Major Problems in Upgrade

[mrmike]

I replaced 2 wrap boards, 1 with 2 cm9s, and 1 with 2 senaos (APS) on the main tower with the new 4 radio war board, with new CM9s. I had 2 sector antennas channel 11 and 8, and 2 2.4 links, channels 1 and 4. This worked for 16 months, and then one of the wrap boards was degrading on one of the sectors. The replacement has not gone well. The 2 links are now on channel 1 and 3, at 4x cloaking, and the sectors are still on 6 and 11. The link on channel 3 is at -64, and passes through data quite well, but the other link is at -85, and the ping times are from 10 ms to 4 or 5 seconds. Up and down like a yo-yo. I have tried to lock the rate in at 6 and 12, but that hasn't helped any. I took off the extra settings, and put them back on.
The sectors don't have any throughput to speak of either. They are set to auto transmit, 12 miles, and default power. The customers that are at 100% will have throughput speeds of around 3k, to 150k, and others are 350k steady. There isn't any consistancy.
I have been up for the last 2 nights, trying to figure out the problem. We have replaces the board and the power supply. I'm at my wits end. This has been extremely discouraging, to say the least. Any help or suggestions would be appreciated.
The only thing I can try tomorrow, is to bond the grounds together, as this is on an AM tower, and there might be some potential problems.
I need some suggestions, please.

[go.fast]

You say the wraps were degrading over time?

I have seen instances of when I have had a high signal quality but terrible throughput. Changing channels has helped me bring the speeds up to what I was expecting to see. Another ap on the same channel can degrade your performance significantly.
So my first hunch is interference. Could be why the old ap was degrading over time.

But this is just guesswork to me. Too dificult to trouble shoot from here.

Have you tried shifting each ap over 1 or 2 channels to see if performance improves?

George

[mrmike]

I took out 1 of the links, and seperated the channels to 1, 6, and 11. Then I dropped the 2 APs 1 and 2 channels down. That didn't help any at all. I downgraded to version 1.0, and got a significant improvement over the 1.01 version. I then dropped the one link, and rearranged the channels a bit, and got somewhat better. One of the previous clients that were about 50k are now 300k.
I don't know where any interference may be coming from, around the access point. The APs are in the middle of nowhere, about 6 miles from town. I don't even see any other AP on the sight survey, that weren't there last year.

I took off WEP on the links, as they were cloaked, and I now have 1200k on the one link, and about 400k on the other. We're getting there, slowly but shirley.

[titan_wireless]

Mike,

Have you gone back to WRAP's on the AP's with the same channels as before to see if the issue goes away? If its noise you should get the same results with V2 WRAP's as with V3 WAR's.

[Bossman]

We're seeing some odd performance issues as well, but I don't know if it's due to upgrades. It's nice that with the latest build 1.0.1. b1115, we can see percentages.

What I don't understand is how one link performs so well while another has degraded.

Link 1 - 15 miles... used to be great, now... not so much:
Signal around -79 dbm to -81 dbm constant, noise -95 dbm 2x cloaking. Quality bounces from 100 down to *. This link was running beta 15 until last night and nothing has changed. This link has less antenna cable and more antenna gain than the longer link below.

Link 2 - 22 miles is newer and performs well.
Signal -81 dbm constant, noise -96 dbm 2x cloaking. Quality usually at 100.

What I don't understand is that if it is noise, I should see increased noise... if it's an alignment issue, I should see a signal level decrease, but I see neither. 2 links with very similar stats should perform similarly... or not?

Any suggestions on this?

[lonnie]

The fact quality drops is an indication you have interference. You should get a Spectrum Anayzer and see what is causing this.

[tim]

If either end of a link is a town (of any size) then you will see large amounts of interference if you use 2.4 GHz. A spectrum analyser will show this.

The solution is to upgrade to 5.8GHz, I am now seeing superb performance on all my 5.8 links

[nelson05]

In establishing a new link over the weekend, we encountered a situation similar to what you describe in Link 1. While our link was only about two miles, we were unable to get v3 running 1.0.1 to maintain a consistent link with anything other than 4x cloaking. We thought interference may have been the issue and the 4x cloaking shielded us, but when the issue persisted on the entire 5.x band, regardless of channel, we knew it had to be something else. Under the 1x or 2x cloaking settings, we would see something very similar to what you describe on the station end where the quality percentage reading would show 100, then 13, then *. This sequence would repeat multiple times all while we had a -75 or better signal.

Turns out that I had forgotten to set the distance at one end of the link (it was still set to def). 4x cloaking managed to establish a rock solid link with this setting missing while both 2x and 1x settings wouldn't function reliably (they could still pass data, throughput was just severely handicapped). Once I entered the correct distance settings on both sides, 1x and 2x came up without a hitch.

This may not be your issue, but you might try increasing the distance setting on each side and see if it makes any difference.

We're seeing some odd performance issues as well, but I don't know if it's due to upgrades. It's nice that with the latest build 1.0.1. b1115, we can see percentages.

What I don't understand is how one link performs so well while another has degraded.

Link 1 - 15 miles... used to be great, now... not so much:
Signal around -79 dbm to -81 dbm constant, noise -95 dbm 2x cloaking. Quality bounces from 100 down to *. This link was running beta 15 until last night and nothing has changed. This link has less antenna cable and more antenna gain than the longer link below.

Link 2 - 22 miles is newer and performs well.
Signal -81 dbm constant, noise -96 dbm 2x cloaking. Quality usually at 100.

What I don't understand is that if it is noise, I should see increased noise... if it's an alignment issue, I should see a signal level decrease, but I see neither. 2 links with very similar stats should perform similarly... or not?

Any suggestions on this?

[go.fast]

Turns out that I had forgotten to set the distance at one end of the link (it was still set to def).\

Good point, a question I have is, what distance is the default distance?

George

[tony]

Default distance is roughly 800m (802.11 standard default)

[rbolduc]

Default distance is roughly 800m (802.11 standard default)

That could be a problem that's less than 1/2 mile and you should allways add 1-2 miles on for good measure.

Reed

[mrmike]

I disables one of the 2.4 links for now, and rerouted that section onto my dsl for time being. I got the last link onto channel 1, with 4x cloaking, and placed the APs at 6 and 11. The links wouldn't stay steady. I then reverted back to ver 1.00 and the link stays steady. One of the AP' has serious lag time on pings, and won't let me push too much data through it (like 600k on a customer 2 miles away). The other AP is fine. I have changed the channels around, but that AP stays the same. I am going to change cards tomorrow morning, and see if that helps. If not, I am going back to the wraps and senaos.

I am also seeing the infamous 'N' connection of these APs (some are wrap boards), and they won't go away until I reset the system.

[ninedd]

I wonder if this is the same issue I'm seeing. I have a WRAP/Prism on a hilltop that's been there for 1 1/2 years or so. I also have two WAR's with Compex cards. The WRAP is 16' AGL and connected with LMR-400, while the WAR's are at 32' and 48' up a 75' tower, and directly connected from the pigtails to the antennas, which should be better. This is quite a few miles from town, and noise levels are -91/92ish on the Prism. I currently have 30 clients on that Prism AP. The WAR's report -96 on their Host card display, but there are no customers associating, so hard to say what they hear.

While driving around with a Compex WAR-Rootenna, I can easily connect to the Prism, even in areas that I know it's partially obscured by ground clutter. I can sometimes link to the WAR/AP's, but not always. When I'm close, it connects redily, and with similar performance to the WRAP, so it seem not to be antenna or pigtail. Even so, we've been up twice more to change antenna/WAR/radios/pigtails, and on the 2 WAR's that's LOTs of combinations. From anything past a couple miles, they show up in the Site Survey list with similar RSSI's as the WRAP does, but I usually can't link.

I'm sure I've tried all possible settings 5 times over, but maybe just missed the right combination. However, it seems I can link to the V2/WRAP with pretty much any combination. :) My latest settings were: 30 Miles, 23dbm, Short, Super and B/G on both the Client and AP. The WARtenna show the WAR/AP in the Site Survey at 4.2 miles at -78, but won't link. It also shows the WRAP/AP/Prism at -79, and links right away at 11Mbit/100%. Noise levels

I also tried all different channels of course and the best results I got were with 4x cloaking at 2462Mhz. With that, I connected fairly regularly, but past 8.3 miles or so, I could not get it to associate, even though changing the WARtenna client to 1x and the SSID of our Prism linked in seconds at 9.36 miles out. It showed -88 but a 100% Quality. The WAR/AP was still in the Site Survey at -87 when set to 4x, but I could not link.

It 'smells' like a distance timing problem, but I've got both sides set at everything from 10 to 15 all the way to 30 miles, so I'm confused. :)

[tog]

I've got 10 mile links running with v3 no problem, no cloaking...

Sorry, I can't be of too much more help than to say "works for me."

I haven't had a single strange issue like seeing an AP and not being able to associate to it and almost everything is still uncloaked right now. Everything has been working as expected. See an AP at -85, associate to AP right away, I get expected performance.

[ninedd]

I've only really used them with the Compex cards. Is that what you're using, or is it CM9's, or both?

[dc2005]

Is there any chance you could be affected by RF interference from other equipment nearby or on the same tower - not necessarily wirless LAN based or even using similar frequencies? I've seen CM9 cards in a Wrap stop passing data every time a nearby Ham Radio repeater on 145.625MHz keys up. This may not have affected your previous setup as the cards or AP setup may be less sensitive to RF interference. If this is the problem then the solution is to make sure everthing is well shielded i.e. use a grounded metal box. May not be at all relevant but probably worth considering.

[tog]

I am using mostly CM9s but I have deployed a few Compex cards with excellent results.

[ninedd]

Is there any chance you could be affected by RF interference from other equipment nearby or on the same tower - not necessarily wirless LAN based or even using similar frequencies? I've seen CM9 cards in a Wrap stop passing data every time a nearby Ham Radio repeater on 145.625MHz keys up. This may not have affected your previous setup as the cards or AP setup may be less sensitive to RF interference. If this is the problem then the solution is to make sure everthing is well shielded i.e. use a grounded metal box. May not be at all relevant but probably worth considering.There is some Shortwave stuff nearby. It's never bothered our 3 WRAP's running V2 with Prism or CM9 or Aries 5354 card. However, since the new tower is higher, I guess there's a possibility that it's hearing more interference than the old Prism one that's hidden in the trees.
The numbers on the SOS GUI don't show an increase in noise, but I'll see if I can duplicate / emilminate that. The real test will be to move the existing Prism AP up the tower higher, but that's the 'live one', so we didn't want to do that.

In years and years, we've never had any linking problems with V2 that couldn't be accounted for with just not enough signal, or an easily identifiable source of interference or something like that. In this case, the V3 boards see each other in the SiteSurvey, with nothing exotic (no WEP, no cloaking, no too short distance, etc.).

Just so I'm clear, I'm not thinking this is a fundamental V3 problem. :) I fully expect that V3 actually works! I'm just trying to brainstorm here and figure out why a link that visisble in the site survey can't link. The only time I've seen something like that in V2 was when we went past 12 miles and had to manually adjust distance, which isn't the case here.

[lonnie]

The only reason I have ever seen that a unit cannot link is because of th4e distance setting. Even with high noise it will link. Do you see the client scanning channels? When you say you had to manually adjust the distance past 12 miles, I have to remind you that with Atheros you have to adjust the distance. Period. Every time. Set to actual plust 2 miles. or even 10 miles more for a close shot. Just make sure it is not too low. It will not link if it is too low.

There is some Shortwave stuff nearby. It's never bothered our 3 WRAP's running V2 with Prism or CM9 or Aries 5354 card. However, since the new tower is higher, I guess there's a possibility that it's hearing more interference than the old Prism one that's hidden in the trees.
The numbers on the SOS GUI don't show an increase in noise, but I'll see if I can duplicate / emilminate that. The real test will be to move the existing Prism AP up the tower higher, but that's the 'live one', so we didn't want to do that.

In years and years, we've never had any linking problems with V2 that couldn't be accounted for with just not enough signal, or an easily identifiable source of interference or something like that. In this case, the V3 boards see each other in the SiteSurvey, with nothing exotic (no WEP, no cloaking, no too short distance, etc.).

Just so I'm clear, I'm not thinking this is a fundamental V3 problem. :) I fully expect that V3 actually works! I'm just trying to brainstorm here and figure out why a link that visisble in the site survey can't link. The only time I've seen something like that in V2 was when we went past 12 miles and had to manually adjust distance, which isn't the case here.

[ninedd]

The only reason I have ever seen that a unit cannot link is because of th4e distance setting. Even with high noise it will link. Do you see the client scanning channels? When you say you had to manually adjust the distance past 12 miles, I have to remind you that with Atheros you have to adjust the distance. Period. Every time. Set to actual plust 2 miles. or even 10 miles more for a close shot. Just make sure it is not too low. It will not link if it is too low.Hi Lonne. Yes, I've got distance set to 30 miles on all our WAR stuff. I've never been one to tweak the distance low to eek a few more K outta it.

That's exactly the way it's acting. I can see the AP in the Site Survey, but it won't link. It scans right on by. If I change the SSID to match our Prism AP, it links straight away.

The other settings are really defaults. Since it's WAR/Compex to WAR/Compex there's no reason to fiddle with settings, but I've still tried everything 5 times over. My latest settings were: 30 Miles, 23dbm, Short, Super and B/G on both the Client and AP. At 4.2 miles, my WARtenna shows the WAR/AP in the Site Survey at -78, but scans right by the channel. It also shows the WRAP/AP/Prism at -79, and if I change the SSID, it links right away at 11Mbit/100%. I don't get it. I've never had trouble getting anything to link with StarAP's.

In any case, thanx for the feedback. I'll drive back out again tomorrow with a Roo19 and fire up the GPS and get some better numbers of what will and won't link.

[lonnie]

On the site survey, what do you see that is close to same channel? What versions are you using? We have almost ALL of our AP units converted to either WLM-54G or CM9. All clients are also a mix of the two cards, depending what was on hand at the time.

Can you get me a login in to the units? If you can get a WRAP or something else on the Ethernet of the client I would like to play around with this.

[mrmike]

To fix my problem, I had to eliminate one of the 2.4 links. The new system would allow for the cross over, even though 1 link was on channel 1 cloak 4x, and one on channel 3, cloak 4x. the 2 Aps were on channel 6 and 11.

[dc2005]

To fix my problem, I had to eliminate one of the 2.4 links. The new system would allow for the cross over, even though 1 link was on channel 1 cloak 4x, and one on channel 3, cloak 4x. the 2 Aps were on channel 6 and 11. Can you explain a bit more i.e. which link did you have to eliminate to sort the problem? Presumably V3 allows more non-overlapping 2.4 cloaked channels to co-exist at the same site - but I guess all the channels have to be cloaked for this to work. We were planning to use this precise feature in a new installation, so I'd like hear a definite answer as to whether it works or not. Has anyone else tried cloaking yet with multiple channels at the same site? Of course there could be some overlap between the cloaked link on Channel 3 and the uncloaked AP on channel 6, so maybe that's what caused the problem.

[ninedd]

I'll take a stab at this... :)

A normal WIFI Radio on Channel 6 is really broadcasting to some degree on channels 4, 5, 6, 7 & 8. Likewise a radio set on Channel 1 is really broadcasting on 1, 2 & 3, as well as channels '0' and '-1' (for lack of a better description) that are below channel 1. In more techincal terms, each 'chirp' on nomal WIFI is 20 Mhz wide, so a transmission that is centered on 2437 is really going all the way from 2427 to 2447.

With 2x cloaking, the channels are 10Mhz wide, so something centered on Channel 6 (2437) is broadcasting from just 2432 to 2442Mhz, so it's really only stomping on channels 5, 6 & 7. This would allow you to run another 10Mhz wide AP on channel 9, and it's only broadcasting on 8, 9 & 10. With 4x Cloaking, channels are only 5Mhz wide, so you really could use all 11 Channels next to each other and not stomp on yourself. This may mean that if there is someone else broadcasting radiation on channel 6 and channel 11, you could use Channel 9 with 4x and probably escape unscathed between his inteference.

However, just because your cloaked doesn't mean you'll be immune to all interference. If your next door neighbor is on WiFi Channel 6, and you select Channel 6 with 4x cloaking, you'll still be right in the firing line, and it may not help at all. However, if you choose channel 7, or 8, it'll help to a greater and greater degree, because the lower end of your 5Mhz chirps are not hitting the center of his data.

I think the ideal situation is to have a tower that uses directional antennas with a fairly narrow beam width, which will eliminate much of the background noise, and then to use more radios on smaller channel widths.

While this might be an extreme example, in theory, you could have 3 - WAR4's, with a 5.8 Ghz Backhaul as well as 11 - 2.4Ghz radios connected to 11 - 30 degree sector antennas each on their own 5Mhz wide channel, and they would both be relatively issolated from the WIFI world, as well as from each other.

At least, that's my story, and I'm sticking to it. :)

[palmczak]

I'll take a stab at this... :)


I think the ideal situation is to have a tower that uses directional antennas with a fairly narrow beam width, which will eliminate much of the background noise, and then to use more radios on smaller channel widths.

While this might be an extreme example, in theory, you could have 3 - WAR4's, with a 5.8 Ghz Backhaul as well as 11 - 2.4Ghz radios connected to 11 - 30 degree sector antennas each on their own 5Mhz wide channel, and they would both be relatively issolated from the WIFI world, as well as from each other.

At least, that's my story, and I'm sticking to it. :)

It will be very interesting to see this in production. Since someone with a WIFI set up and 1 watt amp and a 12db omni can blast 6 radios off the air. The inference rejection and mitigation will be tested when PMP 4x cloaked systems get loaded and need clean environment to transmit in.

Joe

[ninedd]

It will be very interesting to see this in production. Since someone with a WIFI set up and 1 watt amp and a 12db omni can blast 6 radios off the air. The inference rejection and mitigation will be tested when PMP 4x cloaked systems get loaded and need clean environment to transmit in. JoeSure - that's always been the case. We've had OK results using channels 4 and 9 in some cases to get between all the LinSys's on 1, 6 and 11, but obviously there's still some overlap. With cloaking, there COULD be zero overlap, depending on what's in the area.