The antenna itself is a turnstile, what really is just two inverted vees perpendicular to each other on a single mast, driven 90° out of phase.
The antenna has four arms, let's call them A, B, C, and D. A-C and B-D are the two inverted vees. The arms should be arranged so if you look at the antenna from above, and read A-B-C-D, you go counter clockwise (see on rajz-antenna.jpg). This direction is also called "right handed", because if you grab the mast with your RIGHT hand, your thumb will point up, the other fingers will point to the direction of the rotation.
The antenna mast should be 5.5m high (anywhere between 5-6m is fine). The length of the antenna arms are 13,19m PLUS the 3m non-conducting thread to tie the arms to the ground. Done properly, the antenna wire ends should be about 1m from the ground. This arrangement gives near 50 ohms impedance over reasonable grounds (no deserts or salt water, sorry :) ). According to my experiments, no tuning is necessary over average / good ground, SWR were less that 1.5 every time.
The two inverted vees are driven 90° out of phase to produce circular polarization. The direction (left hand or right hand) is decided by the direction of the phase shift. if A-C is ahead of B-D, the direction is RIGHT, otherwise it's LEFT.
There are two auxilliary circuits ("rajz-aramkorok.jpg"). One is a switched bias tee sending or not sending DC power via the antenna coax (found at the left side of the drawing). The other circuit is a 90° hybrid implemented with a resonant circuit. This is called a "twisted wire hybrid", it features two capacitors and an inductor / transformer. This circuit also has relays that switch the output so the direction of the circular polarization can be controlled (found at the right side of the drawing). When the relays are NOT energized, the B-D vee is ahead of A-C, therefore the antenna is LEFT handed, if they are switched on, the antenna is RIGHT handed.
Circuit 1 uses some generic PNP and NPN transistors to combine the DPDT switch and the key input "signals" and send (or not send) power up the antenna coax. C1 is non-critical, it shouldn't have much voltage across it, and since the antenna shouldn't be used over a 100 watts, there shouldn't be large currents either. Jellybean ceramic types suffice. Same goes with C2. R3 and R4 are bleeding resistors, also non-critical. D1 is there to prevent inductive kickback from the relay coils and the cable.
Circut 2 separates the DC and RF with C3-C4 and L2. These are the same types as C1-C2 and L1 in Circuit 1. The DC signal turns the relay(s) on or off. You can use two SPDT or a DPDT relay, as long as coil and contact voltages and currents are OK. R5 is also a bleeding resistor, and is also non-critical. R6 should be a beefy 50 ohm rf resistor. Normally, and in theory there is no power on it, but if the antenna is not tuned absolutely perfectly (so 100% of the time :) ), some of the power will be lost to this resistor. If the antenna wires break, a significant portion of your 100W can go here, so better size this to at least 10W (mine is build for 20W just to err on the safe side). C5 and C6 should be quality hihg voltage RF capacitors, such as silvered mica or good plastic foil (WIMA for example). T1 and T2 are common mode chokes, should be wound with 50 ohm coax, and have sufficient choking resistance. A solution can be seen on balun.jpg (attached).
L3, C6 and C5 are critical. Ideally you should use an oscilloscope to check the phase shift and balance. A much time-consuming solution would be to use a whimpy little resistor for R6, dummy loads for the antenna, and check R6 for heating (or smoke :) ), while adjusting the turns & spacing on L3.
Circuit 1 should be placed close to the radio, and plugged into a 11-15V power supply, the appropriate connectors hooked up to the radio's tx/rx antenna port so it's between the radio and the antenna (like any other bias tee). It can be switched by hand, using the DPDT switch. There is also a "key" input that sould be connected to the radio's key/ptt out port.
Circuit 1 is basically a bias tee, that has two controls: the radio's PTT and the manual DPDT switch. It in essence operates like an XOR gate: the DPDT switch decides wether it's sending the power to circuit 2 when the PTT is pressed, or when it's released.
This is necessary for two CP stations in QSO because during NVIS propagation the reflected wave changes direction, right circular polarized waves become left CP and vica versa. If a station is receiving in RHCP, the other must transmit in LHCP, or the signals will be greatly attenuated. (The attenuation depends on the propagation and the antennas, during experiments ~20dB attenuation could often be observed.)
If both wish to transmit LEFT handed, both should receive RIGHT handed, so they MUST switch directions on receive and transmit. Circuit 1 does this. You can transmit RIGHT and receive LEFT of course. The other station must do the same of course.
This looks kinda complicated compared to conventional "antenna systems" (I mean wires thrown up trees :) ), so better be some advantages right? :) There are. Any wave entering the ionosphere from below produces two CP waves, an "ordinary" (O) and "extraordinary" (X). These waves propagate somewhat differently. When NVIS is just kicking in or going away, often only the X waves are reflected back to earth.
Transmitting RHCP and listening to LHCP uses the "O" path, otherwise it's the "X" path.
If someone is using linear polarization (LP), much of their energy is lost: first the O wave will go to space, and second, the X wave being circularly polarized, the other (LP) station will lose about 3db again. Listening to LP stations with CP antennas at the "happy hour" (when there is only X propagation) gives an instant 3db plus. Not much, but it's there.
If the other station is also CP, it can transmit so you only use the X path, when it's the only one available, avoiding the X/O splitting loss and polarity mismatch loss entirely. One should expect about 6dB improvement. Still not much, but sometimes it makes a difference.
What happens when both X and O paths are viable? Well, they often have different properties in terms of attenuation and fading, so two CP stations can choose freely which path to use, and often actually makes sense to change paths. This can also be useful with CP-LP QSOs.
Distant QRM could arrive as either the X or the O wave, and therefore could be greatly attenuated by switching to the opposite path. (I did this indeed in practice. Only once, but still. :) )
Noise levels can be different on the two paths, this is something we experienced during the experiments.
So wrapping it up:
- CP stations can receive LP stations better during certain conditions (a minority of the time, but still).
- CP stations can reach other CP stations just plain better.
- Sometimes distant QRM can be attenuated.
- If both X and O paths are viable, the best one can be chosen.
"rajz-antenna.jpg" is a sketch that shows the sizes of one arm.
"rajz-aramkorok.jpg" is a sketch showing the switching and hybrid circuits.