Last edited 01/04/2003.
Diversity (is the spice of life)
When you look at those neat diagrams in the antenna books about how waves propagate through the ionosphere, it's tempting to think that there is just one path in operation at any one time between two points on the earth. It also always looks like the ionosphere is just like a mirror, with a smooth flat surface.
Neither of these suppositions is correct. In the HF region, communications that involve ionospheric refraction are frequently propagated by several paths, all of which vary in effectiveness continuously. Also the ionosphere is frequently turbulent, not at all like a mirror.
All of these effects mean that signals tend to vary in strength, often by 20dB or more. If the signal is strong, the receiver's automatic gain control may take care of the fading (although not the distortion effects). However, if the signal is weak, you will simply loose copy for a few syllables (or Morse characters).
Space and Frequency Diversity
To combat fading, there are two commonly used approaches:
Frequency diversity relies on the fact that the fading is different at different frequencies. It is sometime said that it is not correlated. Hence when there is a fade at one frequency, there may not be a fade at another. To make use of this, you simply transmit your signal on two frequencies, perhaps 100KHz apart. At the receiving end, a circuit measures the signal-to-noise ratio in two receivers and automatically selects which is best at any instant in time. This works well but... it is rather inefficient to have the same information transmitted on two frequencies.
Space diversity relies on the fact that fading is different at different points on the earth, so two aerials a few wavelengths apart will have uncorrelated fading. Again, two receivers are needed and some way of selecting the best S/N. A simpler approach is just to combine the output from each receiver. This is not as good (presumably by 3dB?) but it still helps. By the way, even fairly close aerials will show some degree of uncorrelated fading so with, say, stacked beams on 20m its a good idea to feed them separately into the shack so that you can combine them or use them separately. In this case, the difference in performance is probably mostly due to the different take-off angles from beams at different heights.
So, is there a third method? You bet! It is called polarisation diversity.
A plane polarised transmission does not come back polarised in the same way. In general the polarisation is elliptical (linear and circular polarisation are just two special cases of elliptical polarisation. So how can you make best use of the returning wave?
Let us assume that you have two aerials with different polarisations. They might be plane polarised 90 degrees apart (vertical and horizontal) or circular polarised in opposite senses (left-hand and right-hand). Polarisation diversity gives useful improvements but maybe you don't have space for two aerials. I wondered if it could be done with one aerial - oh and to make it easier let's use one receiver too!
Getting a quart into a pint pot
This sounds daft, but here is the G3CWI solution.
First the aerial. What you need is an aerial that can be fed in two places to be either vertically or horizontally polarised. Choices might be a quadloop fed in a horizontal or vertical side (could be a quad beam of course) or perhaps an inverted L aerial with a half-wave top and quarterwave vertical. Feed this either at the base or at the far end for two polarisations.
Now for the clever bits. To do the polarisation switching, use some pin diodes. But don't just use manual switching, switch them at high speed (say 10-20KHz) using a squarewave source. Then, at the output of your receiver, use the same squarewave source to separate out the receiver audio from each aerial. Bingo, there it is with a separate channel for each polarisation.
Another method could be to have two dipoles at right angles in the horizontal or vertical plane and use the switching method that was very popular with the satellite crowd a while back to give Right-hand circular, Lefthand circular and two linear polarisations. Maybe this could even be done with loaded aerials?
Now, let's think further. Could this trick be done with the transmit path? Well maybe. But the pulses would have to be very carefully shaped to avoid a wide signal. Then with a synchronised diversity receiver at the far end (with some sort of phase slipping circuit to allow for varying path lengths) even more gains may be possible. However, I have a niggling feeling that there is something I am missing that will stop the transmit method working. I suspect that to work effectively it would need a feedback system so that the RX end could inform the TX end of the s/n in real-time.
It's just an idea. Will it work? Can you do it?
Post Script 14-10-2000
You know how it is when an idea grabs you. I got to thinking some more about this. A classic method of getting circular polarisation (often used by the satellite/space enthusiasts) is two dipoles at right angles fed 90 degrees out of phase. I modelled this with EZNEC and here is the result.
This is two dipoles in free space. At 0 degrees the array is equally responsive to horizontal or vertical components (circular polarisation).
You can download the EZNEC file here.
Now let's look at that in the vertical plane over a real ground.
Not so impressive. The aerial is equally responsive to vertical and horizontal polarisation at a wave angle of 15 degrees. It's not bad below that but poor above it.
At 14 degrees elevation here is an azimuth plot (above). Fairly circular for about 25 degrees either side of the boresight.
Here is the file.
I started to consider a practical implementation.
Take a tower and run four dipoles off it at 45 degrees to the vertical and 45 degrees to each other and you have a classic sloper array. This could also be a good antifading array. Two opposite dipole should give low angle antifading characteristics and two adjacent dipoles should give high angle antifading . Phase the dipoles pairs for +/-90 degree shifts to give RHCP and LHCP.
I will model this another day.
I have been trying to find some information on polarisation change characteristics so that I can run a simulation of these aerials to see what gains I would get, but as yet I have been unable to find anything. I will trawl the professional literature.
Article found here.
Comments Received - thanks!
-----Original Message----- From: Darrol Draper <firstname.lastname@example.org> To: email@example.com <firstname.lastname@example.org> Date: 29 October 2000 01:25 Subject: Diversity reception
>Hi, >In theory, I suppose you could switch a receiver >rapidly between the two polarizations, so you would >not need two receiver chains. Someday, when HF >receivers are implemented as wideband chains feeding >into 16 or 24 bit A to D converters, with a DSP doiing >the filtering, this could work. In fact, that might be the >easiest way to offload the diversity reception to semiconductors. > >Actually, with the DSP, you can do better than Switched >Diversity. The DSP can perform the calculations to remove >the multipath effect which damages HF sky-path signals. >That is, the DSP can find the exactly optimal phase (time) >and amplitude weights on the two polarization signals. >As I recall, switched polarization on multipath gives an >average 7 dB gain, roughly, and with full matched phase >and amplitude you can gachieve an average around 9 dB >improvement. > >But in the meantime, unless you have the [$ 15K ?] >for a Pentek brand software receiver, and the >expertise and time to program the DSP, the only game >in town is to actually use two separate signal chains in >the receiver. two RF stages, two mixers, two IF strips >and two detectors. The detectors can't be product detectors. >The detectors must strip out the phase information, or you will >recreate the destructive multipath cancellation when the two >audio signals get summed. AM, you can use a product detector. >CW, you can use heterotone, which is AM modulation of an IF >amplifier, followed by an AM detector. I don't know what you >can do for an SSB detector. I don't know of any filters which >can pass the wider signal that would come from the two two >polarization channels, and still give a decent passband. > >As for doing this at the TX, you have to know the path, >which means the far-end receiver has to be telling your >eqauipment in real time, what it is receiving. This scheme >has been proposed for some forms of mobile phones, >where the fading happens at a relatively sedate pace due >to the fading NOT being Rayleigh fading (since these are >extremely wideband CDMA signals). For an HF path, >I have never heard of anyone attempting it using the TX. >You could do it if you have a fast enough Internet connection or >a phone line back to the TX. But that would be a stunt that >kind of dfeats the purpose of a wireless connection. > >I'm glad to see people trying to push the state of the art. >BTW, QST in 1936 [April or May ?] had a diversity receiver. > >Darrol