Hiya Tom,

I've seen a few things do this.  In order of likelihood:
Water in something.  INCLUDING a customer site.
Interference.  It could be self inflicted.  See above.
Bad radio.
Junk hardware that's not as good as the followers say or has previously unknown problems.

Even with fresnel interference I'd expect you to get very nearly calculated values.

Assuming that your amp is a 500mw version you should have 36dB at the wpop. With 24 dB grids on the cpe and assuming 50' of lmr400 you should have an rssi of -61. You are WAYYYYY short of that.

My experience is that for the short distances and type of technology that we run fresnel issues are normally not an issue. I even have links that have a tree IN the path at 8 miles and we still get good service to that customer. (note: he was supposed to chop down the tree but never did.)

I have seen interference cause low signals like that, but normally the rssi stays high but throughput sucks.

Just for kicks I'd try using some of the gear more likely to be found in the ISP world. Tanzeo or SB comes quickly to mind. Knowing things like fcs error rates etc. would probably help.

As an FYI, I try to build all of my links in the -65 to -75 range. This way you have a good fade margin but not so high as to get a lot of multipath (the real problem with fresnel issues).

Another thing you could try is to drop in another ap with a smaller antenna, say a patch of some kind, and point it at the furthest customer and see what happens. If a different ap fixes things up you know it's either interference you've not spotted yet (netstumbler misses almost all of the really important stuff) or you've got a bad hunk of gear.

Hope that helps,
Marlon
(509) 982-2181                                   Equipment sales
(408) 907-6910 (Vonage)                    Consulting services
42846865 (icq)                                    And I run my own wisp!
64.146.146.12 (net meeting)
www.odessaoffice.com/wireless
www.odessaoffice.com/marlon/cam



----- Original Message ----- From: "Tom DeReggi" <[EMAIL PROTECTED]>
To: "WISPA General List" <wireless@wispa.org>
Sent: Monday, December 19, 2005 6:03 PM
Subject: [WISPA] Inconsistent Signal strength


My configuration is: Station Server OS, Wrap board, Pacwireless 9 dbi Omni, CM9 (Atheros), RFLinx OEM 250mw Fixed Gain -802.11G AMP (800 mw if in B mode). Im using 2.4Ghz G mode.

The environment: The AP site is on a tall hill and has 4 subscribers, each spread out over 270 degrees surrounding the hill. Distances range from .5 to 5 miles. All sites ALMOST have LOS, but don't quite have it, with pine trees blocking the freznel zones severally. I can see the subscriber antenna from the AP in every case, but its usually looking through a hole in the trees.

Why I chose what I chose? The noise level was pretty high, and several WISPs were near, so I did not want to chew up spectrum by using a sectorial deployment, expecially with only 4 people to serve that would not use the full bandwdith of the precious spectrum. Thus I selected the Omni. Initially I used low power CPEs (13 dbm) and APs, but it couldn't survive the noise or get strong enough signal to combat the trees. So we upgraded to directional 24 dbi grids on the CPEs. This helped alot, but still was not good enough to get good links to 2 out of 4 homes. We attempted just adding a high watt TX card to the AP, and it did not solve the problem. Although it cut through the noise, it did not compensate addequtely for the low signal strength of the low TX powered CPEs. Although adding an AMP at the AP, amplifies noise as well as intended signal, we learned that it amplified the signal of the CPEs the amount necessary to combat the loss due to trees. High sensitivity cards were not the answer because the noise floor was to high. Of course, I could have replaced teh CPE radios with high power radios, but did not want to eat the cost of replacing the gear, when I rarely use Wifi. But the AMP was sufficient as it solved the problem and left us with all good links. The only problem is that the 5th house still just not able to get good signal. So we added a high watt range2 card (400watt) at the fifth house. All worked well for the rest of the summer and fall.

So whats my problem? Two weeks ago links started failing. The signal to noise ratios were to narrow, because the RSSI of the CPEs had dropped almost 15 db. So I assume it has something to do with the recent snow and Ice or movement of equipment. We investigate, and find nothing. We finally learn that it was interference and the channel we had selected. When we changed channels we were able to gain about 10 db back. However, all the links still had lower power than we calculate they should have. So as a precaution we thought we'd start swapping everything out just in case. We were embaressed we even needed amps and high gain antennas to begin with. Maybe it was a failing cable?

We found some odd results, so this brings me to the point of this post.....

When we changed channels, the signal strength did not change equally amount the 5 radios served. But what was the most odd, was that if we changed from antenna B to Antenna A, there was absolutely no change in RSSI for two of the CPEs. ZERO CHANGE. This of course was ODD, because the second antenna port on the AP's CM9 did NOT have an antenna plugged into it. However, on the other two CPEs we did see about a 5 db change in RSSI.

This blew our mind. How in the world could there not be a signal level change for CPEs when we removed a 9 dbi antenna, and 10 dbi gain Amp from the AP circuit? And only a 5 db loss for the other two CPEs? I just didn't believe it, so I sent a second tech to repeat the tests. Same results.

Just in case we swapped, all patch cables, pigtails, the AMP, and OMni, with all new gear. The same results were found.

Take note, the worst quality link is the one that is using the 400w Range2. However, I did not suspect that device, because it also was the farthest away link with the most freznel zone blockage. The CM9 on its own, instead of the range 2 in the CPE could not even get associated, so we went back to the Range2. However, I heard negative things about Range2's and being outperformed by 200 mw prism cards.

To clarify our current results...

The range2 CPE has -88 signal with 5-7 db over noise floor now, and surviving.
The middle link has -81 with 14 db over noise floor
The best link is -74 with 21 db over noise floor.
The links are all currently survivable as long as heavy weather does not come, and our neighbors do not start changing channels again.

Also understand, that a minimum of 5 miles in every direction around the hill is land owned by the client, so none of the interferrers are close neighbors.

So here are my questions....

1) Is there any protection on CM9s or Station Server, that does not allow the change of the ANTENNA PORT, when a change of the antenna port is made via the software selection in Station server? For example, if it detected an antenna was not on it, it would auto-switch back to the orginal antenna port? What I'm looking to confirm is, if my antenna is plugged to antenna B, and I switch to antenna A, am I definately without a doubt broadcasting and receiving off the antenna port A without the antenna isntalled?

2) Is it possible that the RFLinx amp is compensating for a difference in the signal output of the CM9, and not amplifying the signal proportionally? The model amp I chose should have been ideal for the power output levels of the CM9.

3) Any known problems with the Station Server drivers for CM9 or Range2?

One of my ideas just for troubleshooting was to individually substitute and try a 200mw prism card inplace of the 400W range 2 on the CPE, and to remove the AMP on the AP and substitute with a 200w prism card, just to rule out something wrong with the radio cards. However doing that, brings us to the B mode apposed to G mode. Two of the links were able to acheive high capacity using G mode. We wanted to maintain that because we were attempting some high bandwdith video applications across those links. It might be required to install two of the links with B and two of the Links with G using a dual sector model.

However, what I'd just rather do is understand the weird results we were getting, so I can avoid the situation in the future.

I know one way to fix the problem, and that is to replace everything with Trango 2.4Ghz and avoid the wifi headaches. But to be fair, and to not waste the money spent....

Any ideas?

Tom DeReggi
RapidDSL & Wireless, Inc
IntAirNet- Fixed Wireless Broadband
--
WISPA Wireless List: wireless@wispa.org

Subscribe/Unsubscribe:
http://lists.wispa.org/mailman/listinfo/wireless

Archives: http://lists.wispa.org/pipermail/wireless/


--
WISPA Wireless List: wireless@wispa.org

Subscribe/Unsubscribe:
http://lists.wispa.org/mailman/listinfo/wireless

Archives: http://lists.wispa.org/pipermail/wireless/

Reply via email to