How do I measure/calculate antenna isolation?

[fbooth]

I have two antennas - one transmit the other receive. They both work in the same band (GHz). The beamwidths for each antenna are known for all frequencies. The beams point parallel the same direction. How may I calculated the isolation for a given displacement in meters? Then, can someone suggest a site where a measurement procedure for antenna isolation can be found?

Thanks,
Fran

 

[VEBill]

Assuming that you arrange the antennas for maximum isolation, it is possible that the remaining coupling will be via paths and mechanisms that involve external or secondary factors (reflections, diffraction).

In general, depending on the scale of the problem, you can use either GTD/UTD (ray tracing) methods or various NEC-like methods. NEC-like approaches can be defeated if the problem space is too large.

There is a lot a free NEC SW available:

http://www.si-list.org/swindex2.html

Here is a link on ALDAS for coupling:

http://www.maasdesign.co.uk/aldcou.html

Depending on your purpose, it might be a lot easier to conclude that the isolation is much better than an acceptable number than to calculate the exact number. Such an approach can save a lot of time...

 

[Higgler]

Here's a quick formula for calculations in your head, when you have no handy calculator or internet.

1222 or 12-2-2,
you get about 12 dB isolation at 2 ghz at 2 inches spacing. This set of number was chosen as an easy way to remember, or think of lunch, it's about the middle of lunch (12:22 hrs).

Every time you double the distance or double your frequency, add 6 dB to the 12 dB starting point.

Example; Calculate isolation at 32 inches at 8 ghz = (2">4">8">16">32" is 4x6= add 24dB) and 2>4>8 ghz is another 12 dB. Isolation is 12+24+12=48 dB for two zero dB gain antennas 32 inches apart at 8 ghz.

If the antennas aren't zero dB gain, and are both +4 dBi, then 48 -4-4= 40 dB isolation. You need to add the antenna gains into the summation. The forty is actually -40 dB coupling.

Antennas must be in the far field of each other, or your isolation will be greater than zero in some low frequency instances.

kch

 

[VEBill]

This '12-2-2' rule-of-thumb doesn't mention the relative orientation of the two antennas. For practical (non-isotropic) antennas (such as dipoles or monopoles), orientation obviously matters. There must be an assumption in this rule-of-thumb - probably that the antennas are arranged for maximum coupling ('face-to-face').

For this application ("one transmit the other receive"), it would make sense to arrange the two antennas in a co-linear arrangement (both vertical, stacked one above the other) so as to minimise the coupling. This co-linear arrangement is commonly done for VHF/UHF repeater stations that use separate R & T antennas.

This co-linear arrangement is what I was intending as an obvious (?) assumption when I wrote my opening line, "Assuming that you arrange the antennas for maximum isolation..."

 

[Higgler]

I should have used the phrase "space loss" for the 1222 quick calculation.
True isolation does take in the orientation of the antenna and their relative gains plus other factors.
kch

 

[fbooth]

Perhaps I should clarify the situation.

The transmit and receive antennas are on the same platform, with the boresight of each in the same parallel direction mounted on a vertical metal mast - transmit above receive - separated by a distance of x units. The receive is always "on". The objective is to determine how much energy from the transmitter will get into the receiver for a given displacement between the antennas.

 

[VEBill]

Assuming that these "GHz" antennas are relatively high gain (perhaps parabolic dishes for this example), then the coupling becomes a matter involving small details - for example:

-The side lobe performance of the feeds.
-Positioning of the feeds relative to dish edge.
-The feed taper, how many dB down at the dish edge.
-The details of the edge (diffraction).

In other words, the total coupling is not something that can be ~exactly~ calculated from simple characteristics.

On the other hand, if the feeds are in direct view of each other, then the main coupling mechanism might be as simple as from one feed's side lobe to the other feed's sidelobe.

If this is true, you might want to shield the two antennas with a metal plate. If the plate itself causes difraction, then you could investigate adding anti-diffraction curves to the edge(s) between the antennas. Sometimes large TVRO dishes will have little petal-like curves to the edge of the dish to reduce diffraction.

Or set one antenna a few feet further back than the other.

As mentioned earlier, it is likely easier to set bounds than to calculate the exact number. Also, measurements might be relatively easy (depending on details...).

Can you tell us more about the exact type of antennas?

 

[Higgler]

That's easier to measure than to calculate. Detail calculations for higher isolation setups are not usually possible in todays antenna software.

Sample data results:
6-18 ghz 20 dB gain multi-beam antenna arrays, spaced 4 feet apart, one atop another will have 70 to 90 dB isolation just on their own and about 100 to 120 dB isolation if you try to block RF and surface currents between them. This is measured data I have on a real program from about 15 years ago.

Here is some estimated numbers of isolation for a pair of 15-25 dB horns one atop each other, just to give you my feeling for isolation levels versus difficulty (I've worked alot of isolation problems and have a good feel for the results);

60 dB isolation: easy to do, 3 foot separation, multi-path probably not an issue, unless you point at something within 50 feet? tree, etc.
80 dB isolation: Not too bad at ?6 feet separation, you may want to orient/move the antennas relative to each other, i.e. offset them, straight up and down is likely to be worse isolation than if they were say 3 feet vertical and 2 feet horizontally displaced.
100 dB isolation: difficult, multipath comes in
120 dB isolation: severe multipath problem. Example: One of our antenna array TX to single antenna Rx had 117 dB isolation worst case in the 6-18 ghz band, we received more energy from trees at 1100 feet away than we did from antenna to antenna coupling space 4 feet apart. Birds flying by would put blips on the recorded data when testing.

details on your setup and approx. frequency range would be nice.
kch

 

[fbooth]

About the antennas:

Transmit - pyramidal horn

Receive - spiral

 

[VEBill]

(Higgler: that data is great)

Transmit: pyramidal horn Receive: spiral

Depending on the dimensions of the horn, it is possible that diffraction from the edges of the horn might be a dominant radiation mechanism in the direction towards the receiving spiral. If so, then it might help to add some 'edge treatment' to reduce the diffraction.

It might be even easier to simply mount the transmit antenna out front by several feet and with a screen behind it. It should be simple to achieve very high isolation.

 

[Higgler]

If the transmit pyra midal horn is linear polarization, then orienting the spiral in the H plane of this horn will provide 20 dB more isolation than in the E plane. i.e. if the pyramidal horn is oriented for vertical polarization, worst isolation occurs with the spiral located above or below this antenna, better to have it side to side (horizontal orientation). Much more energy (20 dB) turns the corner in the E plane than does the H plane for those horns. If you have to orient them Vertically and the Pyramidal horn is V pole, locate the Pyramidal in front of the Spiral. Of course that depends on the amount of isolation needed.

If you can orient the pyramidal horn for Horizontal polarization, then having the spiral below it gives best results.

Antenna's crosspole to each other can get very good isolation if you really need to improve things.

How much isolation is needed and was that number properly derived from system requirements. If you have a narrow bandwidth, life is much easier.
kch

 

[fbooth]

I want to thank everyone for their help.

I found the calculation part of what I was looking for in Richard C. Johnson, Antenna Engineering Handbook, 3rd edition, pp. 26-38 to 26-40.

 

[SolderingGunslinger]

Being a hands-on type person, I would probably do the math, then take a Network Analyzer and sweep the system with an antenna on each port.

An advantest R3767CG Network analyzer would be a good choice.

If a Network analyzer isn't available, a spectrum analyzer with tracking generator will work, but not give you as much information about the antennas as a complex circuit.

I remain,

The Old Soldering Gunslinger

 

[tlh0598]

Higgler,
Does that rule of thumb only work with omni antennas? Was the 2" spacing vertical or horizontal? How can you determine what the near field actually is (in meters)?

TIA

 

[Higgler]

thl0598
The isolation quick rule of thumb 12-2-2 is for "spaceloss", a term which implies isotropic, zero dBi gain antennas. The 2 inch distance is from antenna to antenna, any orientation. If the antenna gains are truly isotropic, it doesn't matter which direction you go, up or down.

This is a "far field" calculation. Near field is usually 2d^2/Lambda which is for accurate antenna measurements for standard antennas (not low sidelobe antennas).
If you have a horn antenna pair side by side, and you really are just coupling edge to edge, then the distance can be much closer than 2d^2/Lambda because the effective size of d somewhat changes as you change orientations of the antenna.
Not an exact science and in most situations Isolation calculations can only be accurate (+/- 1 dB) for antennas pointed at each other. Off in sidelobe areas the multi-reflections from support towers, etc. make isolation calculations les accurate.
kch

 

[ipurple]

I think that if you'll take a sight to the Evenode Technique, this will be usefull for you.