted line apparutus suffered somewhat for the inability to make good electrical connections to aluminium.

The top cover of my coupler was originally intended to be completely covered to eliminate the slot. The slot is an unwanted artifact of having to use commercial off the shelf extrusions  which allways conspire to be the wrong size !  But now I have second thoughts.  The precision of my perspex sheet insulators was sufficiently good that maybe the slot could be used as the slot in a slotted line apparutus  for UHF.  Why not ?  I do like slotted lines having made a really nice one about thirty years ago and long before played with Lecher Lines for frequency measurement in the days before I had biult my DFM. Yes, indeed. Next little project will be a inductive probe and slider arrangement. The perspex insulators will need a little slot cut for the pickup loop.  Later I did precisely this. See below.

What I really needed to calibrate my line was a vector voltmeter.  One day, I hope, the Great Cargo Gods will leave a HP vector voltmeter for me in a dumpster somewhere, stranger things have happened. But, in the meantime...

Calibrating my coupler.

It can be done without a VNA or vector voltmeter  !  Heres How !
This method absolutely astonished me, that it was possible with using  nineteen seventies  vintage equipment that had been scavanged from dumpsters and flea markets.  I had a early 70s  vintage pulse generator and a 100Mhz  scope.  The pulse generator  was easily repaired as it used only discrete components.  It has a believeably good rise time specd at 5 nanoseconds.  It used a tunnel diode pulse shaper. Remeber tunnel diodes ? Probably not as they became a forbidden technology along with the Backward Wave Oscillator. (Paranoid, delusional conspiracy theory warning !)  in the 80s.  Could I use time domain reflectometry  to calibrate my coupler ?   It is a big ask. I have to be able to observe the time of flight of light itself along the length of the coupler. It is only half a meter long, the time of flight is only about 2 nanoseconds.  That is not a lot time !
In the TDR principal, I have to launch a well formed pulse, trigger a scope, the pulses' leading edge travels to the remote end of the line, is reflected back along the line and is absorbed in a 6dB attenuator after being displayed on the scope.  Would this remarkable signal be visible/discernable  on my primitive equipment ?  It was !  If the reflected pulse was to be absorbed by the  CORRECT termination then there would be NO reflected pulse  as mandated by the Laws of Physics.  In my workshop, we obey those laws.   Thus was born my precision variable resistor.  It is nothing more than a 100 ohm carbon miniature pot mounted in a nice little brass box that was also salvaged from a dumpster at work.  All internal connections was performed with large sections of brass shim stock to minimise as much stray reactance as possible.

All I had  done was to observe the reflected pulse  dissappear at the pot was adjusted.  When satisfied, the pots resistance was measured.  Bonza, it said my main line was 52 ohms and that is close enough. To confirm I replaced it with a precision 50 ohm terminator  and the reflected pulse vanished.

Now to calibrate the sampling line. That matters just as much as the main line.  One end of the sampling line must be terminated in its impedance before we can seperate the forward and reverse energy flow components on main line.
The sampling line was calibrated  by   sending pulses down the correctly terminated  main line  and now monitoring the sampled pulse with the scope.  The sampled pulse would also now run back and forth along the sampling line. When one end is terminated in my variable resistor  we adjust it  to eliminate the reflected pulse.  It measured 85 ohms in my device.  I then made a precision terminator with a metal film 82 ohm (closest resistor in my cabinet)  in a crimped 50 ohm BNC crimp plug.
With my terminator applied, the pulse reflections vanished in the sampler.  To be sure, I now attached a little chain to it so that it does not go on an unauthorised expedition !

My sampler is now complete.  I will document the dimensions of this thing.  It uses standard extrusions for all lines. You might have trouble getting copper line for the sampler. Mine happened to be 1/8th inch  pipe. This may be hard to source but you might find that a length of silver bronze brazing rod  is just right !  The inner line was 1/2 inch copper water pipe.  The perspex spacers should be spaced in a non uniform way so that the impedance disturbances caused by it do not add coherently.

In any case, do not attempt to make one of these things unless you are able to calibrate it. There are two many variables in construction that would render a like for like reproduction  dubious.

Still needed to be done, calibrating the line to line coupling coefficients for various target Amatuer frequencies. This will make the basis of my own Vector Network Analyser  on one impossibly distant future day.

Variable resistors of 100 Ohm of this kind are now extremely difficult to locate. Nearly all of them are wirewound which is totally unsuitable for this application.  If you see one at a hamfest do snap it up. These days only 1Kohm and above are still manufactured.

Material dimensions are
U Section extrusion is 20x40x3mm
Slot cover is 20x20x1.5 mm
which are standard extrusions
Copper pipe is a length of hard drawn 1/2 inch  standard pipe.
Sampling line is 1/8 inch  hydraulic copper pipe.  Brazing rod would do just fine.
Main line placement is actually not that critical, neither is the sampling line as we will need to calibrate it anyway.
Insulators are made from perspex scrap.

Update ! Newflash ! Stop the Tinternet !
I have modified my coupler.  You can see in the pictures, that the input and output ports are rather clumsy. The connections are made with only a bit of thin braid.  The impedance of this bit of twirly  braid made this ineffective as a slotted line apparatus.   I found a longer length of copper tube and have extended it to the ends of the housing, where now decent RF connections have been made to the copper conductor and the BNS centre pins with the medium of brass shim stock.  There is affixed to the housing a springy brass sheet keeper for the proble slider assembly.  It makes a satisfying click  when the slider is stowed in the brass click.  I have also made a delay line. This line has the time delay of the probe line.  This is to equilize the delay so that I can use it my recently aquired HP vector voltmeter.  I really should have a line stretcher for this . I have only once seen a pulse stretcher for sale at a hamfest, dilly-dallied, and lost it .  Someone smarter than me snapped it up ! Bolloks.
The pulse delay line is also attached with a little chain.  I have found it good to chain little detachable accessories as they allways grow feet and march off into the world, without even the courtesy of saying goodbye.
The coax has been  cut to match the propagation delay of the sample line.  We first measure the velocity factor of the coax  and cut to length. the vf is measure by finding the 1/4 resonance with a grid dip oscillator with a long length of the same coax from the spool.  The longer the measurement, the greater precision.  The finished assembly is also chained down to stop it wandering.  75 ohm cable was used, this is a moderately good match to the 82 ohms of the sample line. It is a good as it gets,  they dont make 82 ohm coax !  

the VK3ZZC  Precision (TM)   directional coupler
completed coupler with variable resistorhome made 82 ohm terminator in BNC crimp plug. complete with chain to prevent it going walk-aboutcompleted coupler with variable resistor
The brass box was beautifully made by the long gone  Monash Physics Metal shop for a  forgotten purpose and sadly discarded but now finds a second life here.super-duper variable resistor for HF/VHFhang on to your 100Ohm carbon pots. they are getting extremely rare.
Above images shows a  pickup slider assembly.  Does it work. Yes. Does it work well ?  Not really . The pickup loop does not pick up a lot, loop is too shallow, even at 500Mhz.  To give meaningfull   slotted line measurement at UHF will require me to re-engineer the metal work  that connects the BNC jacks to the main line.  The current connections do look like a significant amount of series inductive reactance at 500Mhz  which effectively nullify any meaningfull  measurement  by the slotted line  method.   I should not be so dissapointed as this sampling line was never intended for VHF/UHF. I was sorely tempted by the natural slot created by these standard extrusions.

New improved ! it is November 2020  allready.  We have made some needed modifications and improvements
slider is in its stowed positionhere we see the coax delay line equalizer. This is used for VNA measurement with a vector voltmetera better UHF line to BNC connector  transistion
IMG_0861-coupler.jpgLeft. The brass slider stowage clip

Making this bent piece of brass was difficult. The three differant bends were very hard to make without the benefit of a proper finger panbrake. After this, I shall be buying a proper bender from Hare and Forbes. This would be easy with a brake press.


page modified  on Thu Nov  5 18:05:23 AEDT 2020      . Improvements added