Has anyone had success with these kinds of antennas? Tell your success and fail stories here please.
Yes! Very interesting theme. We want to do some tests soon using Mars Dualpol R5H.
Esp. what do you do to get maximum separation between the 2 radios. I’ve heard
that some guys put 2 complete boards into the compartment of a dualpol antenna
just to separate the Wireless cards/pigtails.
Does it improve speed to use one radio per direction?
What channel separation is neccessary?
Stefan
I did not tried cross-pol antennas (yet) but here is some numbers:
-
Isolation between ports aprox 25 … 30dB (if a good feeder is used): this means that a +20dBm signal from one port will show up at the other port as -10dBm better case. Is there any Mini-PCI card that can handle that, even operating at the band edges? Probabily the receiver will desense due to front end overload.
-
Single braid cables must be avoided between cards and antenna connectors: choose dual shield or semi-rigid ones.
-
Better isolation with two completely sealed (RF point of view) boards + mini-pci cards. RF is like water
. Leaks thru power supply wiring or ETH cables from one radio may show up at the other.
Now: if someone make both cards that shares the same antenna run at the same TX-RX cycle (e.g both TX and both RX at the same time slot) the answer is yes - you can share the same xpol antenna at adjacent frequencies or even at the same frequency if receivers can handle a -22 … -25dBc interfering signal (e.g ideal attenuation introduced by oposite polarizations affected by some propagation effects).
If cards can handle the -10dBm signal from a different channel at RX input and if modulation scheme is not affected by an interfering signal which is -35dBc (intermod products from on-air TX at the adjacent channels) may be system run without time-slot sinc using different channels.
I don’t know how good those mini-pci cards are under co-channel interference but I saw PTP full-duplex systems using the same channel with 32 or 64 QAM modulation with zero BER if the oposite polarization signal was at least -25dBc. I don’t know this numbers for OFDM. And there is the Sat TV business, sharing analog or digital video at the same channel but oposite polarizations (linear or circular). Feeder isolation is the same. The path propagation phenomena are not.
Software sometimes (firmware) can handle that, a routine called XPIC (cross-pol interference cancelling).
Regards;
No problem at all.
With bonding we have a link with 24km running with the mars antennas. The only thing you have to regard is avoiding overlap of the scan list. Stations searching for AP will kill the already established link.
I was not able to setup a perfomant NS-Dual-System (RB433). The bandwidth was very low (4 Mbit/s). The duplex distance was maximum possible (5180 vs 5850).
Now I am waiting for 802.11n to build a system with it.
What is your running configuration? Mars with compartment or external antenna? One Board with 2 cards or 2 Boards
with one card each? What card? 20/40 Mhz channels? What is the real bandwidth?
Stefan
We have a 93km link with Dual pole Andrew Dish’s (900mm)
Both pol’s have a Nstreme link with AP bridge and station_wds , the bridge at each end has rapid spanning tree to resolve packet storm issues.
At certain times of the year it gets bad signal fade so hence the failover between the polaritys, space diversity would fix this but we cant be bothered at the moment.
Signals are -59 and -69 - Difference being mainly one is running 5300Mhz and the other more in peak of dish at 5800mhz
Xr5 cards too.
Been like for for well over 1 year.
So to answer the initial question, it works well and have had no problems.
we have many other shorter links configure the same working fine, the x-pol seperation is good on the Andrew dish so you can run the frequencies quite close without any forseeable negative effect.
Im about to try some dual nstream soon and looking forward to see what that performs like.
Cheers
Justin.
Single RB433AH, Bandwidth is with turbo about 60 MBit/s HDX (regarding eirp). But this is only an emergency solution (while I wait for 802.11N) cause we are loosing a lot of spectrum.
ähh, I forgot:
Antenna is mars with compartment.
Hi Justin,
OT but where do you buy your Andrew’s dishes from in NZ ?
Regards,
Andrew
Hi All
Try this antennas! they have 53 Db isolation betveen H & V pol!!http://en.jirous.com/antenna-5ghz/jrc-29-duplex
We will be starting to sell this unit soon.
You cant get the px#f series of 5ghz dish’s any more sadly as Andrew stopped making them, they have been the best units we have ever had, quite pricy but best quality all round by a long shot.
Otherwise for other andrews stuff Ideal electrical can get things for you (if you know what your after)
Cheers
Hi,
Have you had any experience using two radios (full duplex) over a single antenna with narrow band filtering like the Tranzeo TR-CS1?
Regards,
Daniel G.
Poynting’s dual-pol 5Ghz enclosure/antenna has worked great for us. An RB600 fits inside.
Hi Normunds.
Successful with 2 oiwtech 32dbi 5GHz dual polarization antenna, using RB600, Mikrotik R52H and nstreme dual.
1 frequency on 5.8
other on 5.3.
72km spot and 20megabits one way.
10megabits ± both ways.
if you need more description and screenshots I can make.
best regards from brazil
vince de luca
what about no dual-polarization but those antennas which have two on one panel? I think there are such also available, with two connectors.
mramos, has exactly the right concept it is all about the isolation of one signal from the other, it doesn’t really matter the configuration of the actual antennas only the raw numbers.
I used to use signal-pol 29 dBi dishes and would put 2 of them up wired to 2 seperate radio enclosures. Antenna seperation was about 10ft, boxes were aluminium but close together (1.5 ft apart). Using this method and any form of fdx (routing, bonding, etc.) was a mess. I typically got better performance when runnning one antenna at a time.
However after the initial mess, we decided to wire the antennas to one enclosure and run dual-ns for comparison purposes. Now remember Nstreme dual is a RouterOS only feature and the beauty of it is one radio is locked doing only Rx, and the other only Tx. This was a huge advantage as the 802.11 ack commands were always sent on the same radio (this is contrary to the seperate enclosure where even though each radio must be doing both Rx and Tx). Performance jumped and we got a much more reliable fdx link.
After that test we evaluated 29 dBi dual-pol dishes, repeated both scenarios, and came out with similar results. Why? Well if you do a signal scan in both setups you would find there is not alot of additional isolation with the antennas seperated the vertical signal is still being picked up by the horizontal antenna and vice versa.
Rule of thumb… max isolation between signals = better performance, it doesn’t matter how its achieved.
To answer normis, the panel style dual-pol antennas we use, are built with two antennas on one PCB and have seperate connectors for each, the performance is the same as the dual-pol dishes.
The only approach I have wanted to try but haven’t due to the loss in flexibility was to use a signal filter (bandpass filter). Most products that are built for FDX use this to achieve signal isolations >45 dB on top of what the antenna is already doing, and as mentioned earlier some custom FDX solutions would use software to futhor isolate the signals before the actually radio would ‘hear’ it.
Cheers
MyThougts et al.
I had severe problems at a X band instalation (8GHz), where the customer did not want to spent on bandpass filters at the TX side because “TX frequency was 650MHz away”. Make him understand that the transceiver generates background noise, the SSPA generates out of band bg noise was a hard task. At the end, the receiver was completely blocked by wideband bg noise from PA + xceiver. A 135dB isolation was necessary. So … we added the TX filters and solved the problem.
On those 802.11A/B/G radios I think the safest way to put them at the same antenna - and - keep all the avialable frequencies or reach predictable and repeateable results is: both V&H on one side TX and RX at the same time slot. May be using a single clock source for both, I don’t know. Have those cards a “clock pin” or an external TX/RX pin for TDM sinc?
Microwave filters are basically mechanical stuff and to be good need to be machined at a CNC. Tuned ones have losses between 1.5 and 3dB if you use the adjacent channels as stopband transition. Waveguides ones when silver plated internally can exibit loss as low as 0,3dB and can be built for a single channel with this loss and good group delay figures (modem taps can handle that). But we always needs two filters, one for TX and another for RX (4 total at one side). The RX filter kills the TX channel. The TX filter kills the out of band noise at the RX frequencies.
And the leaks … on bandscan I can detect full scale signals from a Ubiquiti SR9 (its 2.4G stages) using an SR2 3 meter away. Note that the SR9 is not connected to any antenna but to a dummy load good up to 12GHz. And the SR2 uses a bi-quad antenna with a 0.45m² reflector. My idea was a 900MHz NLOS link and a 2.4AP at the same X86 card. But with those leaks …
All PTP microwave radios have sealed compartments for TX & RX. Diplexer isolations ranging from 135 to 160dB depending on TX power.
Regards
Currently we have 5 dual polarity antennas available:
- Jirous - parabolic dish - 29dBi/29dBi gain @ 5400-5900
- Jirous - parabolic dish - 23dBi/23dBi gain @ 5400-5900
- Mars MA-WA56-DP25N - flat panel antenna - 25/24dBi gain @ 4950-5850
- Mars MA-WA56-DP25 - same as above but frequency range is only 5400-5850
- MTI 485025/NVH - flat panel antenna - 23dBi gain @ 5000-5950
My personal favorite is MTI, because it has about 50dB of separation depending on frequency, which allows you to make nstreme dual link with high power cards (maximum power without rx errors is 24dBm) and has very wide bandwidth from 5000 to 5950 - you can find many clean chanells
We have about 100 links with these antennas, without any problems, from 30m to 14km, performance is about 80Mbits full duplex with strong&clean signal @ 5GHz turbo
Note on Mars antennas - they have two models, unfortunately, new mode (with N) has more bandwitdh, but speparation is not enough (only 25dB) witch makes this antenna completely unusable with atheros cards.
The “old” MARS without “N” has about 38-42dB of spearation, depending onf frequency, this is enough in most cases but tx power should be as low as possible 14-17dBm works good.
About Jirous, personaly i don’t like dishes, but 29dBi model is very good antenna for nstreme dual, especialy over long links (over 15km). We have case with 18km link, @ 82/78Mbits signals -57dBm
About frequencies, with spearation over 40dB (especialy with MTI & Jirous) , you can easily use chanells with only 40-50MHz separation, for example for turbo 40MHz chanells: 5500MHz (5480-5520) — 50MHz space — 5590MHz (5570-5610)
Are you running RB433’s with N-Streme dual in these MTi’s?
Are you running RB433’s with N-Streme dual
theoretically it’s possible to make one, but i’m using boards with faster cpu 433AH/600A or ALIX