Router upstairs or down?

It would probably be better off, as far as the radios alone are concerned, if the AP was in the middle, on the main floor. That would be only one floor away from whatever you have on the top floor or in the basement. No path would require going through two floors.

That said, radio paths are not the only consideration, and if it works as is there is *no* point in changing it just for better signals.

Another possible configuration, which should work just as well as what you have, is to put the AP in the basement. That has exactly the same relative problems with radio paths as your current setup, but it also has one other characteristic that might (or not) be an advantage. If the AP is in the basement it will have less coverage horizontally away from your house. In other words the exposure to the neighbors is less. That may mean they can't crack into your network, or it may mean your network won't interfere with theirs as much... or it may do nothing.

The radiation pattern for a typical omni-directional antenna looks something like a fat doughnut with the antenna in the center hole. The doughnut represents signal strength, so the more you go through, the better.

Straight off the ends of the antenna there is less signal, and perpendicular to the antenna, all around the antenna, is best... regardless of whether the antenna is vertical or horizontal.

If your antenna is vertical and sitting on the 2nd floor in the center of the building, that means good coverage on that floor, and poor coverage directly below it on the 1st floor but getting better at locations on the 1st floor which are not directly under it.

That is a bad location, generally, simply because there is no way to orient the antenna and have it cover everywhere well.

Think about horizontal orientation and right up against an outer wall. The worst client locations then would be the two corners off the ends of the antenna.

Or perhaps if the antenna is placed in a corner of the building, with horizontal orientation and angled 45 degrees to the walls, so that perpendicular to the antenna is a line straight to the center of the building. That one perhaps gives the best coverage of all.

Of course if you have only two or three locations in the building where you'll have a client, then you can optimize for just those locations.

Also, if all else is equal, the client antenna should have the same angle of orientation as the AP antenna. However, things are *never* equal... reflections off any metal surface larger that approximately 4" square will affect the polarization. So it's always a good idea, if the signal is anything less that very good, to try moving the client antenna around a bit to see if things can be improved with different angle.

Vertical is fine *if* the direction to all of your clients is horizontal. That isn't the case in a multi-floor building though, so horizontal is almost guaranteed to be better for that application.

Oh, it matters!

First you've got the directional characteristics of the antenna itself. Perpendicular to the antenna is the best signal, and directly off the ends there will be a null in the pattern that might very easily approach 20 dB or even more. That can simply mean the difference between a nice solid connection with a -68 dBm signal strength, and an intermittant or non-existant connection with a -88 dBm signal strength.

Then there is the matter of what is called polarization of the signal. Which is the relationship between the transmitting antenna and the receiving antenna (assuming nothing in between reflects the signal and changes its polarization). There can easily be 25 to 30 dB of difference! So the two antennas should at least start or default to the same orientation. It's a good idea to try moving the client antenna around a bit, because any signal reflections off of metal objects will change the polarization. You might find 5 to 15 dB of extra signal that way, just by tilting an antenna 30-60 degrees!

Sounds good, and it works.

Radios are commonly accepted to work on the basis of FM. FM, of course, means Freaking Magic, or something close to that. That's why you set them up the way their should work, and then twist every knob and move everything you can, just to see what happens. They almost always work a little better once you've misadjusted them! (The trick is knowing *which* misadjustment will improve things.)

"LT" wrote in news:mM2dnT snipped-for-privacy@comcast.com:

The 'received wisdom' is that the higher the better. However, in the case of a typical two-storey house with a wood/plasterboard first floor, I doubt it makes any real difference.

I have mine on the top floor and can receive a "good" signal 2 floors down in the basement. Top floor, main floor, basement is the home configuration.

surely the orientation, vertical, horizontal diagonal matters not. After all its only a signal.

I have mine horizontal, lower floor 5' from ground level and all seems fine. The air vents are at the top for heat escape mind.

dj

So for a 2 antenna AP in a hard to get to loc (attic), how about setting set one ant. vert and one horz? Seems like you would get the best coverage.

Reflections can kill the S/N ratio and pollute the data stream with inter-symbol interference. OFDM is very good at recovering from this kind of multi-path (reflection) interference, but not with a very weak signal. You can approach the problem of in-building coverage two ways:

1. A whopping strong signal and use OFDM to recover from the reflections.
2. An antenna and polarization insensitive configuration with fewer reflections, but less signal strength.

The first is easy. Disable 802.11b and set the access point to

802.11g. Even at the slowest connection speed (6Mbits/sec), OFDM is much better than CCK at dealing with bouncing signals.

The 2nd method required circular polarized antennas. It can be right and or left hand sense. The clients still run vertical or horizontal polarization, but the access point doesn't care which one arrives. The catch is that there is a -3dB cross polarization loss which reduces the signal strength somewhat. Considering that the deep fades caused by multipath cancellation and cross polarization are *MUCH* larger than -3dB, methinks this is a good tradeoff.

I haven't done any testing with circular polarized access point antennas at 2.4Ghz. However, I once did quite a bit with 440MHz ham repeaters and 850Mhz cellular on mountain tops. Raleigh fading causes the signal to get "chopped" as a vehicle moved through the multipath cancellation nulls. It was also called "picket fencing". Replacing the conventional vertically polarized antennas with circular polarized antennas on the repeater or cell site drastically reduced the picket fencing with only a slight reduction in range. Most FM broadcast stations have been using either circular polarization, or split their signal between horizontal and vertical polarized radiation to accomidate home users with horizontal antennas, and moving mobiles with vertical whip antennas. The PCS cellular providers are also mixing the polarization to accomidate handsets with hidden internal antennas, which are horizontally polarized. It's not like circular and mixed polarization is something new.

Trying to get reliable coverage through multiple floors is a waste of time. It may work but it's not going to be consistent, stable, or reliable. Almost all the signal between floors is the result of a bounce off of something, which tends to be unstable. The signal usually appears miraculously at the distant end of the house, but disappears when something is moved or changed. The result is constant repositioning of the antennas which can get irritating rapidly. If you want reliable coverage of multiple floors, use multiple access points on each floor with a wired backbone. If you're cheap, run coax from the 2nd antenna on the access point to the floor above or below to another antenna.

Drivel: One of my friends claimed his access point would go through 3 concrete floors in a college dorm. A bit of direction finding demonstrated that the bulk of the useable signal was bouncing off the adjacent building and not going direct through the floor. Of course, the bounce path was not very reliable.

I find considerable success with the ap on the ceiling downstairs.

all the same , normally vertical . ( thats right angles to the earths surface . :)

I guess I haven't been paying attention. Circular polarized 2.4GHz antennas:

I dunno about this antenna that claims to be "multi-polarized".

Downstairs high up on a table or upstairs on the floor.

No, you want the antennas to be horizontal to each other. You have two antennas for a reason. Signals are getting to the antenna at different times and cancel each other out; that's why you have two antennas. They are there for diversity.

Somebody please correct me if I'm wrong.

Yes. That would be particularly beneficial if your clients actually move around. If you only a desktop, for example, which stays in one place you can pretty much fool around and get optimum results. You only have to do it once and the benefits last. One antenna would be just fine.

But a laptop that is used just about anywhere you happen to be at the moment is a different story. You don't want to have to fool with it each time you move.

The two antennas operate absolutely separate. There is a switch, and only one of antenna is connected at any given time. The unit will periodically switch to see which antenna works best, and then lock on that one antenna for that interval. Which antenna to use is determined by the quality of the received signal, and it just hopes that means the next transmit operation will also be best on that antenna.

If everything stays the same it means that if there is only one Client connected to the AP, it will probably spend virtually all of the time connected to one antenna. If there are two or more Clients, or if the one Client moves, or if anything affects the signal path (for example, someone opens the refrigerator door... and that just happens to provide a better path for the "other" antenna), then it will be switching.

But in some cases it will switch to the antenna best for one client and instead transmit to another client that would have been better with the other antenna.

So it is not perfect, but is probably a significant overall improvement.

Except they *don't* cancel each other out, because only one antenna is connected at any given time. That means that if one is horizontal and one is vertical, you can change the polarization of the signal from the Client and it will cause the AP to switch antennas to get the best signal.

Diversity is indeed what it is. But diversity requires some way to switch to which ever is better, based on some difference that can be detected. There are *many* different diversity systems, all of which fit some but not all circumstances. For example, the system used by 802.11 wireless radios works well only because the data consists of digital packets that can be stored and retransmitted, even out of order, if errors happen. That is one of the reasons that while the maximum data rate might well be 54Gb, you are never going to get throughput approaching that rate.