First, thanks for your time if you contribute to answering my questions. This is a very important topic for me because we pass a lot of voice traffic on our network. We have almost all Trango equipment and have been very successful with the handling of voice traffic so far. As our network grows I find the temptation and the opportunity occasionally to use Mikrotik for some particular backhauls increasing - for many reasons. I have not yet put into production a Mikrotik backhaul link to any primary towers because I have been scared - bottom line.

Two things that concern me are reliability and performance... I'll be more specific on both so i can get some valued feedback from you all who use this stuff out there for real.

Reliability. I have heard of the minipci boards having problems with lightning with certain antennas. The only antennas I plan to use are either a pac wireless generic (many vendors have the exact same dishes) 5.3-5.8 2-3 foot dish with radome or a deneral dynamics (gabriel) high performance or standard dual polarity dish. In every case there is a stainless equipment box at 300 up the tower with poe's fiber connections, power filters/suppression/supplies AC/DC 24 or 48v. This box is sealed extremely well against the elements, EMI/RFI, is grounded directly to the tower very well. The radios, including the Mikrotik RB532r5 I plan on using with R52 wireless card would be mounted on average 10 feet away from the equipment box also to a non-painted grounded standoff. Ethernet cable is shielded and grounded on both ends as well. My preference is to mount the RB532 in a pac wireless DCE case with metal hinged cover and Ethernet connector system they also sell. Each radio only has one ethernet connection with PoE 48v and one r52 wireless card with ufl to NFB jumper to bottom of DCE case. From there, a 3 foot LMR400 cable connects to feed on dish - I may connect two ufl jumpers to the r52 and both to the dish in the case of dual polarity options (can switch polarity by clicking alternate r52 antenna port remotely). Finally, for my question... Is this reliable in lightning frequent area (my other gear is all 100% never a problem). I hope I can expect the same from this setup.

My other primary concern is the wireless throughput. Same exact hardware and setup as mentioned above - PTP link using 2.4 5.3 or 5.8 with R52 cards. I setup the link using ap bridge on one side and station wds on the other - no routing, connection tracking off and minimal services to perform only passing of traffic though this setup as layer 2 bridge. Wireless setup as 20Mhz nstream disabled compression off short preamble. I have found i my testing the throughput to be about 20mb, give or take a few depending on what is happening. No problem - even if I only got 10MB throughput I'd be happy. I have detected small pauses or delays in the throughput though while testing (testing from pc to pc not from RB532). While in winbox the screen usually updates about once per second (interface screen, wireless throughput screen etc). When passing a 65MB test file over and over almost always the transfer stops for about 5 seconds (it does vary a bit, this is avg) - and when the throughput drops I noticed the second updates in winbox also stop - when the updates in winbox resume the transfer resumes speed as well. Winbox is unresponsive while in this paused state - to clarify, the windows app is responsive - no communication from winbox to RB532 that I can detect. This cannot happen if I deply this gear on my network. For data it would be fine, probably never notice it - but for voice, the voip data does not pass while this pause is happening - I testing it with voice calls across the test link while passing the 65MB test file. Audio resumes when the data throughput resumes as expected. I tested this with 2.9.43 and 3.0b9 Only difference I noticed is the throughput of 3.0b9 is a couple of MB higher - same pause and audio dropout though.

Can anyone explain why this is happening, if you have experienced this before and is this just something that I have to deal with suing this equipment... I hope not, but I'm not sure what to do next except replicate the entire testing scenario with 100% different gear to ensure none of the pieces of this test system have issues. Thats painful due to costs... Thanks again for reading all that and for your time in responding. I hope to be able to use this setup successfully.

Scott Carullo
Brevard Wireless
scott (at) brevardwireless.com

First, thanks for your time if you contribute to answering my questions. This is a very important topic for me because we pass a lot of voice traffic on our network. We have almost all Trango equipment and have been very successful with the handling of voice traffic so far. As our network grows I find the temptation and the opportunity occasionally to use Mikrotik for some particular backhauls increasing - for many reasons. I have not yet put into production a Mikrotik backhaul link to any primary towers because I have been scared - bottom line.

Fear not, "fear is the path to the dark side".
For 4 years i'm using MT wireless, for 3 years i'm MT distributor, so i have "little" experience regarding wireless
Everything depends of course, but the most important is _good_ quality hardware - that's something you shouldn't save on.

The only antennas I plan to use are either a pac wireless generic (many vendors have the exact same dishes) 5.3-5.8 2-3 foot dish with radome or a deneral dynamics (gabriel) high performance or standard dual polarity dish. In every case there is a stainless equipment box at 300 up the tower with poe's fiber connections, power filters/suppression/supplies AC/DC 24 or 48v

The problem was regardng to omni-directional only, since i'm producer of antennas i won't say any word about pacific ones, but how long your bridges will be, do you seriously need dishes ?
BTW. Nice setup, just like my own

Ethernet cable is shielded and grounded on both ends as well

As long as i remember you should ground it only at one point - where you have minimal ground resistance.

No problem - even if I only got 10MB throughput I'd be happy. I have detected small pauses or delays in the throughput though while testing (testing from pc to pc not from RB532).

considering that RB 532 have only 400MHz MIPS, you should calulate that maximum real throughput will be 20Mbps full duplex

LMR400 cable connects to feed on dish

LMR400 is quite popular, but i'm using Andrew (cinta series) CNT-400 with original andrew CNT-400 N connectors & waterproof kits - and for 2 years or more, i didn't ever heard about water in connector
If you want to use LMR - try to find Telegatner ones i think that they are second after the andrew ones, but they probably have series for LMR-400 as well (to be honest LMR-400 is almost the same as CNT-400...)

2.9.43 and 3.0b9

Use 2.9, even not the newest, 2.9.39 is the "one" which 80-90% of my routers have and i don't have any issues regarding wireless


If you want i can configure your MT's & help you, feel free to contact me at m.burzynski (at) cyberbajt.pl - on monday i should have some time to answer (now im preparing logistics for my next training )

is that periodic calibration setting causing it to drop periodically ?

Sam

The grounding of both end of the ethernet cable isn't really a problem since I'm only going about 10 feet through the tower horizontally. The difference in potential from a point 10ft horizontally through the tower should never be able to cause a problem (and never has)

Thanks for disolving the lightning worries with minipci cards/antennas. It should be fine the way I have everything arranged. I did notice most of the previous problems were caused by grounding problems with certain omnis - I just wanted some verification from different sources.

I do not use periodic calibration so I wouldn't think that could cause the problem. But... thats why I am asking. I'm not sure what the problem is. I see it here so it does exist in my setup any way - reproducable over and over. Actually I just checked... says periodic calibration is set to default. (Manual says: periodic-calibration (default | disabled | enabled; default: default) - to ensure performance of chipset over temperature and environmental changes, the software performs periodic calibration) Not sure what the manual means by the default setting ebing default. Default should be on of off.... I'll set it to off and run my test again.

Scott