# Diversity - Distance between Antennas

Oct 14th, 2008

Considering both antennas are similar-type and gain the two imp criterias i read in achiveing Diversity is

1. Antennas should be placed close enough to each other so that the RF coverage area is nearly identical.

2. The receiving antennas are spaced sufficiently apart to achieve independence (no coupling) between the received signals.

So the distance between both the antenna would be important.

In a cisco doc,

http://www.cisco.com/en/US/tech/tk722/tk809/technologies_tech_note09186a008019f646.shtml

For 2.4GHz wavelength = 12.5cm and for 5GHz wavelength = 6cm.

Hence for 2.4GHz wavelength,

Dist bet diversity Ant = 12.5cms Or 25cms or 37.5cms or 50cms

now from below POST:

The spacing can be in multiples but is best in multiples of odd numbers. 1x or 3x. This allows for the phase to be always 100% out for the opposing antennae. 2x could cause phase shift overlay and could result in both antennae receiving the same phase shift differentiation.

Is this true? then my choise would narrow down to "Dist bet diversity Ant=12.5cms or 37.5cms'

But from below POST iam confused regarding point 1 and 2 (I have put them pointwise just for reference)

1. Never should you have the antennas exactly one wavelength away from each other. For the frequency of 2400, one wavelength is 4.92"... so any distance that is not a multiple of 4.92 and no more than a multiple of 4 is recommended.

2. You should use 1/2 wavelength distances and to be safe no more than multiple of 3.

My Questions are

1. Should the distance between my Antennas for diversity be any len more than 12.5 cms and less than 25cms or more than 37.5cms and less than 50cms?

2. Is it that the distance between my Antennas for diversity should not be exactly equal to the multiple of wavelengths? (12.5cms, 25cms, 37.5cms, 50cms)

3. Is it that i have to use only odd wavelengths multiples? (12.5cms, 37.5cms) and should not use even multiples?

4. The use 1/2 wavelength distances between the antennas is diversity. Is it ok? or is the minimum 1 wavelength as specfied in the cisco doc.

Also from the below POST iam confused regarding the MAX distance:

"Distance beyond 1-2 wavelengths is not critical. Since the antennas are not transmitting and/or receiving at the same time,there is no real chance of overloading the other antenna or radio front-end. "

There's not really a "max distance" as far as diversity is concerned. By placing the antennas some distance apart (the minimum, I believe, is ~one wavelength), you optimize the chance that a signal that is null (or interfered with) on one side, isn't on the other.

my question is

5. Is there a max distance between the antennas in diversity (like we read 4xwavelength) or there is no max distance? then wont there evolve a different RF coverage area for both the Antennas?

Another important question

> If my wireless card supports 5GHZ and 2.4GHZ and hence my antennas would be dual band. Then to achive antenna Diversity, what would be the distance between the two antennas? The minimum distance should be 1 wavelength, so For 2.4GHz wavelength = 12.5cm and for 5GHz wavelength = 6cm.

For a dual band antenna? how much should it be? Whats the min and max.

someone please can help me solve the huge confusion in my mind :)

Overall Rating: 3 (1 ratings)

## Replies

scottmac Wed, 10/15/2008 - 21:16

First, let me say that this is certainly a well-done post, good job of back-tracking the links and expressing your questions.

The next thing is that there are more than a few nuances when dealing with RF and signal transmission and propagation, so to cover a specific topic fully would frequently cover more space than provided.

With all that dancing out of the way ;-), I'll take a shot at answering your questions...

"Is there a max distance ... ?"

Well, no, not really ... but there is a max *effective* distance and a fairly specific point of diminishing returns (as mentioned in the linked posts).

The design and implementation goal is to create a sort of "binocular view" with the antennas, such that the chance of a "null" is reduced, such that the signal / pattern developed from the diverse pair of antennas will create at least one clean and clear (reduced multipath, reduced nulls, possibly better line of sight)signal to (ideally) every client in the coverage area.

Even though only one antenna is active at any given time, having any conductor within the radiated field will distort the transmission envelope, either positively (in the case of a Yagi) or negatively (in the case of a grounded conductor, like a sprinkler head).

The amount of the distortion, and the shape of the distortion are functions of the size and shape of the parasitic element. The effect is so variable that it is one of the foundations for the contention that a comprehensive site survey is not really an option, it's mandatory; it's truely the only way to see what adaptations are necessary to make the "rules of thumb" function in the specific location.

Regarding antennas handling multiple bands:

The antennas (in this case) are close to resonant multiples. The differences are such that the logical / electrical length can be adjusted by adding inductive or capacitive loads at the transmitter or feed point.

Even though the physical shape of the antennas may be different from model to model, the electrical load characteristics for a given band / antenna remain similar, so electronically compensating to make an antenna primarily designed for 2.4GHz function well for 5.6GHz is not especially difficult. Going from the longer wavelength to the shorter is more effective and easier to design than trying to go the other way.

For example, a quarter-wave 2.4GHz antenna is easily compensated to create a good half wave (or loaded 5/8) 5.6GHz antenna.

So, (finally) to provide the proper spacing for a diverse pair of multi-band antennas, figure for the longer wavelength (2.4GHz in this case) and the (roughly double) distance is likely to be acceptable for the higher wavelength. The spacing doesn't have to be exact; personally, I used to shoot for ~18" (variable with the mounting location and antenna type)and I always had pretty good luck, according to the follow-up survey.

Because of the jillions of possible variables, you can't just calculate and hang the antennas and be done ... a survey should be done to verify the installations and to make educated adjustments that will optimize the system for that specific location and environment.

The initial calculations are ideal guidelines, then you make adjustments to cover things like metal wall studs, sprinkler heads, the boss' favorite "wall o' vines" ... you will rarely get a perfect environment, adjustments are almost always required.

It's late and I'm starting to babble ... ponder the above and let us know if it helped any. There are a great bunch of experienced folks here, and I think this will be the start of a great discussion.

Good Luck

Scott

gamccall Tue, 10/21/2008 - 11:13

With all that said- a large majority of Cisco antenna solutions take all of this decision-making away from you: If you use dipole antennas, then they're going to be spaced however the connectors are spaced on the AP, and if you use the double-pigtail or quad-pigtail diversity antennas then the spacing is whatever is built into the antenna.

So unless you have a particular reason to use pairs of non-diverse antennas rather than just buying a diversity antenna, spacing is not something you have to think about at all.

Richard Bradfield Tue, 10/21/2008 - 19:44

this is a great conversation, I have a bit of a dilema, I agree with what has been said above, but we had an independant company come in to do a site survey at one of our manufacturing plants. Now their recommendation was for each antenna to be 50 cms from the AP(therefore a metre apart) and 50 cms out from any wall

Now they said they have many installations using this setup ok, but we are not convinced!

it does not seem to be true diversity if the antennas are that far apart.

david.kelsen Wed, 10/22/2008 - 17:47

This is a topic that has plagued my thoughts for far too long. I've attached a photo of an AP set up that I've used on several of our APs.

The antennas are Cisco AIR-ANT2506 mounted on either side of the AP enclosure. The AP enclosure is approximately 10 to 12 inches wide (26cm - 30cm).

Does anyone have any comments on the way the antennas are mounted?

Is this good or bad practise?

Would this achieve 'effective diversity'?

Attachment:
Richard Bradfield Wed, 10/22/2008 - 18:30

Further thoughts

It seems to me there is really no limit to how far apart yo can have the antennas, but at what point does it stop being diversity and each antenna covers a different area?

scottmac Wed, 10/22/2008 - 20:03

Some of that question can't really be answered without considering the radiation pattern of the antenna.

Cisco's official documentation seems to indicate that maximum effectiveness goes out to ~ four wavelengths (in round numbers) about 20 inches.

What the Cisco documentation doesn't discuss in any depth is the type of antenna, the pattern of the antenna, the mounting surfaces and other pattern modifiers.

IMO, the "same area" spec exists as an indication that having one antenna inside for an AP application while the other antenna is outside for a distance bridge application is not the intended implementation.

Where having a pair of antennas configured for diversity to cover the "same space" will do so effectively from one wavelength ( ~5 inches) out to 36 inches or so ... given the cable length permits it, the type and orientation of the mounting, the type of antenna, and the surrounding materials (wall, ceiling, beams, etc).

Unless the hosts and the antennas are absolutely static and absolutely all exactly the same distance from the antennas, then there is no optimum absolute "diversity spacing" of the antenna pair.

In brief, general terms, diversity is a mechanism to reduce the chance of 1) multipath (one or more reflected waves interfering with the desired signal) and 2) nulls (a point in space where the phasing of two signals negates one signal with the other).

If you put two (identical) antennas up as a diversity pair there will always be somewhere within the "same space" where there are nulls and multipath ... changing the lateral relationship of the antennas only moves the points of the multipath and nulls ... they'll still exist, likely in the same numbers, they'll just move.

Without looking at specific antenna type (pattern - omni, sector, patch, panel), and the mounting surface and location (wall versus ceiling versus mast), and the contents of the space (hard / soft / plants / windows / wood / metal ...)you cannot come up with an accurate "rule of thumb."

And, trying to hang a pair *exactly* one or three or five wavelengths apart is, IMO, an exercise in silliness because it is only effective for hosts located in some very exact locations in the space.

As always, a survey is your best friend. As for working out a location for a diversity pair ... light up one side, do a room sweep ... light up the other side, sweep the room again ... if you can move the test host within the space and maintain acceptable connection for both, everywhere, you're good.

If you get some areas with poor Signal Quality, move the antenna(s) ~2" and try again.

Try to gain a mindset that stops seeing APs and antennas and starts seeing fields ... like light, (i.e., where are the shadows, if the surface was a mirror, how would light reflect from it? Does the "light" from one antenna cover a shadow from the other?)

My installs may not have been "by-the-book" perfect, but the hosts got good signal strength and quality for the entire coverage area (and not much more for security), with a reasonable AP count for the budget.

Good Luck

Scott

scottmac Wed, 10/22/2008 - 19:22

Maybe the picture's angle misrepresents the reality ... but it would appear that the lower few inches of the antennas are blocked by and / or mounted to, the enclosure. Then, in addition, everything looks to be flat-mounted to the way, using omni antennas.

If that's true, it would be a Bad Thing ... not really dangerous or likely to damage the equipment, but just a pitiful bad design with lots of wasted power and pattern.

If the coverage area is only 180 degrees / hemispherical, then use antennas that project a pattern of 180 degrees / hemisphere (for example).

You will not get 2.4GHz through brick, certainly not to any usable distance. As a worse case, wet bricks will kill your pattern, iron re-rod used to reinforce the wall will distort the pattern. The reflection or multipath off the brick (wet or dry) will kill your pattern.

As far as "practical diversity" it's OK, it'd work (if the rest of the installation / implementation wasn't so bad).

Just my .02

Scott

david.kelsen Wed, 10/22/2008 - 20:04

Thanks Scott,

Do you think the area of the pattern that isn't wasted (ie., the 180 degress) would still provide good coverage in this scenario?

The antennas in the photo are mounted a little low, I'll be correcting this in the near future so that the PVC area of the antenna clears the top of the AP enclosure, and only the aluminum base of the antenna is mounted to the enclosure, would this be better?

I take your point with regards to the wasted pattern and brick wall, unfortunately using omnis was the recommendation from an external vendor and was implemented a few years ago.

I suppose my question was really about the spacing apart of the antennas, being an AP enclosure width apart, which you've answered.

Thanks for your time, you're input is appreciated.

Dave.

scottmac Wed, 10/22/2008 - 20:20

Mounting the antennas so the radome (the radiotransparent fiberglass part) is clear of obstruction would probably help.

The spacing is OK. "By the book," ~5 inches out to ~20" in roughly 5" spaces would be most desirable. I sort of disagree with the book on this one ... if its not too much of a hassle, set some stationary hosts in "must have" locations, and tweak the distance to your application.

Some perforated metal strap, or a (bar...) attached to the top (or bottom) of your enclosure would give you some adjustable spread and a way to mount by the base away from the enclosure.

If you can stand the antennas off from the wall some (about 3", which is 5/8 wave at 2.4GHz), you may see some positive gain from the wall bounce (which will vary with wetness, rebar, bugs, humidity...).

If you can place an ungrounded conductor (like an aluminum sheet) on the wall behind the antenna, you might get some positive effect from that as well, and it would likely be more consistent in performance.

Good Luck

Scott

## Trending Topics: Other Wireless Mobility

 client could not be authenticated Network Analysis Module (NAM) Products Cisco 6500 nam reason 440 driver failure Cisco password cracker Cisco Wireless mode