Nstreme link scenario and questions

Great stuff MikroTik, keep it up. First post to the forum everyone so sorry if I’m a bit noobish here… I am currently looking to deploy a wireless backbone using MikroTik RouterOS and associated hardware but I have a few questions regarding Nstreme. I have played with the Nstreme 2 or dual mode and it seems work rather well but I have some major limitations for my link and I’m hoping someone here can help solve them.

When this link was first planned we though we could go 58km on a single hop using 2.4 GHz 30 dB dishes on each end, and we can, it works (finally after discovering MikroTik) but there are some serious fresnel zone issues with a small hill in the middle and one end of the link at a location where the spectrum is highly saturated. Consequently, we have moved our raidos down into the US amateur band as we do have amateur licenses and are avoiding the majority of the interference.

Before we discovered MikroTik, we were using a site on the hill in the middle with 24 db dishes pointed each way rebroadcasting the signal (we actually still are until we get a perm MikroTik solution in place, we just tested it last week). With MikroTik in single Nstreme we were able to make the entire link in one hop which had never been done before, but the throughput is not as high as we will require in the future. The link is using 1 Watt amplifiers which I am quite sure is one of the major sources of problem of throughput as they are time division design. By going to Nstreme 2 would this be eliminated as the amp would be only on the Tx radio and never have to switch modes?

Without the site in the middle amps are required for the link to work and have any fade margin as our radios are all located indoors and the cable runs are 25m and 50m respectively. We are trying to get away from the site in the middle as it is in a very very remote location with solar power only and extremely inaccessable in the winter time. Here in Colorado, it’s tough enough to get some of these places in the summer much less with 80 mph winds and snow. We are worried there won’t be enough power at that location to run four radios to do the Nstreme 2 through that location.

Here is a rough layout with some info that will help responces. Thanks in advance.

Site 1 (South):
30 db dish located at 22m (Top of the tower) with 1 Watt amp
25m LMR-600
Staturated radio spectrum on the back side of the dish (site overlooks a populated area with serveral WISPs)

Site 2 (Middle) @ 28km from South, 31km from North:
Crest of the hill between the two sites, clear fresnel both ways
2 x 24 db dishes
~10m LMR-600 on each dish
Saturated radio spectrum on south facing dish (Channels 1, 3, 5, 6, 8, 9, 11 between -77 and -84 signal on each)
Power is solar/battery/DC Generator with barely enough power currently to run 2 low power radios.

Site 3 (North):
60m tower with 30db dish @ 45m with 1 Watt amp
50m LMR-600
Weak signals on the the other channels (-85 and lower channels 1-11 as the hill in the middle blocks the broadcast location of the WISPs)
Marginal fresnel to south site, great to middle

RX -63 from end to end but throughput is only ~7 Mbits both ways tcp with bandwidth test tool.
With the middle site RX is -54 both ways but only ~14 mbit throughput in bandwidth test.
ACK was tested both dynamic and static with the above results recorded as best case.

Based on the above layout, what’s your input everyone? I’m sure a a lot of our issues revolve around the amps but I don’t see how we can fix that and keep the gear located indoors. Outdoors is not very feasible due to extreme weather (-25F winter temps, +110 summer temps, extreme lightning, etc..) and the few number of days per year the towers are climbable. If there is an outdoor equipment failure it could be days before we have weather permittant to climb to the gear. Can just standard broadband 2.35 ghz amateur TV amps (staying well below the 1500W PEP) be used on the TX radios if Nstreme 2 is used or do I not understand how Nstreme 2 works? This would allow enough power to reach the dishes to have all the gear inside and get away from the crapo bidirectional amps that are boosting our noise floor but I’m not sure if it will work, that’s why I’m asking you guys. Or are we better off to just invest serious $$$ into the middle site and put up a large solar array to provide enough power…

Thanks again.

  1. loose the amps, they are going to be the source of endless problems, try ubiquiti’s XR2 radio’s

  2. move the equipment outdoors, power it by PoE over outdoor rated shielded, gell filled CAT5e cable. MT boards are designed for pretty extreme weather, as long as you have a good grounded metal (not plastic) enclosure for the box, it’s unlikley you’ll have any problems.

  3. if you’re worried about redundancy, do not use NStream 2, setup 2 individual NStream links, with OSPF and configure them for full duplex routing by manipulating the interface costs, that way if one link does drop, it will nearly instantly fall back to half duplex communication on the remaining link, yet when functional you have almost the same full duplex effeciency as the NStream 2 link.

Aside from your Nstreme question I would like to make a suggestion which was made to me a couple weeks ago by a buddy of mine In Texas pertaining to escaping some of your busy 2.4 spectrum.

Set the Country Code to Japan and try using Channel 13 on these links.
I had a 2.4Ghz AP that I could attain a -60 Signal Strength from at 4 miles out but the CPE wouldn’t connect due to the extremely noisy environment created by an irresponsible operator in our area which uses one channel and walks on the other 10.

After setting my AP’s and CPE’s Country Code to Japan and using Channel 13 my CPE’s connect reliably solid at 24Mb.

Just a suggestion though…


Have a good one.

Regards

  1. if you’re worried about redundancy, do not use NStream 2, setup 2 individual NStream links, with OSPF and configure them for full duplex routing by manipulating the interface costs, that way if one link does drop, it will nearly instantly fall back to half duplex communication on the remaining link, yet when functional you have almost the same full duplex effeciency as the NStream 2 link.

Can you expand on this? We are just about to set up NS2 to backhaul a DS3 to our primary tower, but redundancy is very important for us, and your idea sounds logical.

We have our fiber point 2 miles from our NOC, so we have a two seperate links, using OSPF for fallover.. Works flawlessly. The primary link (R52 cards in mini-itx system) holds 35mb tcp each way, simplex. About to switch to Nstreme dual to get full duplex. The backup link is two RB532’s. Then an RB532 is on each end with ether2 and ether3 going to the wireless units… The primary link has never gone down to my knowledge on it’s own, but we’ve taken it down numerous times to test the OSPF fallover…

The OSPF setup was done by Wisp-router for us.

Isn’t this illegal in the US?

expunge, get rid of the amps. They just add noise.

We are running routerboards at -40F in Winnipeg without problems.
Also, consider 5.8 GHz.

I’m quite sure it is illegal in U.S to utilize Ch 13…

First you have to be noticed/caught using that frequency which is slim to none in most remote locations.

However I am surrounded by other operators which operate well above the allowed FCC EIRP considering the power output of amps they use in conjunction with Omni antennas.
I don’t see these guys getting taken down by the FCC even though they tend to walk completely over the other remaining 10 2.4Ghz channels in 20Mhz channel-width settings.

Well we are already running channel 13 on the link between the south end and the midpoint because if we run anything else the throughput goes to hell. This location is remote, I mean really remote so we aren’t worried about being caught and we’d have to be causing intereference first but we’d like to get to a legal operating mode because I have a feeling that we’re going to piss off some of the local competition when we get this setup and they’ll look for any excuse to get us shut down.

As far as the outdoor gear goes, we’re also concerned about throughput, I’ve read that most of the MT boards are bandwidth limited and we’re looking to get a DS3 at one end of this link and be able to deliver to the other. I do like the idea of two seperate links with OSPF, can you get the configs of your setup so myself and other like DesertTek can attempt this?

How close do these radios come to running as they are advertised? I know they state like -72 @ 54mbit but we’re running in the -50’s and only seeing 18/24 on our links. Could very likely be from the amps but I’m still wondering. In your expert opinions everyone, would you just move all the gear outdoors everywhere? 600mW radios, bandpasses, a few feed of cable and ditch the amps? Are bandpasses even really that beneficial in this scenario?

Thanks for the replies

Expunge…

I have a suggestion that just may be an economical solution to your problem-
You are already running on Ch13; so this means you may or more than likely are at present time the only user on that Ch??

OK, what I would suggest you try is an 8 cavity 2.4GHz Outdoor Bandpass Channel Filter.
You would install one of these filters at each affected location in the P2P link.
http://www.hyperlinktech.com/web/bpf24-8xxa.php

The BP Filter needs to be purchased for the Ch you intend to use, one important thing to keep in mind when using the filters though, the effectiveness depends entirely on you being pretty much the only user of that Ch/Freq.
If you see other users of that Freq you would need to see signal quality no better the -90 coming from users other than yourself for the filter to do it’s job most efficiently for your need.

Also, make sure and use NO AMPLIFIERS, this will only intensify your problems by amplifying the unwanted noise from other operators.
Be certain that you are using as narrow beam-width antennas as possible on your links too.
Another thing to do for link antennas if they are parabolics is to do as we used to do for big ugly dishes affected by Out Of Band Interference.
Block side-lobe interference by shielding the side-lobes of your link antennas, this affords greater protection from the noisy environment and provides much cleaner C/N levels for you.

Regards

Been there, tried that. We’re using a pair of the Hyperlink Ch13 filters right now and they aren’t helping much. The width on them is too wide so they don’t attenuate anything above channel 9 very strongly. With 2472 (Ch13) as the center freq and the filters being 22 MHz we are having trouble because a WISP in the area is running on Channel 11 and killing us. There is about 3 db of insertion loss as measured in RouterOS with the filter in place and we cut about 10 db off the RX level on the noise on Ch 11. It cuts 3 db off our own signal and 10 off the neighbors, not to mention it cuts the through put for some reason, either they have a bad SWR or the side lobes are too steep and cutting down some of the bandwidth but we loose about 2-4 mbit/sec through the bandpasses. I am going to be able to play with this more this weekend and we will remove the amps from the system and see what happens (weather cooperating). Until we get some outdoor radios with the XR2’s we can’t play too much as the cable loss will kill us… We are using the Hyperlink 30dbi dishes on both ends of the long link and 24’s on the site in the middle as explained above. Side lobes aren’t an issue, on our south site we can’t even see the radios that are behind the dish but the north end sees our south site and all the other sites as the towers are within 2 degrees of in line with each other.

The amps are about the only thing I can think of to clean this up without going down into the 2.3’s and getting some custom bandpasses and stingers for our dishes. Going to 5.8 would be an option once we get some outdoor radios… Thanks for the input though giganet, any other ideas?

Hmm, bummer…

I would then strongly suggest to first place your radios within no more than 10’ of their respective antenna.
Make sure and run VSWR confirmation on each and every cable 1.5:1 or better.
Convert the MINPCI’s to 5.8Ghz.
Replace your current 2.4Ghz antennas with PAC Wireless 28db parabolics.

I know all too well that this isn’t what you want to do or hear, but I have been in the exact same ship already!
I had nothing but 2.4Ghz gear and due to another operator I can’t even connect to any of my towers from as little as 3 miles out with -60 signal strength, he also killed my inter-tower P2P links.

SO I converted my entire system to 5.8Ghz for AP’s and P2P.
Just last week I deployed 900Mhz AP’s to 4 locations to also provide NLOS connectivity in addition to LOS area coverage enhancement.

Regards

Thanks for the reply giganet, we talked last night about this very same thing and I think that will be a plan here in a few weeks. This weekend we are going to put the 2.4 link back up with the amps and push it up to 2484 using the superchannel license and hopefully then we will have 20 mhz from our center to channel 11 possible interference. It won’t be ideal but at least we can get a link between the sites again and hopefully then later this spring get 5.8 GHz gear deployed. Any reason for suggesting the PAC 28dbi’s? We were looking at the Hyperlink 32’s last night as possible dishes.. Are you using the XR5’s in your 5.8 links and how do they perform if that’s what you’re using?

You’re right though, not what I was looking forward to, those 6’ dishes aren’t too much fun to haul 150’ into the air but it’s not doing any good anyway so we might as well us it on another link. It’s interesting reading through the forum how many people in the US are experiencing so many issues with other WISPs way beyond power limits or saturating the spectrum with poorly designed installations and making it a nightmare for everyone. I guess that’s the price one pays with Part 15 gear and ISM bands…

Thanks again.

Heya Expunge!

So, in reading this post, I have to concur with a few things. First, 5.8 is the way to go if you have LOS. We too have a saturated field in 2.4, so we just don’t mess with it for backhauls. We use 5.6-5.8 for long hauls, and 5.1-5.3 for short hauls.

Second. I personally love amps. I find them extremely useful in REMOTE areas with little or no other 2.4 usage. I use CM9s in them, and we get great results in wooded areas. However, I would never use them on a backahul. I am not a subscriber of brute force, especially if I have bad neighbors/competition.

What routerboard are you using? I would say the minimum you want is a 532. Use a pair of XR5’s and use this antenna http://www.pacwireless.com/products/HDDA5W.shtml

We just completed our 4 mile DS3 Backhaul test. We used 4 R52s, dual Nstreme. RB333, and the above dishes. I am happy with the quality of the link… -56, 97/100 CCQ, but unhappy that the most I can get is 29Mbit full duplex speed. This will work for our needs, but we wonder what routerboard options will give us more. ( I know a PC at 2.4Ghz is better, but the 333s are mounted on towers )Pings are MOSTLY great, averaging 6-9ms. However, about every 20th ping we get a 40 or 50. We almost never drop a ping, even in our 24 hour ping test with 1024, we only dropped three. This is tested from one pc to another through the dual nstreme.

If this is an important backhaul, invest in it. You won’t regret it. And regarding your current cable situation, I see no problem retaining the LMR 600. Myself, I would prefer to service the routerboard at a comfortable level, instead of way up on the tower. We currently have a few with POE all the way up the tower, and following I have some photos of the adventures we’ve had with Mikrotiks. Not all pleasant, for sure! At least my business partner is adventurous

Here are some fun photos of us and some of our towers. http://www.desertgate.com/modules/myalbum/viewcat.php?num=10&cid=3

You still out there forepoint?

Hey Expunge

Personally I chose to use the PAC Wireless 28db parabolic because they provide a substantial amount of gain without too much aperture size, a 6’ dish isn’t fun to deal with high on a tower.
I have had excellent results using the PW 28db parabolic on links in excess of 18 miles LOS at -60 with throughput averages of 48Mbps faithfully.

The XR5 coupled with the PW parabolic works very well also IMHO, lots of head-room and I find that they deal well when there are other 5.8Ghz gear from other operators near your Freqs too

I won’t say anything negative regarding the HL 32db’s, however, as with anything, too much of some things are not always the best scenario or solution.
Another thing is that you will in most cases find that the higher the gain factor the narrower the beam-width, while narrow beam-width is the way to go- you must remember that you will need to have a person at each link point to fine-tune the parabolic for the best possible S/N ratios.
Your partner will need to make minute adjustments while you watch your readings and then vice-versa for yourself making adjustments while your partner watches his end.
When tuning a parabolic it is always best to start with each reflector setting at negative Elevation and work into the sweet-spot beginning with the Azimuth first.

You can find yourself sitting in the side-lobe of the antenna’s radiation pattern thinking you have the best possible signal, but don’t stop there, keep tuning your Azimuth to the max in each direction until you absolutely validate that you are in the sweet-spot of the reflector.
Only once you and your partner have validated that each antenna is near bore-sited is when you would begin the fine-tuning in Elevation.

I’m sure you have already devised something as I am going to suggest here also Expunge- get a-hold of some structural aluminum tubing which is 1-1/4" in diameter, I located some old hang-glider tubing.

Cut the tubing long enough so it can protrude the outside of the tower at least 1-1/2’, then cut a slot in one end of the tubing wide enough to receive the lattice in your tower and create a hook-point.
After your slot is completed drill a 5/16" hole clear through the tubing a couple inches from the slot you cut earlier, this is where you will use a hardened 5/16" bolt and nut assembly in conjunction with a length of heavy chain which will basically wrap around the back tower leg next to where the slot is receiving the lattice keeping it from just pulling free with weight applied.
Make sure and use a double-nut on the bolt so when you are hooking to the tower you don’t have to worry about losing the anchor chain bolt, then you just remove one nut by hand and warp the chain around the tower tube then onto the bolt then just place the nut finger-tight

On the opposite end of the tubing you will drill another 5/16" hole and another slot will enable that end to receive a pulley which is to be rated to at least 200lbs and can receive from 3/16 to 7/16" climbing rope.

At one end of the climbing rope secure a snap hook capable also of at least 200lbs, this will be used to secure the aperture for the lift process.

Once completed when you set the hoist onto the tower the slotted receiver end with the chain will secure the back side and the opposite end will be supported by the tower lattice on the pulley end.
Now your partner can fasten the aperture using the snap ring and then pull it up to you easy as can be, once lifted into place your partner can tie off the rope at the bottom so you can safely install the antenna to the tower.

It’s interesting reading through the forum how many people in the US are experiencing so many issues with other WISPs way beyond power limits or saturating the spectrum with poorly designed installations and making it a nightmare for everyone. I guess that’s the price one pays with Part 15 gear and ISM bands

Isn’t that the truth!
I have two other operators in my area… one guy has had his system going for 3 years and can’t seem to figure out why his clients can’t connect reliably, he actually doesn’t even bill his clients due to poor connection.
This poor guy is too meek in my opinion, I spoke with him regarding the issue at-hand and suggested he have a sit-down with the other guy but he doesn’t like confrontation and chooses to suffer.

Me on the other hand; I love confrontation!
I have called the offender in our area to the table on more than one occasion but my requests go unanswered.
I know this guy is doing this on purpose too!
When I first found him to be the problem the noise floor was -80, now he has upped the anti and the average 2.4Ghz noise floor stands at roughly -64 presently.

This guy has been in biz in our area for a little over 4 years now and doesn’t like competition!
When we first turned our towers up in 2.4Ghz my answer machine was barraged by the offenders calls begging me to change frequency and his final call sickened me as he sounded like he was going to cry.
He uses channel 11 in one valley and totally walks on the other 10 channels due to over-amplification coupled with omni antennae.
This guy I guess believes in brute force and kill thy competitor with unusable spectrum.

The offender of my area purposely avoids all forms of contact with me.

He’s not only ruining the business aspect for others but also making problems for himself as he has created so much noise in the valley that it is now working against him and his clients also.

Considering the fact that I continually take calls from his clients complaining of no connectivity when they need it and his failure to return their calls for support work into my favor as I convert a large number of his clients over to my system regularly.

Good luck on your venture at either rate Expunge, I hope to hear your success story regardless of hardware options you choose to go with.

Regards



Site 1                                                                           Site 2

            RB333-5Ghz H-Pol-AP Mode    <----NStream---->   RB333-5Ghz H-Pol-Station Mode
           /   wlan1 OSPF cost 100                                            wlan1 OSPF cost 10   
Router<                                                                  
           \                                                                   
            RB333-5Ghz V-Pol-Station Mode    <----NStream---->   RB333-5Ghz V-Pol-AP Mode
               wlan1 OSPF cost 10                                             wlan1 OSPF cost 100

Assuming both links are operational:
Traffic from Site 1 to Site 2 takes the lowest cost path, which is across the Vertical link
Traffic from Site 2 to Site 1 takes the lowest cost path, which is across the Horizontal link
Traffic flows in a full-duplex path across the 2 half-duplex link, significantly improving overall link performance and throughput

If one link drops:
It immeaditly fails over to the remaining link in standard half-duplex mode, while overall performance may suffer significantly, there will be no outtage or downtime.


Note, this is only works if you are routing, not bridgeing.

I would like to see your ospf configuration. Please contact off forum. steve at airsignal dot net

Hi,

I currently have a 73 mile link, using 10mhz channel size, running 5.8ghz that will do 12Mbps of actual tcp throughput.

(1) Change to outdoor mounting using PoE
(2) Change to 5.8ghz. The Fresnel zone clearance on 5.8ghz at that distance is 90ft, compared with 140ft at 2.4ghz. More signal will get “over the hill” that you are now fighting with.
(3) Use either SR5 or XR5 cards and 3ft solid dishes. This should give you a -58 signal (or better).
(4) Turn off connection tracking

Using RB532’s or better, you should be able to get 25Mbps or better across this link.

email sent.

note, I realized in copying pasting I made a typo on the diagram on my previous post indicating that both sides of the Vertical link were running in station mode, I edited to correct this…

Hi ForePoint, i would like to see your configuration too. I need to decide between ospf and nstreme2 and i want to test in lab to verify if ospf is better in my scenery. Please sent to npbrasil at gmail dot com. Best regards

same here forepoint. I would love to see this config. Thanks for your contributions to the community! I wish more were as generous. gcs85 at hotmail dot com