Ryder, as levels of overkill go what you propose is overkill OF overkill! (.and yes, I do acknowledge that your input is from 'WUG background 'n these are the tools you know to use)
..I mean, 12dBi yagi's to FIFTY metres??! I've used those things (in my Sentech HGA work, which skews things somewhat, but still..) to talk to towers 3-5 KILOmetres away! Seriously, unless you plan on overcoming an olympic swimming pool length of co-ax to recover cable signal loss, there's really no need to go to that expense ..or, for that matter, visual eye-sore/evidence of Doing Stuff That May Be Nefarious(tm)
At that range, if we're even going to do HGA action of any kind, I'd use li'l 8dBi flat panels - at 8cm on a side they're nigh on invisible to the untrained civilian eye
..which should neatly avoid problems with body corporate busybodies! And, at 8dBi they're STILL overkill for that distance, but at least they're damn small and more than enough to do the job.
And then, a router on each side? I can understand using one, or even just an AP, really as wireless NIC's calm right down with an AP to talk to; instead of chattering madly back and forth like co-dependant kugels out of their comfort zones, but still...
fergus gives us FAR too little info (..that'd be a hint!) as to what hardware's where, as in: is his/either PC by a window, such that his WG111/a USB wi-fi adapter can be put on the end of 5m worth of USB cable? (that being the spec'd max length one should use USB at; it *may* be possible to get more, YMMV). If so, it may take no more than putting 'em on the end of a long USB cable and leaving it at that!
If he wants to go the co-ax route, a PCI card would be preferable as there's that nifty li'l reverse SMA connector one could put an 8dBi panel onto, on the end of some length (again, input please fergus!) of cable. Also, he'd be able to get full wireless-G speed (with an external antenna, at any rate)
And then we could get *really* creative!
..I googled for "hack WG111 external antenna" and landed up on netstumbler.org, specifically on this discussion:
http://www.netstumbler.org/archive/index.php/t-16003.html titled "WG111 soldering fun!" ..check it out, specially the posts by Phunky Monkey - and definitely his post of [06-04-2005, 09:17 PM] where he posts links to pix of how he HGA'd his WG111!
fergus, for the questions you did have the presence of mind to ask (hint, asking good questions (like these) is notatALL silly and is the preferred method of getting a geek to open up!
)
1. antenna gain is a calculated as a logarithmic function ..which sounds horribly complicated, so ignore that and just remember: the gain (amplification ability) of an antenna is doubled or halved every 3dB you go up or down. IOW, if you have your reference 2dBi antenna (the thing you're likely to get on the back of a PCI card) and put in a 5dBi dipole (looks like a bigger stick, really); you'll have put in a part that has twice the ability (in terms of sending a signal farther, as well as being twice as sensitive at picking one up). If you then put in an 8dBi omni (staying with the one kind of device here), you'll have doubled up *again* - so you're now 4x stronger than you were with the original one. PS, this thing often looks much like around 40cm of white-painted broomstick-sized rod.
2. yes, but not as you may think. You can achieve "something more powerful" by tossing out the dipole/omni (which radiate pretty much evenly all round, not unlike a candle (..and so waste light/energy/radiation everywhere you DON'T want it!)) and using something directional (..put a reflector around 3/4(ish) of that candle so light comes out only in the remaining 1/4) instead. You're now passively (as in, not putting energy in to get the apparent amplification) 'amplifying' your usable signal.
3. for what you've told us, and as per my op-ed above: an 8dBi panel should be more than sufficient; hell, I've even got one of these somewhere near the Cresta region aimed at the MyWi Mintek tower- a distance of 2.59km, according to my GPS!
4. depends, more data please!
5. you want to focus your (radiated, i.e. off the antenna) energy ONLY in the direction you actually want to use it; 'lighting up' every place that's NOT the flat you're aiming at is a) wasted energy and b) possibly a nuisance to other people who may be wanting the ether for whatever floats *their* boat. This would make dipoles and omnis *bad*! Any of the 'focussing' types would do the job, but it now becomes a question of cost: sure, a 40cm dish would make for a *really* narrow beam (getting to be like the light from a torch, rather than a 1/4 circle of illumination) but it'd be HUGE overkill for your distance and are a hassle in many ways, too. Depending on the yagi (look at a TV aerial sometime, yagi's look a lot like that (unless TV aerials ARE yagi, or are they log-periodic? ..I forget)) you'd end up with a ~40cm grey plastic thing, or a ~60cm piece of metal -and again these are massive overkill for your job ...as least, going on the little data you provided.
As I said above, the smallest you can get away with (which has the pleasant bonus of being the cheapest too!) would be more than you need anyway, so I'd personally go for an 8dBi flat panel (which are, you know .. flat, square things!) Check
http://www.miro.co.za/stage/intranet/datasheet.php?id=50 for the part I mean.
..ahem. OK, I got *far* too involved in this one- time to go and do something that actually helps pay the rent!
HTH
-bdt
- well yes, I have this tendency (sometimes regrettably) to get (just a little bit, mind!) carried away ..and then it ALL has to be worked through or it backs up the system and I end up blurting out nonsene to people who have NO idea what I'm blathering on about!