I think using a digital signal processing would help, instead of beam-forming network and couplers

in this way the magnitude of open circuit voltage at each antenna can be sampled and added to each other with some weighted coefficient in digital manner.

(v1*exp(0*j/N))+(v2*exp(1*j/N))+(v3**exp(2*j/N))+(v4**exp(3*j/N)+...)

so the proper phasing of terminal signal can be calculated in a small board.

in your case:

V_Antenna1*1+V_Antenna2*(j) will produce +90 phase shift between port 1 and 2 and

V_Antenna1*1+V_Antenna2*(-j) will produce -90 phase shift between port 1 and 2.

all you need to do is digital multiplication, and doing it in HF band is not a hard task compared to UHF band

Hi Ahmed,

Appreciate your thoughts, thank you! 

I will use that technique at the receive side, as soon as I have my dual channel synchronous direct sampling HF receiver ready. Reception at HF is challenging due to the large dynamic range needed, but I found a good candidate (www.openhpsdr.org).

But I also want to set up a number of small (2 Watts) HF beacons with simultaneous dual circular polarization (e.g. 7.0380 MHz RHCP, 7.0385 MHz LHCP).  I thought that would be easiest with an analog combiner. Therefore the question.

Any ideas how to realize a cheap and simple digital synchronous dual channel transmitter for HF?

you can make a nice rat race coupler yourself by just using 4 coax cables correct length which are connected via some BNC or N type T piece (connector)

Thanks, Fritz, will try to make one and measure it on the NWA.

You can create using strip line in a "snake motion shape" this will be sturdy & better package. Will have reliable o/p wrt time.

Just connect a series capacitor from your RF output to one antenna, and connect a series inductor from the same RF output to the other antenna.

If the antennas are tuned so that they have a real impedance (no reactive component) of magnitude R and the reactances of the capacitor and inductor are also R, then the load on the RF output will be R - easy to match an amplifier to.  Also, the two antennas will be driven 90 degrees out of phase.

Of course, like a rat race, this only works exactly at one frequency.

If teh antennas take a differential (balanced) input like for example a center-fed approx half-wave wire dipole, then you will need a balun transformer at each antenna in addition to what I described in the previous response..

Hi Larry,

Thanks for the response. I want to feed both dipoles of the turnstile antenna with two transmitters via the network, one creating a left-hand circular polarized wave, the other creating a right hand circular polarized wave. So that will not work with your method, it is intended for use with only one transmitter for one rotation sense, I think. Or am I wrong?

So I'm looking for what is called a rat-race hybrid coupler or quadrature hybrid coupler, which has two inputs for the transmitters (tx1 and tx2) and two outputs for the dipoles. The phase difference on the dipoles is -90 degrees for tx1 and +90 degrees for tx2.

On UHF/SHF you realize such circuits in strripline, and they are narrowband. On HF these striplines are not practical and even a solution with coaxial lines the solution is bulky. So I'm looking for a design where the quarter wave lines are replaced by bifilar coils on toroids or so. Preferable a wideband solution using ferrites. Do you know of any design?

And yes, I use good quality baluns, matched for identical phase delay.

Kindest regards,

 Ben

Ben,

In that case drive the antennas from one RF output as I described, and in parallel drive them from the other RF output similarly but with the capacitor and resistor swapped.

Also the reactance of the inductors should be R * 1.618 and the reactance of the capacitors should be R / 1.618

The 1.618 is (1 + sqrt(5)) / 2

It comes from solving a quadratic and assuming the two capacitors are the same, and the two inductors are the same, and the geometric mean of the inductive reactance times the capacitive reactance should be R.

I arrived at this using a lot of symmetry arguments.  The load on the RF outputs should still be resistive although probably different from R.

I wish I had more time to double check this...

Larry

I meant to say

In that case drive the antennas from one RF output as I described, and in parallel drive them from the other RF output similarly but with the capacitor and inductor swapped.

There are two aspects to be given thought in case of ferrites.

At high frequency it is difficult to get. Rare manufacturer

What amount of power to be want to handled  by ferrite? For higher power, you will have saturation problem

Striplines are definitely bulky, but 'Bridge' made out of stripline is perfect solution. Using Bridge of stripline you can create combiner/divider as required. about 54 degree electrical length for each arm of bridge will provide you impedence match/isolation. You can create combiner, divider or phase shifter as required using Bridge

Hello  Amipara,

Thanks for the discussion. Appreciate your interest in the subject.

Can you send me a schematic of the bridge that you talk about? My email is [email protected].

How do you suggest to realize such a circuit in strip-line? Intended frequency range is from 3 to 10 MHz, so 54 degrees in free space would be between 15 meters and 5 meters, in strip-line probably 80% of that length? I have no clear view how to realize that in strip-line technology. Please clarify.

Also it would be narrow-band, preventing any frequency change.

Ferrite toroids are not really that difficult to obtain, they are used in balance/unbalance transformers and power splitters on HF (3-30 MHz), where a 1:10 frequency range is achieved for powers of 1000 Watts or more.  But you are right about the proper selection of the core for the intended frequency range and power flux density. My Tx power in my experiments will be 0.1 to 100 Watts.

Kindest regards, Ben

The easiest to make quadrature 3-30 MHz coupler can be found in the QEX Nov. 1982 isssue article by G. W. Horn (How to Design Wide-Band RF Quadrature Network).  A copy of this article can be obtained from the ARRL for $5.

Although more complicated, a quadrature 3-30 MHz coupler can be designed and made from the book, "Lumped Element Quadrature Hybrids", by  David Andrews

I think you can create different sections of rat-race with lumped element circuitry. I saw lots of lumped element power dividers, filters and impedance transformers that are low-frequency versions of high-frequency strip-line designs.

different sections of circuit with different electrical lengths are easy to create with L-C circuit. But power capability would be an issue and need more attention.