Antenna Education Needed

[sfx2000]

Does an antenna's nearness to a wall create a 'splash-back' effect for WiFi signals? (Let's assume it's a transmit-thru capable wall, too.)

I think I need to find out if a signal, once it leaves the antenna, is broken into some packet-sequence that requires "flight distance to form up", like a 300-plane bomber raid takes an hour for all to take-off, so they circle and wait for everyone and then 'form up'.

This assumes, then, that a WiFi signal can self-validate in-flight (NO, IT CANNOT) and thus it could send back re-transmit requests to the WiFi origin point. I prefer to believe packet-validation is only performed upon Destination Receipt, or perhaps in Transmit Checks (the router knows "I coughed - I'll just retransmit that signal now, automatically").

The nearness to a wall might create a validation-error at the Antenna itself if the "flight of WiFi bomber signals" required a flight-formation transit.

I just don't think that's true, though.

Therefore - does an antenna's distance from a wall matter? Is six inches better than 1 inch? Is 18 inches better than 6?

1/4 wavelength in free space is 30.7mm with a dipole antenna...

In Imperial measurements, that's about an inch and a quarter and some spare change...

Most households are sheetrock, with has about 3dB attenuation... Plaster/Lath (chicken wire) is about 5 dB - no "splatter"/self-jamming by the AP. None of these reflect anything back to the antenna of the transmitter... just a harder path to punch thru for the Tx/Rx path.

Best guidance here for 2.4GHz and 5Ghz - don't put the AP on top of a metal base (e.g. file cabinet), as this can impact the beam patterns.

Plastic/Wood base - this is ok...

sfx

 

[TonyH]

Hi,
Chicken wire or mesh opening is less than 1/20 of Lambda, it'll act like reflector.

 

[sfx2000]

Hi,
Chicken wire or mesh opening is less than 1/20 of Lambda, it'll act like reflector.

Not enough to matter... Path loss is greater here...

 

[Just Checking]

I am not an antennna designer or even an expert. Here is a little FAQ that I found which is food for thought:

http://www.wlanantennas.com/faq.php

The placement of an antenna near a wall and its effect on the signal strength/quality will, as other posters have already stated, depend on many factors. Here are a few that I think are most important (IMO and not necessarily in actual importance).

1. Surface Reflectivity to the wavelength signal - Dish reflectors do just that. They reflect signals that are coming to them from the source and direct them to an area where they are more effective. Dishes are curved with dimensions that put the transmitter/receiver at the focal point. Walls are (generally) not as effective at directing the signals.
2. Surface Smoothness of the wall. Smooth surfaces (like Sheetrock or Plaster) are better reflectors. Rough surfaces (like brick or stone) difract the signal (send it back in many different directions).
3. The exact distance between the reflecting surface and the antenna. If, by chance, the distance between the wall and the antenna is an exact multiple of the wavelength, it might be possible to get interference. That distance is different for the 2.4GHz and the 5.4 or 5.8GHz frequency bands. I think the wavelength for 5.4GHz is like 1.3cm (probably wrong about that), so having a antenna more than 15cm (6") from a wall would be safe for most applications. This is backed empirically by looking at many commercial wireless antennas on buildings where the antennas are placed about 25cm (1') away from the building.
4. Antenna design - Omnidirectional antennas are usually vertically polarized. This means that the signal is concentrated in a roughly horizontal plane. Different antennas broadcast with a particular vertical angle or spread. That spread or cone vertically can be anywhere from 10 degrees for narrow cone (YAGI) antennas to 60 degrees for a lot of the small antennas like the OEM antennas that come with most home routers. If the antenna has a narrow vertical dispersion and goes against a highly reflective surface, the chance of signal interference is increased. For normal home router antennas, this is not a big factor if the antenna is more than a few wavelengths away from the wall.

I am sure that the radio guys can explain this much better than I just did.

Another thing to consider is how the commercial equipment is designed to be placed and operated. All the repeaters are made to be wall mounted with the antennas close to the wall (within 10cm of the wall). If there was a significant negative impact on radio performance, I would bet a lot of money that the commercial equipment would be designed to move the antennas farther from the walls.

 

[TonyH]

1/4 wavelength in free space is 30.7mm with a dipole antenna...

In Imperial measurements, that's about an inch and a quarter and some spare change...
sfx

Hi,
Dipole ant. radiator is 1/2 Lambda, cente-fed. Broken into 2 1/4 Lambda radiator. In case of vertical mount 1/4 Lambda is in the air, the other half is projected into ground or counterpoise or graouind plane ,etc. That ia really dipole. We ca also use center-fed dipole in open air. This one will be 12 Lambda long(often not practical) DBd has little more gain compared to reference DBi.

 

[TonyH]

Reflector is either parabolic or hyperbolic curved focusing radio beams to feeder/receiver(LNA) horn. If object like wall reflected wave is in phase with radiator it'll
yield positive gain, if out of phase, the opposite.
Commercial antennas are usually mounted using brackets which can maneuver antenna for
proper positioning.

 

[sfx2000]

In most MIMO systems, Lambda/2 is ideal as a rule of thumb in design... This applies to both SM and TxBF - challenge here is matching for dual band usage at 2.4/5Ghz...

But this is done, easily enough. Been there for CDMA handsets, Wimax, and WiFi designs.

Going back to my original comment, chicken wire is a good match for both, and that's why we see strong attenuation in both band for Plaster/Lath vs. Sheetrock/Drywall..

2-3 dB is significant between the AP and the STA. Difference between being there or not... esp. at the low Tx levels of WiFi and the narrow link budgets afforded by 802.11..

 

[stevech]

2-3 dB is significant between the AP and the STA. Difference between being there or not... esp. at the low Tx levels of WiFi and the narrow link budgets afforded by 802.11..

Technically correct, but in practice, a link margin of 10dB is needed to accomodate fades and human body blocks and many other variations.

Running with 2-3 dB of margin will cause an unwanted high packet error rate.

 

[TonyH]

Lambda/2 is calculated value but considering velocity factor antenna gets bit shorter up to 85% of Lambda/2 in real life. To have solid link on a fixed path , signal level at RX end should be above quieting knee. If path is refracred vs, LOS, it is different again. (Right, visual LOS and radio wave LOS is not same.)

 

[azazel1024]

Technically correct, but in practice, a link margin of 10dB is needed to accomodate fades and human body blocks and many other variations.

Running with 2-3 dB of margin will cause an unwanted high packet error rate.

For desired link speed, or for minimum SINR for any link, sure. However, 2-3dB can be significant in terms of link performance.

Based on testing with different antennas, it seems like 10-25% performance difference with no other changes with a 2dB difference in antenna gain (depending on location, band and client). So 2-3dB might not be much when taking in to account minimum link budget, but it is pretty significant when determining utilizable link speed.

 

[sfx2000]

Technically correct, but in practice, a link margin of 10dB is needed to accomodate fades and human body blocks and many other variations.

Running with 2-3 dB of margin will cause an unwanted high packet error rate.

Absolutely - the 2-3 dB is the relative difference in the use case discussed (sheetrock/drywall vs. plaster and lath).

10dB is a good target - as a general rule, I actually like to get 20dB above the relative noise to get best performance (20dB is also the spectral mask limits for 802.11).

BTW - I usually calculate body loss as 5dB - some parts are obviously more dense than others, lol, but this has worked well for me over the years - also fades in WiFi have generally been more of the Rician model, as both AP and STA tend to be fairly static - however, Rayleigh comes in to play, esp now with most smartphones having WiFi.

 

[stevech]

Absolutely - the 2-3 dB is the relative difference in the use case discussed (sheetrock/drywall vs. plaster and lath).

10dB is a good target - as a general rule, I actually like to get 20dB above the relative noise to get best performance (20dB is also the spectral mask limits for 802.11).

BTW - I usually calculate body loss as 5dB - some parts are obviously more dense than others, lol, but this has worked well for me over the years - also fades in WiFi have generally been more of the Rician model, as both AP and STA tend to be fairly static - however, Rayleigh comes in to play, esp now with most smartphones having WiFi.

In urban areas, one needs 10 or more dB of margin, to meet/eceed the goal SINR, not just the SNR. SINR = Signal to Interference and Noise ratio. Interference is other WiFi (coherenet signals) and also non-coherent from non-WiFi emissions.

5dB for body-block - I'd go higher - we use 5dB for UHF like police radios in 800MHz. More lossy at 2.4GHz on body block.

 

[azazel1024]

In urban areas, one needs 10 or more dB of margin, to meet/eceed the goal SINR, not just the SNR. SINR = Signal to Interference and Noise ratio. Interference is other WiFi (coherenet signals) and also non-coherent from non-WiFi emissions.

5dB for body-block - I'd go higher - we use 5dB for UHF like police radios in 800MHz. More lossy at 2.4GHz on body block.

You may be right, but my measured body block (I realize InSSIDer is not exactly a super scientific tool) is approximately 5dB for 2.4GHz and closer to 7dB for 5.2-5.9GHz.

Now that may be because of multipath/signal bouncing reducing the impact of body block on an indoor environment, but that is roughly the averaged signal drop I've seen doing a bunch of testing.

I've seen from 4 to 6dB for 2.4GHz and 5-8dB with 5GHz band.

 

[stevech]

measuring a valid RSSI with a MIMO system is a challenge. When MIMO is working right, on BOTH ends of the link, and when the multi-path conditions in a certain area are just right, one can make a meaningful measurement. That's not RSSI. It's packet error rate with fixed-size layer 3 IP packets (e.g., 1/4 MTU). Because some MIMO conditions enable some multi-path to be "favorable", RSSI won't show such.

But I believe that simple power-in-the-channel RSSI, which is what most products display, is good enough to assess how much margin exists, in each direction. For handhelds at high rates, lots of margin is needed.

A curious thing exists in wireless: Higher mobility means lower data rates are NEEDED. (Not to confuse an automobile with an 4G/LTE to WiFi bridge on the freeway - this is an immobile WiFi situation).