Hello everyone,

we bought this Antennas: http://www.wlan-shop24.de/5-GHz-WLAN-Ri ... MIMO-17dBi

and this one: https://www.interprojekt.com.pl/netmeta ... -1641.html


We thought it would be cool to have 3 chains in use for max throughput.

The Link Distance was 2 km clear LOS and free Spectrum aviability

The Test Link was very bad. we did only get max -90 Db

we use these config:

name="wlan1" mtu=1500 mac-address=... arp=enabled
interface-type=Atheros AR9888 mode=bridge ssid="notourrealssid"
frequency=5860 band=5ghz-a/n/ac channel-width=80mhz-eeeC
scan-list=default wireless-protocol=nv2 vlan-mode=no-tag vlan-id=1
wds-mode=disabled wds-default-bridge=none wds-ignore-ssid=no
bridge-mode=enabled default-authentication=no default-forwarding=no
default-ap-tx-limit=0 default-client-tx-limit=0 hide-ssid=no
security-profile=profile1 compression=no


So we think, that the Antennas are not the best chosie.

We have a second Link as with other Antennas :

https://www.interprojekt.com.pl/gold-wi ... p-719.html

there is no problem with the Link same distance in a noisy situation.

Does anyone have any Ideas what could be the Problem?

Bad antenna, very bad antenna for sure. Anyway I tried to use one dual pol antenna + one 23dbi panel and was worse than only 2 chains but it works very good with the pannel in another direction and link up with another station.

Bad antenna, very bad antenna for sure. Anyway I tried to use one dual pol antenna + one 23dbi panel and was worse than only 2 chains but it works very good with the pannel in another direction and link up with another station.

This is what we we thought already. Do u know good stable triple chain antennas?

We dont have the place to use 2 or 3 antennas.

See this antenna ...
http://www.wireless-instruments.com/en/ ... 20hvx.html

We offer this 3x3 antenna, available w/o enclosure, than we use our holder to fix the NM directly at the antenna, and also with enclosure, so you can integrate a RouterBOARD directly.

Antenna only:
http://shop.meconet.de/Antennas-Accesso ... 14004.html

Antenna with Enclosure for RB/411 and similar:
http://shop.meconet.de/Antennas-Accesso ... anguage=en

I never hear about so poor results from customer, but be aware, if you use more than dualpol, you don't have the ~25dB physical attenuation between your polarization level.


Regards
Lutz

Hallo Lutz,

danke für das Angebot, Ich rede mal mit meinem Chef und geb bescheid wenn wir welche bestellen wollen.

There are two importand things in 3x3 antenna system:
- antennas should be mirrored, It means that if one ontenna is HV and +45deg, another side must have HV and -45deg. I hope it;s clear, but if someone is not understandig, I can try to make a picture that explains that. It means also that the antennas can not be the same, but should be paired.
- maybe it's stupid at first look, but it works: chains have to be connecet in wright order. So H-H (ch0) V-V (ch1) +45deg and -45deg (ch2)

It;s also very important that she signal should be strong enought on every chain, Good idea for testing a link is to connect it in 802.11 (without nv2 or nestreme) and tune it for best parameters (analyzing CCQ and packets lost from registration tabel)

I hope my advices will be halpful,
A lot of bandwith and many free channels my friends

There are two important things in 3x3 antenna system:
- antennas should be mirrored, It means that if one ontenna is HV and +45deg, another side must have HV and -45deg. I hope it;s clear, but if someone is not understanding, I can try to make a picture that explains that. It means also that the antennas can not be the same, but should be paired.
- maybe it's stupid at first look, but it works: chains have to be connected in right order. So H-H (ch0) V-V (ch1) +45deg and -45deg (ch2)

It;s also very important that she signal should be strong enough on every chain, Good idea for testing a link is to connect it in 802.11 (without nv2 or nstreme) and tune it for best parameters (analyzing CCQ and packets lost from registration table)

I hope my advices will be helpful,
A lot of bandwidth and many free channels my friends

Can you explain what you mean with; "the antennas can not be the same, but should be paired"?.
I do agree on the mirroring, that what we have done. But I used exactly the same antenna, but only the fixing is mirrored so they both 'see' each other at the same angle.

I used first 2 Jirious 30db dome antenna's and aligned them as good as possible. Only after that we enabled chain 2 that on both ends was connected to a 31db Gold Mesh antenna and we aligned these by looking at the signals on both ends of chain2 only. (Obviously)

These meshes are just not or just touching with one corner the domes and fit to the same pole. But it that means now that the radio waves are hitting at the same distance (in mm or ms measured) as the dome I am not sure. Obviously these antennas don't work exactly the same.

The meshes, although they are 1dB better than the domes, still show almost consistent 3-5db less signal than the domes. (I look at the chain levels, not the total rx/tx, that just the sum)

The ROS only very rarely shows it works with all 3 chains (3S), mostly only with two. And the CCQ is below 50%.
Signals are around -40/-41 but sometimes the tx lowers to -48;

As can be seen in the two screen shots, although the 3rd chain is enabled and therefore the overall signal increased a bit (theoretically 1,5db) the CCQ actually dropped on the tx a lot (from around 95's to 60's) and the actually connection rate on the Tx also dropped.

During all these tests I actually didn't see anytime 3S popping up so obviously only 2 chains are used all the time.


What I was thinking of doing was to put 2 similar domes to replace the mesh antenna but sadly enough I can't fit these diagonal...... so I have to create a second horizontal or vertical polarized chain.... not ideal.

Another problem we have now is that with the new ac radio chipset we cannot see the actual power per chain, nor can I see the actual MCS rate strengths. So I can not start playing a bit with the rate setting and select some lower rates to see it that would help the link.

What I have done now is on the uplink disable the Tx of the chain2.
Immediately the tx rate jumped back up to 300Mbps and the CCQ went up as well.

Yet again something strange; on the signal I still got a tx reading for this chain3? How can, its disabled....




Traffic you see in the screenshots is live traffic though, not a test.

I will try to explain it with picture:
The same antennas, so identical polarisations: It's wrong configuration, because when you face these antennas one polarisation will not match.

This is how it should be: Now it's OK - when you face those two antennas all polarisation will have it's partner.
The biggest problem is with betwen slash polarisation and "H" or "V". Because it's very close in angle, those two streams should be maximaly separated, so that the difference betwenn them will be enought to get third stream working.

@czolo;

I'd fully agree, you give a nice example on how polarization should be done. Off course that is the case in my setup.

After reading hours online (not only this forum) I distilled that indeed the lack of cross pol. separation special on longer links reduces or eliminates the advantage of triple chain setup.

Also, I'd realise now that by having a 3rd chain not working, while the 3rd chain radio still emits energy through a (in this case ) mesh antenna we create self interference. So instead of creating a better link we make it worse.

On our 12km link we found that 3S is almost never to be achieved, even with -40dBm signals.
At the same time we saw that the CCQ stayed around 50% were disabling the 3rd chain immediately sees the CCQ go up to the 90's...

Now I've got one more thing to try. Use a better 3rd antenna that has bigger front to back ratio and almost 'nul' side lobs. Only Jirious meet that standard so far I could find. And such antenna also needs a 45º slant polarization. So far I did not find any antenna meeting these requirements...

Next plan is to use a 3 chain link on a 500meter link. And separate the antennas a bit.
I will do this with a 25dBi Jirious antenna pair in combination with a 25dBi Gold dish that can be fit in 45ºslant setup. I'll let you know how we go on that one.

Same time we are making a 350 meter similar link so we have two to play with.

The present 12km triple chain link will be replaced by a 12km duo chain 45ºslant fit (Ubiquity with 45ºslant adapters!) antenna setup. This will give me a bit better interference rejection and together with 'ac' and possible 80Mhz will give me the throughput I am looking for (150-200Mbtcp aggregated).

The 'short' 350 meter link is now duo chain 80Mhz 'ac' with almost 250Mbps throughput. In achieving a proper triple chain link here I hope to be able to fall back to a 40Mhz 'ac' link that still will transport around 200Mbps (Which is enough for now and near future...)

I'll keep you guys updated. But it takes time....

I'm thinking on this new setups:

1. A triple chain link over 680 meters. (B to C)
2. A triple chain link over 354 meters. (B to A)

Both links spring from one point "B" that also has one 'ac' link to yet another location km's away, a 4rd 200meter 'ac' link, a 5th 'n' link some km's away and 4 ap's.

Hence if we'd use 20Mhz for each, we already use 9 x 20Mhz = 180Mhz of spectrum.
Since 4 of the backhauls are 40Mhz we in fact use 260Mhz of spectrum were we officially can only use 5500-5700 (-5925 for 'ac' wide bands)

The no. 1. link goes to a site "C" were yet again is another AP + a new backhaul to remote location.
The no. 2. link goes to a site "A" were yet again are 2 AP's and one more backhaul to a remote location.

As you can calculate, we use the whole available 5Ghz band, with overlaps and beyond.... its a big puzzle to get it all done.
Tools we have; High gain antennas, even for the short distances. Proper shielding were ever possible. Careful setting of working frequencies in each radio in respect of what its neighbour is doing.


Site "A" has our central with a 300/300Mb backbone connection. So here it all has to go/comes from....
The link from "A" to "B" need to carry almost the same as the 300Mb (mainly download though) so here I am looking for connection rates of at least 450Mb.
"B" to "C" can be slightly less, so 300Mb conn. rate would be minimum.

For future sakes both could be better off with one step more.

Hence my plan is to make the "A" to "B" link a triple chain 'ac' link working in 40Mhz wide band.
In using high gain antennas the signals can be set around -35dBm and I hope to get VHT MCS 9 (or 8 ) stable running.
This would give me on a triple chain link 600Mb conn. rate (or 540Mb). That would keep we from the need of using a 80Mhz channel. (I do now and I have vht mcs 6/7; 585/650Mb conn. rate.)

The "B" to "C" link will be similar, 40Mhz 'ac' but due the slightly bigger distance (and interferences) I think VHT MCS 6/7 is what I can expect. On a triple chain this should still give me 405/450Mb conn. rate)


Both links are made of 1x 25dBi Jirious Dome antenna and one 25dBi 45º slant fit Golddish antena.
Off course the 45º slant's are mirrored into the link. And both slant's on "B" will have the opposite 45º slant angle.

I sue the Gold dish antennas because so far these are the only dished I could find (and I'd happen to have these surplus anyway) that can be fit with 45º angle. Most other cannot.


The questions are now;
Should I fit the extre (3rd) slant antenna fit as close as possible to the Jirious dome on the same radio?
Or should I place them lets say a meter apart?
And if I would put them at different colocation compared to its partner on the other end of the link; on "B" the slant goes above the Jirious where on "C" the slant goes below, would that make any difference?

Length of cables: I would say within a radio setup all chains should have same length antenna cable. But is that really true? Due the fase shifting created by un-equal length cables we might even improve the link?
Any input on this?

To fit the 3rd antenna at some distance (maybe even 2 meters?) from the Jirious I need longer cable anyway. It might even help the time delay needed for face shift. Or am I now completely missing the concept?

Any input in this before I start playing?



P.S.
Yes, cable/fibre could be a solution, but not economical at this point and permission needed from national operator, which will be denied....
Yes, GPS sync will help, but since MT doesn't have it we need to swap most hardware if we'd want. At this moment not feasible. (And if they don't come with it in the near future, we are going to say goodbye to Mt any time...
Yes, some licensed radio's (or 'light' beam units) might do, but not economical at this point in time.

Oh, forgot to mention that I do realize that a conn. rate only give me 50% in real throughput.
And off course I need relative high CCQ's. But a 450Mb conn. rate with 90% CCQ should give me some 200Mb capacity. For now I would settle for that.

I will try to explain it with picture:
The same antennas, so identical polarisations:

wrong.png

It's wrong configuration, because when you face these antennas one polarisation will not match.

This is how it should be:

Ok.png

Now it's OK - when you face those two antennas all polarisation will have it's partner.
The biggest problem is with betwen slash polarisation and "H" or "V". Because it's very close in angle, those two streams should be maximaly separated, so that the difference betwenn them will be enought to get third stream working.

So in effect 3x3 Mimo antenna's should be purchased in configuration matched pairs and not individual unless the antenna polarity can be changed within the enclosure.

In my opinion using a slant polarization for a 3x3 system is a mistake.

Slant polarization is not recommended for use in combination with polarization H and V .Insulation due to very small differences from the other two. Only -3 dB against each of them.
Usually,singular slant polarization is used in cases where a desired mitigate interference from the environment with a minimum of -3 dB.
In MIMO rule of thumb is that, the signal on each connection of the antenna unit must to be out of phase by 90 degrees from each other.
For a 2x2 MIMO system it is relatively easy to obtain that. Polarization V to H is out of phase with 90 degrees. But slant it is 45 degrees from the H and V and give only -3dB attenuation against each of them. Result a very bad CCQ for 3x3 MIMO system , polarized with slant in combination with H and V.
Even when using a second antenna for the third channel solution offered by some manufacturers, not solve it.
A relatively easy solution would be to use of quadrature directional couplers in various combinations.
It will be mounted between the antenna and equipment.
In this way it is possible to ensure a phase shift of 90 degrees for each channel.

Sorry for my English.

In my opinion using a slant polarization for a 3x3 system is a mistake.

Slant polarization is not recommended for use in combination with polarization H and V .Insulation due to very small differences from the other two. Only -3 dB against each of them.
Usually,singular slant polarization is used in cases where a desired mitigate interference from the environment with a minimum of -3 dB.
In MIMO rule of thumb is that, the signal on each connection of the antenna unit must to be out of phase by 90 degrees from each other.
For a 2x2 MIMO system it is relatively easy to obtain that. Polarization V to H is out of phase with 90 degrees. But slant it is 45 degrees from the H and V and give only -3dB attenuation against each of them. Result a very bad CCQ for 3x3 MIMO system , polarized with slant in combination with H and V.
Even when using a second antenna for the third channel solution offered by some manufacturers, not solve it.
A relatively easy solution would be to use of quadrature directional couplers in various combinations.
It will be mounted between the antenna and equipment.
In this way it is possible to ensure a phase shift of 90 degrees for each channel.

Sorry for my English.

"quadrature directional couplers" explain yourself. What is that, and how can you achieve 90 degrees separation in a 2-dimensional space between 3 antennas in the first place.

There are 3 antenna systems on the market, both for indoor as outdoor. How do they manage the issue?
For our outdoor environment these won't work due their low gain. Anything below 20dB is not interesting for backhaul over distance.

I don't get it, 3x3 mimo is promoted, is sold and high end manufactures say it works for them... how come we can't get it done?

I have even seen 2x HP or 2xVP antennas. From professional antenna builders. How come they can fit 2x VP or 2xHP in one unit and not have an issue where 45º slant would give one in combination with 1xHP + 1xVP?

And by the way, I actually use 2 setups with duo chain 'n' radio and both chains connected to a single chain VP antenna. One antenna is 90º sector with only 17dBi gain were the other is 22º directional with 24dBi to reach some distant clients.
It works very good (in the single chain 'world').

So in effect 3x3 Mimo antenna's should be purchased in configuration matched pairs and not individual unless the antenna polarity can be changed within the enclosure.

That's a very good remark! Something you need to think about when ordering these.

Quadrature directional coupler or hybride directional coupler , same thing , are passive devices .
In the link below you can see an example in Figure 21 how can also be used a combination of such devices. With a little imagination you can get an idea of what I explained above .
- https://en.wikipedia.org/wiki/Power_div ... l_couplers

I don't think chain separation in respect to the phase is so important. There are manufacturers that successfully use multi antenna system, up to 27 antennas/chains. Most of them share the same polarization (Ruckus) and polarity separation doesn't seem to bother them...

In small soho 3x3 routers high speed can be obtained by using all three chains between (3x3) devices. Coming and going to similar polarized antennas. How come polarization is not an issue here?

I think the bigger issue in creating long range backhauls with 3x3 systems is the fact that a bigger part of the data stream increase has to come from multipath receipt.
With an indoor 3x3 system their is plenty multipath signal so plenty of phase shift between streams that the mimo technology can use to build reliable high date rate.

On long haul links that need to be obtained by high gain antennas the multipath receipt is supressed a lot by the use of these antennas. Any beam that is not directly coming from the main antenna beam (and that therefore is a multipath signal) hits the receiver antenna under a angle out of its main lobe. The higher gain the antenna (to reach bigger distance/higher signal level) the worse the multipath reception..

Hence a multi antenna/multi stream (4x4, 5x5 up ??x??) on short range that use antenna with sensitivity in wide azimuth range do fine. The chipset is presented with many different streams all in relative same signal level and can build a good data rate on it.

A high directional antenna system will on the other end probably only receive the main radio beam (because any multipath beam just won't reach the other end any more, or to be weak to be usable) and when 3 chains are used the only separation is the polarization. Since each stream also hits the other stream's receiver antenna yes, cross polarity separation is the only thing left in basic wifi systems.

A solution could be indeed to separate each stream's antenna a couple of meters and/or differently point them slightly away from the other end towards a reflective surface. Now we create more time delay and thus phase shift the receiver can work with.
One trade off though; signal strength. Since we need to obtain high s/n level, thus high signal to get to the high mcs rates, we now need even higher gain antennas on both end, that again have higher compressed sidelobs.

Imho the trade off is such that there is a arbitral distance where a duo stream system ('ac' or any other coding system) will benefit more from a high signal than a multi stream system would be from phase shifting due multipath.

So maybe my 350meter link running over many more or less reflective house roofs might benefit from triple chain even if the max pol. separation will be 45 degrees, where my 680 meter link already might proof to be on the edge of a benefit a triple chain will bring me.

In no doubt my 8 km link will benefit more from just a very high gain duo stream antenna setup with 'ac' radio compared to somewhat lower gain triple chain setup.
In using the very high gain (30 + 34dB pair antenna, both in 45º slant setup to avoid 3rd party interference as much as possible) setup the signals will be high enough to achieve vht mcs rate 8 or 9 and thus 360 to 400 Mbit/s. If I need even more I have to go to 80Mhz wide channel.

Its interesting to know that some manufacturers are using circulair polarization in a 4x4 setup and together with some other improvement claim to have much bigger date rates possible than we could think off in using basic stuff.
Probably if I can't achieve what I want with MT or Ubnt I'll probably go for such solution or go for licensed.
But now the trade off is money!

Yes, there are manufacturers who use systems with 27 antennas and more/ channel in the same polarization. Ruckus, Xirrus ... etc, but talk of a totally different technology (beamforming) and a different destination p2mp.
.... I do not want to comment remainder posting. All I wanted to say was, that slant polarization use in combination with the other two H & V,did not help.

...........
Regarding your setup, 8 km, is not very long link, I would go on 2x2 MIMO solution with multiple links , with bonding or OSPF. That way you can have back-up and is not very expensive.
I wish you good luck and tell us the results.