Is there any chance you guys could give us some help on NV2 for real world PTMP applications? It seems like it is working well for PTP links but I know there are a lot of people who can't get it to work for PTMP applications, myself included. If you guys have tested this in a real world outdoor environment or know of someone who is successfully using it let us know the setup, board, wireless card, settings etc.

It seems like a lot of people trying to run it on XR2 cards are having no success with it, is there a compatibility issue with XR2's? I have done my part and sent in spouts from non working AP's with the NV2 package in both NV2 mode and 802.11 mode so how about tossing us a bone and helping us make your equipment/OS work to its potential.

chadd

If you have any real issues with NV2 on point to multipoint links. Please, contact support (support@mikrotik.com), we will need support output files from your router and optionally access to your routers.

chadd

If you have any real issues with NV2 on point to multipoint links. Please, contact support (support@mikrotik.com), we will need support output files from your router and optionally access to your routers.

I have been sending spouts to support and I am willing to give you access to these AP's but it needs to be real time as I am testing on a few small production AP's in the wee hours of the morning here in the states. This is normal business hours for you guys but I would need someway to schedule this with you because I can't leave the NV2 package installed on an AP because it isn't stable enough to run without having issues and causing service problems for our customers. I am willing to do what I can to help with this but there needs to be a better way for me to interface with your support on this rather than posting to forums or sending you an email and waiting for you to get back with me. The email support is not real time enough for me to leave NV2 installed and wait for you to log into the AP.

Thanks,
Chadd

What is your support ticket number?

What is your support ticket number?

The latest ticket is Ticket#2011011266000064

Some previous tickets

Ticket#2011010366000143
Ticket#2010121466000149

Please let us know how this progresses - we also have had no luck with NV2 in PtMP.

MT have given us a rebuff stating that we must have misconfigured or have some other issue.

When we place the links back into A rates - we have normal performance resume so i doubt we have misconfigured anything very much.

chadd, thank you very much for the cooperation.
We will see what we can do.

chadd, thank you very much for the cooperation.
We will see what we can do.

Here are a few screen shots comparing signal levels of V4.9 toV4.16 with the NV2 package installed. There is a huge difference in signal levels between the two, these are the same clients with no changes other than the OS version and NV2 package on the AP and clients.

If you go down the list the difference in RX signal to the AP per mac while running NV2 package is as follows

:AD 12db worse
:C4 17db worse
:E2 16db worse
:5E 16db worse

Average signal loss between V4.9 and V4.16 with NV2 package is 15.52db. That is HUGE!! Is there any explanation for this? Were signal levels not accurate in the old wireless package or are they not accurate in the new package or is there just something not right in the new package?

Also the noise floor is shown @-97 in V4.9 and -80 with 4.16 NV2 package.

The first image is with V4.9 installed on the AP running 802.11 10mhz




The second image is with V4.16 installed running the NV2 package in 802.11 mode 10mhz



Also with the NV2 package installed I have to leave the Data rates set to default otherwise the clients won't register. I do have country code set on all units so I am not sure what is up with that.

I believe Uldis is going to remote log into this AP to take a look at it.

We reported the same signal change issue to MT when 4.13 came out and got a response that nothing at all had changed in wireless and it must have been a bird sitting on the antenna when we ran 4.13 :P

Well I figured out what the deal is with the Signal levels and noise floor. If you change the noise floor setting in the Advanced wireless tab it directly effects the clients signal strength to the AP. Also whatever you have the noise floor set to in the Advanced tab is what shows up for the actual noise floor under the status tab and also if you do a frequency usage survey it shows the same number there.

The clients on this AP are currently "locked up" so they are not passing any data but are still associated with the AP. Hopefully you guys will be able to log into it and check it out. If you cant log into it for some reason please call me. My contact numbers are in the latest email I sent regarding the ticket.

Chadd

If you cant log into it for some reason please call me. My contact numbers are in the latest email I sent regarding the ticket.

When you are awake, we are asleep. We don't call, we are on the other side of the planet. Email is the preferred method.

If you cant log into it for some reason please call me. My contact numbers are in the latest email I sent regarding the ticket.

When you are awake, we are asleep. We don't call, we are on the other side of the planet. Email is the preferred method.

Well I see you didn't care for my reply,

but as I mentioned in my email to support I have arranged to stay up late to try to help get this resolved.

Uldis is working on your issues, please continue communicating with him.

D'oh, you guys DO know of the existence of Skype and the free calling all around the globe that comes with it?

On a related note - are there any news on this topic?
I would like to try NV2 in PTMP as well, but I am quite reluctant from what I read here :/

D'oh, you guys DO know of the existence of Skype and the free calling all around the globe that comes with it?

On a related note - are there any news on this topic?
I would like to try NV2 in PTMP as well, but I am quite reluctant from what I read here :/

Well from my testing if you have very good connections 4.16 NV2 works pretty well but it seems more susceptible to noise than either 802.11 or nstreme. On the AP's I have tested it on if a client doesn't have a SNR of 30db or better they have issues with timing related disconnects from the AP. For example the noise floor on one of the latest AP's I have tested it on is -97 any clients with signal strengths of worse than -67 have disconnect issues and clients with signals of worse than -77 won't hardly pass traffic. These same clients work fine with 802.11 and Nstreme so it is for sure something to do with NV2. I hope that MT can make some progress with NV2 because I expected it "TDMA" to do better with noise than either 802.11 or Nstreme. I think there may be some changes to NV2 in the latest RC that may help but I haven't tested it out.

As far as general issues I was having with the NV2 wireless package I think most if it boils down to the fact that the new wireless package has many changes in how the drivers interact with the wireless interface and when migrating from the standard package to the new package settings just don't seem compatible between the versions. Things that you used to do to get the old package to run well don't necessarily work with the NV2 package and in some cases flat won't work. Three of the biggest culprits IMO are ANI, Noise floor threshold and Data rates. If you are upgrading from the old package to the new package and have a noise floor threshold set from what I have seen you will have problems with it.

As far as general issues I was having with the NV2 wireless package I think most if it boils down to the fact that the new wireless package has many changes in how the drivers interact with the wireless interface and when migrating from the standard package to the new package settings just don't seem compatible between the versions. Things that you used to do to get the old package to run well don't necessarily work with the NV2 package and in some cases flat won't work. Three of the biggest culprits IMO are ANI, Noise floor threshold and Data rates. If you are upgrading from the old package to the new package and have a noise floor threshold set from what I have seen you will have problems with it.

Well I think I jinxed myself, about 10 min after I typed this one of the AP's that I had 4.16 NV2 package running in 802.11 mode when to crap on me. I played with it a bit but ended up rolling it back to the standard wireless package and it was normal again.

4.16 with nv2 crashes everytime I transfer at full bandwidth p2mp. This happens within 10 seconds on test.

I am now trying clients 5.0rc10 and ap 5.0rc10. I will let you know what happens. My problem has been kernel panics.

With 1 client and 1 ap I get kernel panic with 5.0rc10. This happens during bandwidth test within 10 seconds of it. I get kernel panic. Maybe bad wireless card? I sent supout.rif to mikrotik. From dallasweitzel@gmail.com

we are checking this issue right now. Maybe you have some newer board to try out instead of that RB532?

NV2 has been working wonderfully with:
RB800 AP
Ubiquiti SR71-15 or MT R52Hn
Router OS 5.0rc9
Station mode for dhcp hotspot authenticated customers
Station wds mode for customers with public IP's over a vlan
CPE's are RB711's running RouterOS 5.0 various rc's

We are seeing 30Mbps total throughput from the AP to 9 customers.
Great ping times, even when one customer is maxing out the AP.
It has been working well for a week (no Kernel crashes)

This is what I sent to mikrotik:

Broke is 5.0rc9 and rc10 and rc7. However I found 5.0rc10 with rb433 it works good. I have a lot of aps using NV2 and the rb532 boards keep getting kernel panic no matter what I do for settings. Is NV2 compatible with rb532 boards or the mipsle? I am still doing testing but if you use a ap as a rb532 you will see the problem. What are you thoughts? I want to help you fix this because I am putting NV2 on all over our aps. Thats well over 50 aps. I need this to work.

Dallas

My ticket # is [Ticket#2011022166000045]

RB532 boards do not work. Kernel panic daily. Anytime I use more bandwidth than 2 mbps it crashes.

NV2 has been working wonderfully with:
RB800 AP
Ubiquiti SR71-15 or MT R52Hn
Router OS 5.0rc9
Station mode for dhcp hotspot authenticated customers
Station wds mode for customers with public IP's over a vlan
CPE's are RB711's running RouterOS 5.0 various rc's

We are seeing 30Mbps total throughput from the AP to 9 customers.
Great ping times, even when one customer is maxing out the AP.
It has been working well for a week (no Kernel crashes)

Thank you for your information. I will tell my boss about your success.

Dallas

Anyone else have certain routerboards that work and done work with nv2?

I have a feeling this topic talks about NV2 implementation on 2,4Ghz networks? (XR2 cards, I don't know them but the "2" would mean 2,4Ghz?)

Anyway, I have a network in a valley consisting of about 12 A's and as many backhauls. The valley is not more then 15km's wide so basically although I can use 12 free channels in my region (5Ghz band) some freq's are double used, some even triple.
I have to do carefull planning against self interferences. But I managed fine in the last 5 years. Almost no interferences and low noise levels (around -100)

The last year I see some 3rd parties going to use same 5Ghz band so to stay ahead of the competition I planned to make use of NV2, also because ´hidden node´ is a big issue in my environment.

What I noticed now in the use of NV2 is that channels that normally would perfectly live side by side in relative close range (1 Km or some in same towers/buildings) and only a 20Mhz freq. distance, if I start using NV2 on some of these the frequencies in approxime range start having drops. I spend a day re-arranging freq. and I basically found that when I am able to get radio's in an environment where for 40Mhz on both sides (up and down) can be maintained in freq distance the problems are almost gone.

To me it looks like when using NV2 it is so ´aggressive´ towards nearby freq's that frequency separation must be at least 40Mhz on close range radio's not to run into problems.
This same would it then make almost impossible to use on the 2,4ghz radio band. The whole band is not that wide so the moment more then two channels in that band are used and one runs NV2 this would create big problems....

As long as you are the owner of the single 2,4Ghz running radio link that's not such a problems. But what if the competition does do the same? Or you have more AP's or Backhauls in 2,4Ghz that basically can ´see´ eachother signals....?

What struck me strange though is that the problems I had are at regurlar time intervals. Although links were very stable, high signal levels in both directions (-50- -65 range) with fixed low (24Mb airates) and near perfect CCQ's (95+) some of the links dropped every 40, 60, 180 minutes, or some hours only.
Move the freq's away from the other radio's freq's and no more dropouts.
The drops only lasted for some secs. After that links came back very stable again.

Also, remove the NV2 and same link also had no more problem.

I have a feeling this topic talks about NV2 implementation on 2,4Ghz networks? (XR2 cards, I don't know them but the "2" would mean 2,4Ghz?)

Anyway, I have a network in a valley consisting of about 12 A's and as many backhauls. The valley is not more then 15km's wide so basically although I can use 12 free channels in my region (5Ghz band) some freq's are double used, some even triple.
I have to do carefull planning against self interferences. But I managed fine in the last 5 years. Almost no interferences and low noise levels (around -100)

The last year I see some 3rd parties going to use same 5Ghz band so to stay ahead of the competition I planned to make use of NV2, also because ´hidden node´ is a big issue in my environment.

What I noticed now in the use of NV2 is that channels that normally would perfectly live side by side in relative close range (1 Km or some in same towers/buildings) and only a 20Mhz freq. distance, if I start using NV2 on some of these the frequencies in approxime range start having drops. I spend a day re-arranging freq. and I basically found that when I am able to get radio's in an environment where for 40Mhz on both sides (up and down) can be maintained in freq distance the problems are almost gone.

To me it looks like when using NV2 it is so ´aggressive´ towards nearby freq's that frequency separation must be at least 40Mhz on close range radio's not to run into problems.
This same would it then make almost impossible to use on the 2,4ghz radio band. The whole band is not that wide so the moment more then two channels in that band are used and one runs NV2 this would create big problems....

As long as you are the owner of the single 2,4Ghz running radio link that's not such a problems. But what if the competition does do the same? Or you have more AP's or Backhauls in 2,4Ghz that basically can ´see´ eachother signals....?

What struck me strange though is that the problems I had are at regurlar time intervals. Although links were very stable, high signal levels in both directions (-50- -65 range) with fixed low (24Mb airates) and near perfect CCQ's (95+) some of the links dropped every 40, 60, 180 minutes, or some hours only.
Move the freq's away from the other radio's freq's and no more dropouts.
The drops only lasted for some secs. After that links came back very stable again.

Also, remove the NV2 and same link also had no more problem.

I use a cavity filter and nv2 with rb433ah and I get actuall throughput of 10/10mbps at 10 miles. Its great.

Use filters. Its a must. The more the ap speaks the more the channel widens.


Do what I said and you will be able to use nv2.

I use a cavity filter and nv2 with rb433ah and I get actuall throughput of 10/10mbps at 10 miles. Its great.

Use filters. Its a must. The more the ap speaks the more the channel widens.

Well, 10/10Mb at 10 miles is not impressive at all. But the use of cavity filters sounds interesting.
Can you show me some suppliers for these and their specs? How do they fit on a radio etc.?

Why dont you think thats not impressive. 10/10 at 10 miles. I get 20/20 at 7 miles. In town I get 30/30. Plus I dont have to to rebuild my network again. Thus making more money. Try doing this with another product. It wont happen stably.

Well, I thought to have seen examples on this forum with even higher speeds. But ok, I might be wrong. Maybe not on 2.4Ghz?
I can't tell for myself. My longest link is only 7km (5 mile?) and in 5Ghz. It runs some 12,5Mb in both directions tcp and taken from one of the rb's itself towards one beyond the ´other side´.

But my interest is in these filters. To stay ahead of upcoming competition fine tuning my network keeps me ahead of them.
So far in google only found Streakwave has some in 5,8Ghz. But I am in Europe and we only use 5,5 to 5,7Ghz.
So guess I have to search for them. None of the wifi suppliers here have anything in that field so far I could find in a quick search.
So, which once's do you use?

Well, I thought to have seen examples on this forum with even higher speeds. But ok, I might be wrong. Maybe not on 2.4Ghz?
I can't tell for myself. My longest link is only 7km (5 mile?) and in 5Ghz. It runs some 12,5Mb in both directions tcp and taken from one of the rb's itself towards one beyond the ´other side´.

But my interest is in these filters. To stay ahead of upcoming competition fine tuning my network keeps me ahead of them.
So far in google only found Streakwave has some in 5,8Ghz. But I am in Europe and we only use 5,5 to 5,7Ghz.
So guess I have to search for them. None of the wifi suppliers here have anything in that field so far I could find in a quick search.
So, which once's do you use?

2.4ghz is noisy. 5.8ghz yes you can get a lot more. Its a bigger spectrum also. A better noise floor. Most wisp use 5.8ghz for backhauls. My 10 Miles backhauls get 73mbps half duplex at 10 miles.

l-com.com 2.ghz 8 pole High quality filters $125 each.

My ticket # is [Ticket#2011022166000045]

Mikrotik have seen the problem with the RB532 and NV2 not working together. They haven't told me a ETA yet. I will keep you guys updated when they get it working.

I am now trying clients 5.0rc10 and ap 5.0rc10. I will let you know what happens. My problem has been kernel panics.

like i mentionned in this post: http://forum.mikrotik.com/viewtopic.php?f=1&t=48073

I had issue with RB411AH + SR71-15 in 4.16 with package NV2 with WPA2 encryption.
It made RB411AH kernel crashes randomly.
As soon as i pulled off my encryption, the APs stopped to crashe.
I'm waiting news from MT.

For now, i'm using NV2 without encryption with 49 stations in 2.4 Ghz 20 Mhz of channel width.
We get 30/20 Mbps down/up in UDP Btest with almost all client.
Distance between AP and client is up to 5 kms no more.
Performance is better with this NV2 package, but stability as suffer much as client with little CCQ problem (due to multipath or interference) disconnect often and more than with the normal wireless package.
Those clients disconnect with "control frame timeout" message in logs.
Often the distance in wireless registration grows up to 15 km for those clients and we see a lots of variation on the signal strenght.

From the manual:

As Nv2 does not use CSMA technology it may disturb any other network in the same channel. In the same way other networks may disturb Nv2 network, because every other signal is considered noise.

How much is it sensitive ?
Do SOHO interior wireless router from client cause exterior pacwireless rootenna 15 dB disconnect ?

I will let you know if MT give me news,

Michael Plourde
Digicom

I am now trying clients 5.0rc10 and ap 5.0rc10. I will let you know what happens. My problem has been kernel panics.

like i mentionned in this post: http://forum.mikrotik.com/viewtopic.php?f=1&t=48073

I had issue with RB411AH + SR71-15 in 4.16 with package NV2 with WPA2 encryption.
It made RB411AH kernel crashes randomly.
As soon as i pulled off my encryption, the APs stopped to crashe.
I'm waiting news from MT.

For now, i'm using NV2 without encryption with 49 stations in 2.4 Ghz 20 Mhz of channel width.
We get 30/20 Mbps down/up in UDP Btest with almost all client.
Distance between AP and client is up to 5 kms no more.
Performance is better with this NV2 package, but stability as suffer much as client with little CCQ problem (due to multipath or interference) disconnect often and more than with the normal wireless package.
Those clients disconnect with "control frame timeout" message in logs.
Often the distance in wireless registration grows up to 15 km for those clients and we see a lots of variation on the signal strenght.

From the manual:

As Nv2 does not use CSMA technology it may disturb any other network in the same channel. In the same way other networks may disturb Nv2 network, because every other signal is considered noise.

How much is it sensitive ?
Do SOHO interior wireless router from client cause exterior pacwireless rootenna 15 dB disconnect ?

I will let you know if MT give me news,

Michael Plourde
Digicom

I was not using encryption at all. RB532 does not work. With the RB433AH I also do not get disconnect timeouts. Logs so far are clean. Clients from 1 miles to 14 miles away.

Mikrotik sent me 5.0rc11 and I installed it. It fixed the RB532 board from crashing! Mikrotik Rocks! I will let it run over the weekend.

Hey - I want my legacy boards to stop crashing on the rc upgrade too! Would love to have rc11!
Rod

Hey Mikrotik, is it ok if I give them the 5.0rc11 package since it works? I want to ask just incase they dont want it public. I dont know if they do or not. If they dont respond I will post it. Give them a couple days first.

Write mail to support and they usually give test version to try.
Every time when ask they give me test version, sometime you don't need to ask they send you link if you complain before for some problems.

thats how I got mine.

Hey Mikrotik, is it ok if I give them the 5.0rc11 package since it works? I want to ask just incase they dont want it public. I dont know if they do or not. If they dont respond I will post it. Give them a couple days first.

if anyone wants a prerelease, email us. we will usually have a new version anyway.

Hey Mikrotik, is it ok if I give them the 5.0rc11 package since it works? I want to ask just incase they dont want it public. I dont know if they do or not. If they dont respond I will post it. Give them a couple days first.

if anyone wants a prerelease, email us. we will usually have a new version anyway.

Ok, what will this be a v5.0roprc1.11? ("request-only pre-release candidate?) How many version can we have...

In a pre release you will have underscores I believe. 5.0_rc_11 or something like this.

Chadd - I see you have 4.16 running on RB133s. In our experience these boards will lock up under heavy use - you have to only install minimum packages and lock data rates down to no more than 18Mb on AP and clients. We have many RB133s working well with nv2 in PtMP environment, but this is in 5GHz band with R52s.

Yep I do the same and it works.

Me too, up to 27 133C's with 4.16NV2 (leave all three on CPE side, only NV2 on AP) on one AP in 5Ghz band.
All CPE's at default conn rates and set to either 18 or 24Mbps only. (Set basic rate to same)
Also only 4 packages on rb133C's (all with R52's); system, routerboard, dhcp and wirelessNV2.
This works relative good.

All my CCQ's are 95% or better and only on idle sometimes the drop. As soon as traffic runs on radio CCQ jumps to 100/100.

If two conn rates are set on AP, for instance 18 and 24Mbps (basic always on the lowest!) because one CPE has not enough signal to connect at 24Mbps, than all CCQ's drop a bit and ´dance´ in the 90 - 100% CCQ.

Thanks for posting

Chadd - I see you have 4.16 running on RB133s. In our experience these boards will lock up under heavy use - you have to only install minimum packages and lock data rates down to no more than 18Mb on AP and clients. We have many RB133s working well with nv2 in PtMP environment, but this is in 5GHz band with R52s.

Please tell us what config you use on those RB133C? Are you using Nv2 on them? What packages you have enabled on them? Have you tried to use RouterOS v5.0rc11?

Chadd - I see you have 4.16 running on RB133s. In our experience these boards will lock up under heavy use - you have to only install minimum packages and lock data rates down to no more than 18Mb on AP and clients. We have many RB133s working well with nv2 in PtMP environment, but this is in 5GHz band with R52s.

Please tell us what config you use on those RB133C? Are you using Nv2 on them? What packages you have enabled on them? Have you tried to use RouterOS v5.0rc11?

In my case I noticed that the 4.16 (dhcp, routerboard, system and wireless package) runs fine, the moment you install and remove (so not enable and disable..) the wireless package to get the nv2 package running the boards become slow... some worse, some hardly effected (rb133C's with same radio cards. So hardware equal).
In settings it makes no difference if the nv2 is really used or not. Some boards run as CPE in an normal 802.11 network and are slow, some work in a nv2 network and are slow.
But the most weird thing is that some boards don't seem to be effected at all, while others are really so bad they are almost unusable. All hardware is exactly the same and configs only differ in the used freq's and off course the IP's the boards are been given by the AP.

I have on one AP upgraded 9 rb133c's to 4.17 and about half is still as slow as a snail....
I have some rb122's upgraded to both 4.16 and 4.17 in my networks, all with the nv2 wireless package, and they seem to be less bothered by the nv2 then the rb133's.... One became slow, 5 others are flashing fast....

Going to find some suitable boards to upgrade to v5.0.rc11 soon...

Hi WirelessRudy I have been reading your NV2 post and saw u logged on, I am wondering if you have got ptmp to work correctly yet with 5.1. I am using rb433AH AP to rb411 client x2. 1 client is about 1km away from tower and 2nd is 16.7km set radius to 20. But I am getting that control frame and media access errors. I found a thread that talks about noise floor threshold and one about using filters... Any advice would be greatly appreciated. I using 20mhz 2.4ghz-b in this setup. I am lucky enough to be able to test these settings on real-life clients. Which you know that means I don't have free reign to start throwing freqs around.

Two remarks to make as follow up on discussed:

1. Upgraded almost all my units to ros5.1 now. Even 133C and 122 (!) boards that became slow improved considerably with the new 5.0/5.1 versions. They are not that fast as before with pre-nv2-package-era but definitely improved. I even start to use some old 133C's that were set aside again in my network. They get their 2nd life!

2. I have several of my AP networks now upgraded to ros 5.1 and on most of these now also nv2 running. I also ´hardened´ my network with more strict connection rules for CPE's to their AP's and use the security feature in nv2. My network is now better than ever before. Still have some minor issues here and there but so far I am relative satisfied with this new ROS.

rumiclord: a PtMP with only 2 clients is a bit of a special one. I have been reading something about it but can't find it any more. But I have two AP's with each 2 clients (which in themselves where yet again AP's with more clients, so almost always some traffic coming from them) where i have been struggling with same disconnects. But my deviation in the distance was not so big (7km versus 8km), where yours have much more variation. In your case I probably would use two separate links for that situation.
In my case the problems are gone after several changes on the link that I can't tell what actually established the link. Probably all bits helped...

I do have little experience in 2,4Ghz back-haul links. 2,4Ghz has so much limitations in the available spectrum that I don't see any advantage in using it, special for PtP.

Hi WirelessRudy I have been reading your NV2 post and saw u logged on, I am wondering if you have got ptmp to work correctly yet with 5.1. I am using rb433AH AP to rb411 client x2. 1 client is about 1km away from tower and 2nd is 16.7km set radius to 20. But I am getting that control frame and media access errors. I found a thread that talks about noise floor threshold and one about using filters... Any advice would be greatly appreciated. I using 20mhz 2.4ghz-b in this setup. I am lucky enough to be able to test these settings on real-life clients. Which you know that means I don't have free reign to start throwing freqs around.

I had frame timeouts on a customer. I upgraded customer and ap to 5.1. Still same issues. I found my customers was getting EMF from the fuse box. When the ethernet cables gets emf interference. It will go up to the mikrotik and lower the signal or it will cause the registration to cut out. After moving the cable away from the fuse box the issue stopped. This customer had a problem every 1-3 minutes. It was bad. Now works great.

Two remarks to make as follow up on discussed:

1. Upgraded almost all my units to ros5.1 now. Even 133C and 122 (!) boards that became slow improved considerably with the new 5.0/5.1 versions. They are not that fast as before with pre-nv2-package-era but definitely improved. I even start to use some old 133C's that were set aside again in my network. They get their 2nd life!

2. I have several of my AP networks now upgraded to ros 5.1 and on most of these now also nv2 running. I also ´hardened´ my network with more strict connection rules for CPE's to their AP's and use the security feature in nv2. My network is now better than ever before. Still have some minor issues here and there but so far I am relative satisfied with this new ROS.

rumiclord: a PtMP with only 2 clients is a bit of a special one. I have been reading something about it but can't find it any more. But I have two AP's with each 2 clients (which in themselves where yet again AP's with more clients, so almost always some traffic coming from them) where i have been struggling with same disconnects. But my deviation in the distance was not so big (7km versus 8km), where yours have much more variation. In your case I probably would use two separate links for that situation.
In my case the problems are gone after several changes on the link that I can't tell what actually established the link. Probably all bits helped...

I do have little experience in 2,4Ghz back-haul links. 2,4Ghz has so much limitations in the available spectrum that I don't see any advantage in using it, special for PtP.

I agree with you. You need to have strick rules on the cpe. All the cpes need to have the same settings. No rts or noise immunity either. Stable then.

Conclusion: Never put any network cables close to, or even in direct contact with, electricity transporting cables.
Probably lots of forum members also use cheap not shielded utp cable anyway and then it becomes even more important!

This might sound obvious to us more experience and knowledge members but I would not bet on how many of us as ´newbees´ would have used the existing tubes of the house electricity grid to run the network cables?
We have clients that are demanding us to use these since they don't want to see any cables...
So now we have to explain the why not of our refusal to do so....

I totally agree. Just because I tell the installer not to do something. Doesn't mean he is going to listen, lol. I wait until he cant fix something, then I go fix it and tell him what fixed it.

I agree with you. You need to have strick rules on the cpe. All the cpes need to have the same settings. No rts or noise immunity either. Stable then.

What do you mean with that? You don't use rts/cts (on 802.11)? I swear on it!
Noise immunity? You don't use it? I do because in the past is showed improvement. Not sure if it would still do but by default I use it all over my network.....

When rts enabled. I have tested this on rc's. If enabled on NV2 clients. NV2 will have crashing issues. I havent tested since. I just dont use it. NV2 doesnt need rts with tdma slotting.

When rts enabled. I have tested this on rc's. If enabled on NV2 clients. NV2 will have crashing issues. I havent tested since. I just dont use it. NV2 doesnt need rts with tdma slotting.

Yep same issue I saw with RTS/CTS on some units that I upgraded from 802.11 to NV2. I sent in an supfile to MT with the bug so it may be fixed now but I am not sure.

Chadd

When rts enabled. I have tested this on rc's. If enabled on NV2 clients. NV2 will have crashing issues. I haven't tested since. I just dont use it. NV2 doesnt need rts with tdma slotting.

Makes no sense. What do you mean with "NV2 clients". Is this a client running a wireless nv2 package or the fully embedded nv2 option in the latest 5.0series? Or does your client really run the nv2 option because the AP is set to it? (And CPE follows?)
When nv2 is running RTS/CTS is disabled completely. So that can not have any influence on nv2 running radio. The RTS/CTS protocol is replaced by TDMA.


chadd: Same counts for you. Which packages did you use. It is known that the initial ros versions running the early nv2 version had stability issues. For instance, even when running normal 802.11 in nv2-wireless package on 4.16 router had disconnects and on rb133C became very slow. Normal package of same version did not have that issue.
But all these issues are gone in 5.0 and 5.1. I use it now widespread over my network and it has never been that stable! Others reported the success as well and if you look for it, there are ample issues with respect of disconnects left for the 5.x series of ROS.

chadd: Same counts for you. Which packages did you use. It is known that the initial ros versions running the early nv2 version had stability issues. For instance, even when running normal 802.11 in nv2-wireless package on 4.16 router had disconnects and on rb133C became very slow. Normal package of same version did not have that issue.
But all these issues are gone in 5.0 and 5.1. I use it now widespread over my network and it has never been that stable! Others reported the success as well and if you look for it, there are ample issues with respect of disconnects left for the 5.x series of ROS.

This was back with 4.16 NV2. But like I mentioned I haven't had RTC/CTS enabled on anything since then so I may be fixed now.

I do know that the noise floor threshold and data rates still do screwy things with 5.0. I upgraded a remote AP/BH the other day and lost wireless connectivity to it because I had a noise floor threshold set on the BH and none of the clients would connect to the AP because I had manually configured data rates.

Chadd

I do know that the noise floor threshold and data rates still do screwy things with 5.0. I upgraded a remote AP/BH the other day and lost wireless connectivity to it because I had a noise floor threshold set on the BH and none of the clients would connect to the AP because I had manually configured data rates.

Chadd

Ok, I've seen some of you guys using this noise floor threshold. Why? I never used it nor do I actually see a reason. Or is it to keep weak roaming units away from your AP?
But on a backhaul there is no reason to use it. Leave it to default. On a backhaul you need a good signal with >25 s/n. If your backhaul is set under marginal conditions you still don't want it to drop because you set the noisefloor threshold too low. In my opinion it has no value here.
But if you have any other opinion about it please let me know, maybe I can learn something.
I have several backhaul links, most of them are with better than 25 s/n but one or two are marginal. I still don't use the nf threshold. If conditions are worsening the data rates (that are set for 2 options in this case, even with nv2) just drop one level and the link still transports data, only less.

Data-rates: Normal practise is to leaf these at default on the CPE. So the AP sets the rates.
It is also good practise to set only a few rates. In normal 802.11 that should be 3. One on the lowest rate to accommodate the basic support rate, and two higher ones. The highest one should be the one all client can make and keep, the next lower one is to cover conditions this high one still cannot be remained for whatever reason.
Thus in 802.11a for instance I set 6, 18 and 24Mb.

In NV2 running AP+CPE the stability seems to increase even further if only one rate is set. Yet again this has to be a rate that all client can follow.

If you now have a situation where all but one clients could follow for instance 24Mb but you wanted to set 18Mb as well to get that other weaker unit on as well, you'll see the CCQ on all units drop in NV2. If this is not harming the assigned connection speed of the client than in that case it is better to set data rate of AP to 18Mb. Now all links show better CCQ's and are more stable.

By setting the client to default data rates you also guarantee they can always follow AP if the signal strength is sufficient for the AP-set data rate. So in that case you would not have the upgrade issue you mentioned.

In theory what Noise floor threshold is supposed to do is to ignore any signals that are below what your threshold is set for. So if you are operating in an area with a high noise floor it should help with interference from other sources. So if your noise floor is -80 and your clients are connecting at -60 you should be able to set the threshold to -80 to ignore the noise. That would still leave you with a 20db SNR.

Locking down the data rates was helpful with 802.11b/g to keep the AP from constantly rate changing while working with different clients. It also help reduce retransmits once your AP figured out it couldn't send data to the client at the highest rate. The AP would continually drop the rate till it found one that the client could work with, that ate up a lot of overhead. The other benefit is to maintain higher transmit power levels. A card may put out 23dbm at 24Mbps rate and only 19dbm at 54Mbps.

[Ok, I've seen some of you guys using this noise floor threshold. Why? I never used it nor do I actually see a reason. Or is it to keep weak roaming units away from your AP?
But on a backhaul there is no reason to use it. Leave it to default. On a backhaul you need a good signal with >25 s/n. If your backhaul is set under marginal conditions you still don't want it to drop because you set the noisefloor threshold too low. In my opinion it has no value here.
But if you have any other opinion about it please let me know, maybe I can learn something.
I have several backhaul links, most of them are with better than 25 s/n but one or two are marginal. I still don't use the nf threshold. If conditions are worsening the data rates (that are set for 2 options in this case, even with nv2) just drop one level and the link still transports data, only less.

Data-rates: Normal practise is to leaf these at default on the CPE. So the AP sets the rates.
It is also good practise to set only a few rates. In normal 802.11 that should be 3. One on the lowest rate to accommodate the basic support rate, and two higher ones. The highest one should be the one all client can make and keep, the next lower one is to cover conditions this high one still cannot be remained for whatever reason.
Thus in 802.11a for instance I set 6, 18 and 24Mb.

In NV2 running AP+CPE the stability seems to increase even further if only one rate is set. Yet again this has to be a rate that all client can follow.

If you now have a situation where all but one clients could follow for instance 24Mb but you wanted to set 18Mb as well to get that other weaker unit on as well, you'll see the CCQ on all units drop in NV2. If this is not harming the assigned connection speed of the client than in that case it is better to set data rate of AP to 18Mb. Now all links show better CCQ's and are more stable.

By setting the client to default data rates you also guarantee they can always follow AP if the signal strength is sufficient for the AP-set data rate. So in that case you would not have the upgrade issue you mentioned.

In theory what Noise floor threshold is supposed to do is to ignore any signals that are below what your threshold is set for. So if you are operating in an area with a high noise floor it should help with interference from other sources. So if your noise floor is -80 and your clients are connecting at -60 you should be able to set the threshold to -80 to ignore the noise. That would still leave you with a 20db SNR.

Ok, I can confirm now that a set threshold is not working in ros5.1. Independent if the AP runs nv2 or legacy 802.11.

I presume this is because ´noise floor threshold´ is an atheros chip specific add-on where the new ros versios by MT that have nv2 in them embedded is probably bypassing this. Or they simply forgot to have it working in nv2 able packages....? I will ask the question is a separate new topic.

Locking down the data rates was helpful with 802.11b/g to keep the AP from constantly rate changing while working with different clients. It also help reduce retransmits once your AP figured out it couldn't send data to the client at the highest rate. The AP would continually drop the rate till it found one that the client could work with, that ate up a lot of overhead.

Hence you should set them fixed on the AP. Now the AP doesn't have to do all the hard work and the CPE should just follow. If the rates in the AP are now set low enough so all clients can establish that rate and high enough your clients can still have their promised speeds there will be hardly any rate changes any more in the AP-CPE network which improves the stability a lot.

For 5 years I worked with the approach that AP had all rates set and the clients where given all rates up to the highest they could sustain without ´step back´s but since some convinced me to do it the other way my network is much more stable.
But I have to say that my network consist out of fixed client antenna's. No roaming or moving units which would make the whole scenario different.

The other benefit is to maintain higher transmit power levels. A card may put out 23dbm at 24Mbps rate and only 19dbm at 54Mbps.

You need to explain this a bit more.
If I look in my AP that for this test purpose only the two lowest rates has set I still see in the "Current power" field all rates listed with their output power...
To me it looks the radio transmits on all these rates still...?
But, if I look in an associated client I see in the registration window under "Signal" only the two rates of the AP with a ´Strength´ graph. So now my conclusion would be AP only transmits on the two chosen data rates?

So, is the radio transmitting in all data rates, or only on the ones set?
In the latter I can imagine the radio would consume less power. But thinking in what you said it could be that if some of the higher rates are disabled the card puts more power in the left ones?

Ok, I can confirm now that a set threshold is not working in ros5.1. Independent if the AP runs nv2 or legacy 802.11.

Yep I have been saying this since they released NV2 for 4.xx and I have had several email discussions with them about it but they say it is working as it is supposed to.

Hence you should set them fixed on the AP. Now the AP doesn't have to do all the hard work and the CPE should just follow. If the rates in the AP are now set low enough so all clients can establish that rate and high enough your clients can still have their promised speeds there will be hardly any rate changes any more in the AP-CPE network which improves the stability a lot.

I set data rates on the AP not the clients, the clients will only use what you have the AP set to. Again there are some strange bugs with Data rates in the NV2 wireless package, if you have a mix and match of clients with the New wireless package and some with the old wireless package and you set data rates the old clients will not connect.

You need to explain this a bit more.
If I look in my AP that for this test purpose only the two lowest rates has set I still see in the "Current power" field all rates listed with their output power...
To me it looks the radio transmits on all these rates still...?
But, if I look in an associated client I see in the registration window under "Signal" only the two rates of the AP with a ´Strength´ graph. So now my conclusion would be AP only transmits on the two chosen data rates?

So, is the radio transmitting in all data rates, or only on the ones set?
In the latter I can imagine the radio would consume less power. But thinking in what you said it could be that if some of the higher rates are disabled the card puts more power in the left ones?

It should only be transmitting in the ones you have set but it will try to use the highest one.

Chadd

I have a small PTMP nv2 network with 4 clients. Initial tests yielded awesome results, TCP bandwidth tests maxed out at around 71-79mbps. However, during the daytime I started getting quick client disconnects. These would last around 5 seconds, but were enough to make our VOIP calls terminate.
-Initially I switched the frequency of the system from 5825 to a less crowded 5745. This gave us better CCQ (around 95%-98%) during late night testing, but again, we would experience multiple drops during the daytime.
-During the second round of troubleshooting I turned off the "adaptive noise immunity" setting on clients/AP and set the "Data Rates" to "configured"; allowed it to go up to MCS12. So far I've had no drops so far. CCQ is better now at 99%-100%.