Hi.
Can anyone tell me what would be the result of connecting an SR-71 802.11n card with two dishes (one per chain) and pointing them in different directions. My intention would be to hook up two remote 802.11n sites using one frequency only. Will the two chains work independently and give me a reasonable speed?
Has anyone tried this kind of setup before?
Malcolm
I have something like that on 2.4G using a R52n on G mode only.
Chain 0 is tied to a 90° sector antenna, 16dBi horizontal, tilted down (20º) to cover a specific area, down hill. Aprox 30 CPEs there.
Chain 1 is tied to a 25dBi grid, 25dBi horizontal, beaming to another area, creating a “spot” coverage 1.4Km away (there is some houses at a hill top).
Regards;
You didn’t say how well it worked…
How do you use two chains in G mode only? I thought you had to enable N mode when using multiple chains.
I forgot …
Well, until now … seems enough for my requirements. I have a single “plan”: 270Kbps download each client, PCQs with 1M burst ~10 seconds. Some mangle setup to split traffic for some common targets such as antivirus updates, windows updates, adobe etc etc for “free”. MS have 1M, antivirus from 256K to 768K, etc, priority=8 for them.
BTW, instead of sell 1M and guarantee 10% (like our country regulations says) I prefer to sell 256, deliver 270 and assure this speed 95% of the time.
Upload for both geo area is a common 3Mb pcq. All clients have “free” upload speed so they finish their tasks asap with 1.5 to 2M avereage.
CCQ 75 … 100, Pthroughput > 28M , HW retries = 4 => ping < 5ms full loaded. Peak online CPEs on chain 0 = 40, on chain 1 = 12.
With this configs I catch 1 ~ 3 512K ex-adsl and 1 ~ 2 ex-3G clients each 45 days.
Simple like that. One antenna on chain 0, another on chain 1, both enabled TX&RX.
As MT guys explained some months ago, their N cards on legacy modes (e.g. 802.11B, G) chooses the best chain for each client based on the strongest received signal … and use-it.
A great toy … say goodbye for 2.4/5.8GHz RF splitters (-3.5dB loss) and put both radios at N cards to work.
BTW, clicking at any client at winbox wireless registration table you can see the different levels at chain 0 & chain 1 and the level that’s in use at this time.
Of course the max speed still the standard for 802.11G (54M) or B (11M).
Regards;
Ramos, you are going to be my best friend! Such usefull info you give me!
I was just struggling to find out what was happening:
I recently bought my first SXT’s and wanted to use them in a present single chain, single radio, single antenna network V-Pol. Than, over time, the plan is to replace the AP-omni for a ´n´ omni (dual antenna HV-pol) and thus upgrade my network overtime.
First I noticed (like many already did in this forum) that to use the V-pol antenna (=ch1 in SXT) you should hang the antenna on its side. Very odd because all poles are vertical!
First intention was just to switch chain “0” off (=H-pal) to use the chain1 only but that didn’t work. Ch1 only works when CH0 is also enabled. So, if you want to use one chain only this has to be “0” so in that case you do need to hang the unit on its side… 
So I tried to run and test the unit just with both chains enabled while running either “a” or “A/N” mode. It seems to work well! I was only puzzled. Was this really a proper way or not. Since it worked I decided to go for it.
Now I made myself a dual pol directional antenna and also put a ´n´ card in with the same idea in mind. For now it has to work on an ´a´ network but later I am going to upgrade the AP to ´n´.
Off course I can now connect chain0 to the vertical antenna and chain1 to the H-pol. But later in a mixed with SXT enviroment this might lead to confusion so I decided to go for same connection as in the SXT units.
Thus CH0=Hpol / CH1=Vpol
Yet again I tested this unit. First only chain0 (=H-pol) enabled. No connection.
Than only chain1 enabled, also no connection.
Than both chains enabled, connection! And yes, now I look in the status and see the CH1 has the best signal (obvious, this V-pol gets V-pol signal).
I only never knew that MT rb and MT cards are that smart to use the best signal in this situations.
I only don’t know some of the guys on this forums that say they use one of the chains only. It just doesn’t work for me. (Or is this because they then only use CH0 for the polarization they use on the other end? Otherwise I can’t think of a reason why it should work…)
Your input again shows so much value!
Well, once I used R52n at chain 1 and it was ok. Chain 0 was not selected. What I noticed is that from time to time it drops all connected CPEs (may be it lasts 2 or more days without problems). At this time the sector antena was tied to chain 1 just because I didn’t remember which connector was chain 0 or chain 1 so I choose one, disabled the interface, reduced the power to few dBm and do a bandscan after the box was already closed.
Ok, I had picked up chain 1 and that’s it, left this way. But this was ROS V 4.XX.
When all CPEs dropped I simpy do a freq usage or bandscan and things came back to “normal” for a couple of days.
Later I left chain 0 RX enabled and the drop problem gone.
I know, wrong setup but it’s a 55m hight highrise, some problems to gain access to move the ufl plug to the right place. At the same time I put a modified 25dBi grid reflector, removing a sub reflector it has, because some spot distant area was causing me some problems. My idea was use this 25dBi grid aimed to this area and - with some luck - closer clients will keep connected.
Worked but was a temporary solution. I wanted to add a second card - a R52hn was hanging around - R52Hn at sector, existing R52n at grid. But the R52Hn came dead from reseller.
Both antennas was there, so I decided to put both at the R52n. Surprize, TX was at chain 0 but RX at chain 0 & 1. And the wireless table shows the difference in signals. Why not enable TX so? And that’s the story.
The distant spot clients CPEs has then a pair of letters added to its system name (e.g. vv-bet0330, vv-jas0089, etc). So yet on wireless tables when sorting clients using radio names I have all spot area cpes togheter. Aimed the 25dBi grid this way, trying to maximize signals, filtering “vv” CPEs reduce the visual mess (instead of use a single one I adjusted antenna position for the “best” signal combination on all CPEs).
Next step was the closest CPEs x sector azimuth & tilt.
Regards
He Ramos, I want to come back on the option to use one ´n´ radio card to feed two antenna’s.
I have this AP where loads of clients are to be found in a 180 degree sector and than only some 4 or so in opposite direction.
I was planning to setup two AP/radio’s: one on the sector handling some 30-35 clients and one on an omni for the left 4-6 units. But that would mean two cards/radio’s with two frequencies to be used with their interference issues. (I have competition using same band in region)
If I could now use a ´n´ card, but working in ´a´ (5Ghz) mode and connect both sector and omni antenna to each of the connectors, this should work?
Clients all are legacy 5Ghz (a) CPE’s coming from one present omni that becomes absolete after change over.
The advantage would be clear: both antenna’s would work in same freq. (and synchronized!) where the sector will reach further to pick up distant units while omni takes care of some clients at close range but opposite of sector.
But what happens to the output power and sensitivity for both chains in this setup?
I presume transmission power is devided over two antennas now? Thus each gets half?
And sensitivity? The aim of the usage of the sector antenna is to pick up some ´weaker´ CPE’s at distance, and lift all signal levels to below (better than…) -75 values. Do I not throw away this focussed beam antenna (sector) advantage by making it work in combination of other antenna from same card? In other words “is radio listening on both chains at the same time?” How does this work?
Any advice on this?
Hi Rudy
Nope. It’s a dual transceiver, e.g. 2 power amps, 2 preamps, etc. No divided by 2 power nor sensitivity.
Yes, it will be listening both at the same time but picking the strongest one. Even if they alternate received signal along the time between two antennas.
I aimed my antennas this way: labeling (comments) each cpe, adding some pair of letters to those one I want to listen on sector. Then filtered them at winbox reg table and sorted by tx/rx signal (or even by chain-X received signal) and aimed the sector to reach the highest overall signal, doesn’t care if some will drop a bit. Better to have 3 or 4 at -70 than 29 at -65 and one at -80.
Those 4 remaining clients … can they be reached by a focused beam (or some degree sector)? Why an omni?
Regards;
Do keep in mind the advertised power is both chains together, so the R52HN would be 20dBm per chain, 23dBm total only if both antennas are used on the same link.
I see what you mean. Basically you’re saying that in the setup as we discussed both antenna’s are 20dBm where another single chain 802.11a High power card would give full 23dBm available to the antenna.
Well, this might be a slight disadvantage, but the ´hearing´ characteristics of the sector will be still the same for the chain it is connected, and better than if it was a omni antenna. And signals at the clients ´listening´to the sector will be sort of the same as when we had one OMNI only since the sector has bigger gain. So in regard of signal levels for and from clients in the sector range there is no difference between the two options
But THE BIG advantage would be that instead of 2 AP antennas on same tower with both their freq.s and possible interferences to each other and from others I now only have to worry about chosing one free freq. for the both antennas!
Somewhere else in the forums some guys (me included!) discuss the option of having AP radio’s synchronised. I think this is what we are doing here! You could have two sectors connected, both 20dBm power available and both fully synch’d and same SSID! By beeing two sectors the ´listening´ sensitivity would be better than one omni-one 24dBm radio setup and because of the higher gain we also make up for the los of transmit power each of the sectors get!
If MT would now produce cards with 3 chains and 30dBm power you could use these for 3 sector antennas working same freq. and SSID!
Whow, if I only think of it!
Ramos, do you agree on this? If so, should we not break in that sync discussion with this option and get MT involved!
I don’t see the benefit of doing this. What you describe is basically an omni. There are very few locations left in the States that you could run an “omni” without having huge interference issues. Now if you could run each chain on a different frequency and have them synchronized you would be on to something, but the same Frequency for the whole tower I don’t see it helping. What we need is sync for adjacent channel interference on the same tower.
Chadd
This looks like PMPO for cheap audio devices .
Seriously … R52nM … this means that 23dBm on 6Mbps G mode is - in fact - 20dBm per chain?
But it’s G mode, so how can I use 2 antennas to hit the same CPE at -60 + (-60) to have -57dBm? Trimming cables to 5mils in lenght and lab-style antennas to be sure that this power will be combined in amplitude & phase over the air?
Anyway, if I enable a single chain, I’ll have 20dBm. And I’ve been using R52n for some time with a single chain enabled, so its a 20dBm device that talks to 20dBm CPEs.
Chadd, some time ago to solve a particular situation I was ready to add a second card on a 433UAH, same band. I was aware of mutual interference between this two cards (R52n + R52Hn). The R52Hn came dead so … short story … two antennas at the R52n, both chains enabled tx&rx.
Ok, it’s not perfect but at least I avoided mutual interference inside 433 box just because both chains tx & rx are sinched (or multiplexed, I don’t know). A grid and a sector, aimed more or less to the same azimuth. And the noise floor still parked at -105dBm.
Of course a ROS that can implement sort of master clock to make all existing cards, N or G, transmitt at the same time slot will make a huge difference, at least if a site have only yours rigs. But I guess this only can be done at a new generation of mini-pci cards … all local oscilators in phase, locked to the same reference (gps, master clock, etc).
More later …
Regards;
Example: I have this tower mainly with 5ghz a/n links and AP’s. To have a separate completely independent ´monitor´ network (and to use some ´old´ absolute boards I don’t want to use in the main network any longer) I am setting up a ´monitor´ network in 2.4Ghz with 10Mhz channel wide.
This tower needs to communicate with three different locations and supply also a local 2,4Ghz coverage for land owner.
I can now set up one omni to do it all but due the nature of the omni it will ´see´ any 2,4Ghz signal in its 360 degrees radius. In the 2,4Ghz band there will be no channel that is not already used so the change is big I always will have interferences.
So I decide to use strong directional antenna’s (28dB) directed towards the other 3 towers. The idea was original to use 3 different channels, all 10Mhz wide. But I also need to serve land owner with a 20Mhz channel.
So here I already have a problem. At least some ´overlap´ of channels is unavoidable. And I need at least 2 routerboards (rb333’s I still have) to serve it all.
Now with what we discussed here I am going to do the following:
One rb333 with one ´n´ card and both chains connected to two directional almost opposite in direction of eachother. 10Mhz channel.
One other ´n´ card to serve the last directional back to my central which also serves the small sector serving the land owner, running a 20Mhz bandwidth channel.
Now I can connect from this tower 3 remote locations with relative low risk of interferences (10Mhz channel combined with high gain directionals) and serve local with one 20Mhz channel that also connects to the link back to my central (and thus can do with a bit more throughput).
All this with the use of only 1x 20Mhz and 1x 10Mhz channel! By using channel 1 and 11 I also avoid the ´self interference´within the band! (Would be impossible in a normal setup with so many antennas).
One other possible use:
If you need to make a backhaul around an obstacle you normally would put the AP in the middle and either have an omni or wide sector connecting to both leg’s end-points. Or use two directional antennas with a splitter. Or setup two individual backhauls with their own radios and freq. channel.
If free channels are scarce and signals marginal you will have a problem now.
With the use of an ´n´ card you only need one free freq. channel. Just put both directionals on the both chains of the card and héhó, here we go… 2 leg backhaul with only one frequency and no extra connectors etc.
If I would have known this solution some time ago when ´n´ just came out! Would have saves me some radios and channel usage…
Hi …
Playing a bit more with R52nM + Only G 2.4GHz + both chains enabled.
I dropped from a small tower in the middle of lots of tress a pair of RB411ARs and replace them for a RB433 + R52nM(2.4 only G AP) + XR9 (900M Nstreme AP).
This tower is on a hill between two small roads with some small farms, different sectors (180º appart) each one covered by a 90º 16dBi pannel.
I use to fill 2 frequencies on 2.4 (ch1 & ch13) and in the middle the XR9 IF (wich falls on 2442 or 2437).
Now a single frequency and overall performance better than the old setup.
Regards;
Hi Ramos, you ever made this statement, do you have the original post of this?
I’ve now been given a reply fm MT support which sort of contradicts this. (send me a private mail to rudyatmarucom.es to give you more background info)
Hi …
Yes, fortunatelly I have. E-mail sent.
Regards:
Hi Ramos,
Could you post a link to that original thread for the rest of the readers?
Thanks!
-Nelson
About the chains settings. For N cards only ht-chain settings are used.
If you enable both of them and use legacy bands like A/B/G it means those antennas
will have like diversity option. As it is sending and receiving the same data
using both antennas. Note that you always need to enable chain0.Regards,
Uldis
Hello,We suggest to use 1# option (=Hor. mount, R.)as this would be pure one polarization, because if you use 2# option (=Vert. mount ch0RX+ch1RX&TX enabled) your device would receive the noise from the other polarization as well (since RX on chain0 should be enabled to make chain1 work stable). If you disable the RX on chain0 the chain1 might not work stable. This is due to the radio chipset hardware.
Regards,
Uldis
My real world usage (9 SXT now in existing legacy ´a´ network with V-pol) shows the units perform fine with only ch0RX + ch1RX&TX enabled.
I don’t have the issue of fitting horizontal poles, I have no issue with water ingress and network can slowly migrate to ´n´. I don’t notice any of the ´stability´ issues MT is referring to. In fact, SXT performs this way better than rb411 with R52 card with almost same gain antenna. Link is more stable and I have 2-4dB better signal on both ends…
Nelson, Rudy already did . Thanks.
Regards;