> > thanks for your responses! > resending to the list - hope you don't mind > On Mon, 2003-09-22 at 19:40, [EMAIL PROTECTED] wrote: > > On the simple question a vs. b - forget a. It's analagous to running fiber > > to every apartment when you can carry 10mbps over cat 3 phone wiring, and > > your connection to the net is probably going to be 1.544 Mbps, > You're right we don't need the full bandwidth of 802.11a but given the > installation costs, hardware cost is not very important and if we're > likely to see higher rate for greater distance it might be worth doing. > (i understand a falls off very steeply). > You mention that 2.4 is noisy, but given 20 families with microwaves, > 2.4gHz phones bluetooth noise etc I thought it might be more > advantageous. (not many 5.8gHz phones yet)
As for 2.4 noise, my major customer base is people who live up in the mountains who have no other way of getting network access - they have line of site to literally hundreds of thousands of phones and microwaves (and they all summ up in power level at altitude) - I've learned tricks to deal with it (at first - I thought just like you did - I'm doomed to go 5.6/5.8!) - but we started some experiments (couldn't even SEE the distant signal on a spectrum analyzer at -137 dBm and 30 dB of gain in the antenna, but had a 5.5 Mbps link each way - it's amazing how those chips pull data out of pure noise). Don't get me wrong - I *LOVE* 802.11a - and I'd use it everywhere if I could, but it does have a few negatives that hit you when implementation plans get underway. I bought dual band 2.4/5.8 antennas for my backhaul links with the plan to go to 11a - never needed to. 1 - '11a cards don't emit as much power as 802.11b - they're designed for very close distances (50m) in an office where 100 Mbit e-net is being replaced. (at least that was the design concept) - power output on 11a is generally speaking lower than b, especially at same price points. 2 - Ok - so you're going to now fork over the $ and buy an amp for 802.11a. Shopping for 802.11b amps (we've all drooled over the 10W "export only" amps :) you get accustomed to 500mW, 1W, 10W, etc. 802.11a - lucky to find a 1W amp - and it compares with those 10W "export amps" for 11b. So now I'm starting to re-think this - if the amp is needed, I'm in the poor house. 3 - cards. With the advent of 802.11g, 802.11a cards may go the way of fiber in offices. Why spend more for a non-compatible card, when you can spend less for a compatible card that's almost as fast. Then I called a "friend of a friend" who works at intel (pre-centrino days) - he says forget 11a. Go with b/g. Intel is backing b and then g on the motherboards of both desktops and laptops under a super secret name soon to be anounced. 11a will die / fade into oblivion. 4 - power - but but but - 5.8 is SOOOO quiet when I go up there on my spectrum analyzer - and I even bought dual band antennas that do both 2.4 and 5.6/5.8 so I'm ready to go to 5.8 any time!... But then I did the path loss at the higher frequency. Power faster at 5.8 than 2.4. Alot faster. Add relative humidity and God forbid rain and path loss is really nasty at 7-10 miles. How much were those amps again??? Finally - since 802.11b was running with 47 dB of link margin (!) even with the summation of all the silicon valley noise in 2.4 and ricochet's crappy 1W 2.4 GHz radios on every poletop (we had 4500-5000 of those to contend with) -- there was no reason to migrate to 11a, even though funds had been set aside. 11b became the link. Even though I had planned for 11a to bd > I don't know about penetration of walls/windows of 5.8 vs 2.4 > 2.4 certainly penetrates walls/windows in our building since I can use > my own apple airport ap from the gazebo. 5.8 finds paths out that 2.4 can't fit through, but 2.4's normally overwhelming power advantage fixes that quickly. Since you have some 2.4 hardware - I'd take a weekend and move parts around the complex and see how far you can go with the =built in antennas. > > > If you _really_ want to be forward thinking then 802.11g will work with people > > who just have b cards, and still offer high speeds later. Realistically, > > when do you think you'll pull an OC3 into your complex? > never - 8Mbit is the max we'll probably ever install. I want to consider > the connection rates at the low signal areas. I don't know if it's > better to have no signal in some areas or variable signal in all areas > (or if a vs b would do that) > > > > > As for the antenna - it's not so much were you put the antenna as it is > > where the ANTENNA puts the signal. > sure but give that I don't want any wierd lobes i expect we'll use a > pretty standard omni ... Rooftop omnis with electrical downtilt aren't bad - without it, they shoot out to the hroizon. I have a professional (REAL professional - uses a 3" diameter pole mount) 12 dB (conservative) 2.4 G omni that is 4' long. I generally prefer flat panels (the good quality mfgrs don't have all the ugly side lobes) for tight spaces. The flat panels have optional downtilt kits you can buy to give them "mechanical downtilt" -- makes a big difference (all of the pcs / cellular installs are going to sector flat panels with mechanical downtilt now) If you didn't mind turning your gazebo into a cell site for 2.4 G 802.11, you could have one awesome antenna package that would punch through all 3 floors of all buildings. I've been brought in as a mediator between technical antenna installers and city planning commissions and this is usually where things break - the best antenna looks ugly, and the cute tree-patern antenna has lobes everywhere but the population center. > > > > > "professional" can mean anything - just like "deluxe" or "new, improved" - > i was looking at the dlink gear because it puts out 200mw instead of the > 30 that their consumer gear puts out. 200 mw is becomming a standard in high power cards - senao makes one also - look back a few threads and you'll see a discussion we had about these so called "high power" cards - they're honest about the output power (FCC requires that!)- but their receive performance spec (if any) is on some cards highly (10 dB+) exagerated. You end up with a net negatve (-7 dB) link margin, but it's a great marketing tool! (WOW! this one has 200 mW thats TWICE as GOOD as the 100 MW - [3 dB to the engineer] - I'll buy the high power one!) (usually the 200 mW is located in a bright vivid color on the box and in a prominent position on the website, with a big star drawn around it to get your attention). If you really need power - buy an amp. 100mw in - 1000 mw out (+10 dB) - and many amps have a GaAs FET pre-amp - 10 dB boost on the receive side - for a link margin theoretical improvement of 20 dB, versus the 200 mW card's theoretical 3dB, and often actual -7 dB or worse depending on how they built their receiver's front end. In all cases, I emphasize theoretical - the amp boosts the noise as well, and creates intermod products depending on how good the amp design is, so picking a good amp can't be done by the numbers. With all this put together - I ran an 802.11b link from Cupertino, CA (black mountain) to south San Jose (85/Almaden) - a distance of 14.5 miles as the GPS flys - 11 Mbps link speed, solid connection, 1ms latency. Through the worst 2.4G interference silicon valley had to offer. It's all a matter off buying the right hardware to do the job. > I suppose they could be lying but... Politicians lie. Marketing staff selectively emphasize the partial truth. > Is there any sort of conduit between > > > the buildings that you could pull cat-5 as a wired link between two > > APs, one > > on each building? That would be probably the cleanest way to preserve > > spectrum. Otherwise, you can make a spanning tree link out of APs. > > I've > > done these - they really get solid coverage. > Unfortunately I can't run cable between the buildings > Well - pick APs that have a spanning tree capacity and a best route feature. Cisco's do this - the first AP (the one plugged into the landline circuit) knows it's #0 - the best one. From there you place #'s 1,2,3, etc - they all know the route back to #0. And if the route changes as someone moves, well, that's fine - the APs see it, and it's transparent to the user. It's a killer design - I use AP350's on sectorized antennas ontop the mountain to cover a wide area. OOnce ina while a customer "hops" onto another access point (and they log this so you can see what is going on). You can also do link tests to customers without them knowing it, which helps you figure out the contour lines of your network. It's a well engineered system. I'm sure other vendors have done similar designs - I just pick cisco because of their RF performance (In fact I think there are better designs from a support standpoint than cisco - if I didn't need to squeeze the last drop of RF path out, I might try them instead) > > > > When you place the cutsheet radiation patterns on the cad drawing, you'll > > see who is going to get service and where your 20 dB, 10dB and 0dB link > > margin lines are. > thanks for these ideas - i didn't realize that these diagrams were so > accurate. EXTREMELY accurate. People plan 1000' tall towers based on Decibel's or Andrew's cut sheets. They want a beam to serve a city 5 miles away - a small error could multiply quickly. Most of their antennas come with computerized test results to prove exactly how that serial number performs. > > > Is your objective to be able to have complex wide laptop service, or indoor > > laptop service so people can couch potato it and surf the net with their > > 802.11b laptop? > well - both > mostly it's so we can share the cost of a network connection > but i think people will like using their computer all over the place. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ This one is going to require a solid RF design. It's easy to stick an omni on the roof and make people buy yagi / dish antennas to point back at it (we all do this from time to time on this group) - but getting laptop solid coverage in a large venue takes planning. > > > Your signal levels will be substantially different. You > > may have to set expectations with users who just bought the $3000 laptop > > with the builtin 802.11b card! (alot of the builtin cards are worse than > > the cisco/orinoco external antenna-fob cards - they built the antenna inside > > a faraday cage! - but it works the 100 feet the factory specs.) > I've experienced that first hand... > I want to put out enough signal and have enough receive sensitivity so > that we have few to no dead spots. > You will win or loose with the antenna design - the boxes are not the limiting factor for that level of saturation. It will be nice to roam around and stay connected anywhere! > thanks again for your detailed responses! > Good luck with your project! Everett > brad > > -- general wireless list, a bawug thing <http://www.bawug.org/> [un]subscribe: http://lists.bawug.org/mailman/listinfo/wireless
