Orinoco distance figures

[redfeaag]

I have just run up my first two long distance shots and, to say the least, am a little astounded by the results. For those who are interested:

Link one is the longest at 16.6 miles. Each end is using an Orinoco gold, 30' of LMR400 and 19" pigtails into a Pac Wireless 24db grid. NO AMPLIFIER! This came up with a signal quality of solid 10 - not briliant but enough to give an 11 meg throughput. The background noise is -83 and signal is -73. One end sits right next to a 300' TV mast with a zillion antenna on it which, I guess, explains the noise. STAROS distance setting is set to 17.

On attaching a 250mw hyperlinktech indoor amp between the pigtail and LMR400, I get a quality of 20. Belt and braces maybe but not strictly necessary I guess.

I just can't believe that 30mw can shoot that far. I have another link that can only just keep a quality of 5 and this hovers between 5.5 and 11meg.

These cards seem to cling on to the smallest of signal. I have pictures of the installs if anyone is interested.

[bobbyc]

What polarization are you using? If you're using vertical you might try horizontal to maybe cut back on the noise. Also you might want to clip the internal antennas off the orinoco card in case it is picking up noise inside the shed (if it's beside other 2.4ghz equipment.)
We have a backhaul that is horizontal and 13 miles. One end is 24dB Andrews grid and ~40' lmr-400. The other end is 24dB andrews grid and about 110' lmr-600 and it has a constant SNR of 20, -77 signal -97 noise.
Bob C

[lonnie]

That is a great test. Everyone has this fixation with higher and higher power. We see the extra power as a means to overcome an improper installation and poor design for a lot of guys. You got everything right, and the results are fabulous. :D Welcome to the club. You are on your way to doing what some guys before gave up on. We have made solid shots in spots where people had tried and tried and then given up and were saying 2.4 GHz and 802.11 is crap.

To me, the worst possible case is everyone shooting 200 mW around. We know that it will go 50 miles so a 200 mW radio messes up your chance to reuse the channel.

I am so glad you have done this. As a bit of added info, we have 500 mW amps, 24 dB grid antennas and we see -56 dB signals at each end of a 52 mile shot. This is higher than some guys are getting from a mile away. Does this raise any questions that maybe they have bad cabling or something else wrong?

[redfeaag]

OK, I may be stupid but I have to say that I listened. You have probably noticed that I have been asking loads of questions over the past month and have been taking the best of all advice from everyone. Things I have learned from this forum, and bear in mind we had never done this until a month ago:

1. Use Orinoco, and the Gold at that. Reason: They are the best rx card. TX seems to make no difference - its what the card can hear that matters, not what it transmits. On the Gold front, by using all-golds we can maximise security by means of RC4 encryption. Also, staros offers much more functionality for these cards - DISTANCE SETTINGS for example.

2. Keep the antenna as close to the radio as possible and use the best cable possible. Reason: less loss. We standardise on lmr400 - the best cost/performance - and max lengths of 30' This has led to some interesting installations (see below) but WORKS BLOODY WELL! Why have a 100' cable run? Build the PC into a weatherproof case and mount it on the mast. We have a source of DC-DC ATX PSU's for mast installations.

3. Antenna selection is important. Reason: By using all directional ants with as low a gain as possible we have minimised interference which has allowed us to re-use channels. We have one AP that has three antenna, all on the same channel with NO PROBLEMS. Mixing V and H pos obviously helps.

Obviously everyone has their own opinions. I have none of my own, only the best of yours! All I can say is that I listened to you all and am dumbstruck by the results.

[redfeaag]

OK, and now for a reply to the previous post:

I am using V-pol. I can't use H-pol due to the size limitations of the poles and our installation locations. A different channel would probably also have helped but what the hell, it is working.

The noise MUST be localised. Every other location I have has a background of 98/99/100 - there is just no other in-band users over here where I am. (lucky me)

Assuming the noise was -93, not -83, this would give a quality of 20. Unreal for something that has trouble spitting rf across a crowded office.

Whatever people say, you just can't knock reality. It exists. I used to be an athiest. I now prey to the great god Valemount. Seeing is believing.

[redfeaag]

Last rant for the day:

I forgot to mention a couple of other things that may be of interest. It seems that by using pigtails onto a standard orinoco gold, there is no interference between cards at all. On a 3 card setup, ch 1, 6, 11, I have noise of -98 on all cards and no attempts to associate with each other. The antenna are two sectors and a 24db grid. Somebody did say somewhere here that they thought that by looking at the circuit the on-board antenna is 100% bypassed by the external antenna jack. My experience seems to confirm this. I see no need to clip the antenna and risk damaging the card.

Also, all our CPE's are Soekris. I get the feeling that everyone else are doing things that I do not fully understand at the customer end. I am making the assumption that they are putting radios into windows pc's and hooking up external antenna.

Problems I forsee with this are driver issues, radio incompatability, wep issues, user stupidity and every install being different. As we say in the UK, a pig and a poke.

Obviously I have had the luxury of a clean sheet for my WISP design. By choosing the Soekris board - see the pictures of our design in the pictures section - we have standardised on a single, black box piece of kit. I care not what equipment the customer has - PC, Linux of IBM mainframe. I provide DHCP'd ethernet.

The decision for this was largely driven by the acceptance of keeping the distance from antenna to radio at a minimum. This allows us to use lower gain antenna and get better signal at marginal sites.

Other benefits include my ability to enforce an entirely orinoco gold network. Why? EVERY client runs RC4 encryption. Also, it is none of the customers business what is inside the 'black box'. If they don't know then encryption becomes harder to crack, particularly as a hacker would start by trying the usual WEP stuff.

[ddvzlnz]

Great compilation of info. You mentioned RC4 encryption. I don't know if you mean the RC4 used by "WEP". Just be aware that WEP has been broken (even at 128 bits) with free downloadable tools. This is only an issue if the link has to be secure.

gt

[lonnie]

Thank you Tony. You have done well. You could do just a bit better if you paid slightly more for the cable. I HIGHLY recommend LMR-400DB. It will only about 5 cents a foot.

The DB means direct burial and it has a sort of grease inside that repels water. Normal cable will actually wick the water in by the braid. We also use a double walled heatshrink for the cable ends at the connector. The inside wall is a hot melt glue and when the stuff shrinks it squeezes a bit of the glue and that means you now have a near waterproof cover and increased strain relief - the glue is stiffer when it cools.

We use bathtub caulking (silcone seal in a nice squeeze tube) to finish off the end when the cable has been connected. All that taping and mastic is so unnecessary if you do ALL of the above. Do a neat caulking job and it is perfectly waterproof as that is what they sell it for.

Keep water out and the connection will be as good in 10 years as the first day you did it. Not many people have followed all of this advice and from the lists I see they still have water trouble. We have connections from 4 years ago that have never been touched and I have always used this method.

[Anonymous]

Last rant for the day:


Also, all our CPE's are Soekris. I get the feeling that everyone else are doing things that I do not fully understand at the customer end. I am making the assumption that they are putting radios into windows pc's and hooking up external antenna.



I started preaching this idea some time back on a couple of the wireless lists. The soekris boards aren't powerful enough for the "one box does anything" it appears, and are a bit spendy. However, I'm in complete agreement on the idea that a complete computer running Star-OS can be the "one box" WISP setup. You use the same hardware in all places, just configured to do different things.

You can buy a fanless Via motherboard with the 533 processor on it for under $100 US. By the time you add a PS, weather-resistant case, ram, cf adapter, pc card adapter, you're up there close to the Soekris price.

Too bad someone, somewhere, can't see the possibility of the market, and mass-produce a board that has just the stuff needed for WISP use. Imagine the possibilities, when all you need to stock to build your ISP several of one box, licenses for Star-OS, variety of antennas and cables.

Oh, well... Nothing like being totally "out there", off in outer space, eh?

[redfeaag]

Don't get me wrong on the Soekris theme - these are CPE ONLY! I have also listened on the AP front. ALL of my AP's are GENUINE intel boards with a 2.4 P4 with the 500mhz FSB. The theory being no incompatibility issues. I have bought branded DDR and run 512 in each box. The case itself has the standard 350 watt PSU junked and a dual fan 550w fitted. The BIOS has all non standard bits switched off - sound, usb, parallel, serial etc. and a plethora of fans are fitted to keep the internals cool. The hard drive is a 7200 rpm ATA133 for the proxy and a 128mb CF for staros.

The principle behind this expensive approach is one of total reliability and performance. The CF allows for a solid state bootup. If my hard drive fails, so what. I just switch off squid and prepare my hot standby!

As far as the spec goes, it is obviously overkill. However, when I add 4 11.a or 11.g cards plus a couple of 100 meg ethernets, I won't be on this forum asking what spec machine is required for optimum throughput. I will be one of the smartarses with the advice.

I did consider paralleling PSU's but decided against it. the actual load with all the add-ons switched off and no monitor is less than 100 watts. The main thing I have found for reliability is a clean supply and again, I take no chances by using a 650VA APC.

[lonnie]

Just a word of warning about turning the printer port off.

Many of the support chip sets will generate spurious IRQ 7 when you disable the device. If you have a radio card or Ethernet that takes the free IRQ 7, you are dead in the water.

[georgew]

Great compilation of info. You mentioned RC4 encryption. I don't know if you mean the RC4 used by "WEP". Just be aware that WEP has been broken (even at 128 bits) with free downloadable tools. This is only an issue if the link has to be secure.

gt

Orinoco WEP Plus has not been broken.

WEP had a problem with weak initialization vectors would allow decoding of individual packets, and eventually cracking of the wep key itself. However WEP Plus throws away the weak values. This makes the WEP available with the orinoco gold's useful.

This does not get around other WEP limitations, and eventually there will be a successful breaking of wep plus, and all other encryption methods.

George

[ddvzlnz]

1. I was refering to plain standard wep, not proprietary extensions.

Proprietary extensions can be good, I use tkip and mic at home in with my Cisco Stuff.

2. When Star-OS turns on Wep, if there is a gold card, does it turn on wep plus?

gt

[tony]

WEP-Plus is automatically used on any and all Lucent cards that have firmware 8.10 or newer.

Thanks!

[ddvzlnz]

Well that's cool.

When in AP mode, when associating with silver and gold clients, will the ap do plain wep with silver cards and enhanced wep with the golds?

gt

[tony]

WEP-Plus is for all Lucent-style cards including Gold, Silver & Ruby cards. Gold cards only have the added ability to do 128-bit WEP like its lower-cost successor, the Ruby.

Thanks!

[dkii]

interesting, I was under the impression that only the gold cards would support WEP plus. Since I haven't heard of any brute-force methods to break wep, and it seems, at least in the case of the weak key exploit, that once 64 is broken, 128 is just as easy, then silver cards should provide sufficient security then correct? What about dlink's 256bit, is that still vulnerable to the weak key exploit?

[tony]

All WEP-Plus is, is a firmware component that makes sure no weak encrypted packets are sent over the RF, making it a hard, and long process to crack (though is still not perfect). As long as you have firmware >= 8.10 loaded onto your card for Client mode use, then WEP-Plus will be used. Our Lucent AP support includes WEP-Plus, regardless of the firmware version flashed on the card.

128-bit WEP takes much, much longer to crack than 64-bit in any implementation.

The new cards (802.11a/g) have a new security system, which includes strong AES (and other) encryption methods which cannot be cracked in the traditional ways, which will be good for a change.

Thanks!