SR9 range issues using omni-directional antennas

[kf]

Hi,

I'm having some range troubles with the SR9 cards using omni-directional antennas. Outdoors with the antennas ~2 meters above ground and no objects blocking, they only reach about 150 meters before the connection drops (RSSI about 10-12). Tested on channel 3 (922Mhz) with channel width set to 5Mhz and 20Mhz using max TX-power (28dBm).

Setup:
* Two SR9 cards placed in two Avilas from Gateworks
* Custom Linux distribution using MadWifi (madwifi-dfs-r3319) with Ubiquiti HAL (ubnt-hal-0.6.280.tar.gz)
* Two omni-directional antennas - Sirio SCM-918-2 (click for more info) (http://www.sirioantenne.it/prodotti_wir.php?idc=1011281017&idg=1008325576&idp=1073461158)

Correct me if I'm wrong, but these cards should perform significantly better even if I'm using omni-directional antennas. Am I missing something obvious? All tips are welcome.

[UBNT-Mike.Ford]

Hello Kf,

Yes these cards should be performing far better then this. What is the nosiefloor you are reading? Do you have the correct antenna set for SR9? Can you set the Tx Power to 24dB and see if th range is increased?

Thanks,

Mike

[kf]

Thanks for the input. I will perform some tests with another pair of 900Mhz antennas to see if the problems are related to my current ones. I'll also try a lower TX power to see if this affects the range. I'll report back with the result.

[kf]

The other pair of antenna performed terrible. So I abandoned MadWifi-dfs with the ubnt-HAL and switched back to standard MadWifi and HAL (MadWifi trunk 3314 + OpenWRT patches). The range increased and ping was stable up to 500 meters (free line of sight). I haven't had the opportunity to try a longer distance yet. But the throughput wasn't impressive at this range only about 50-70 kBytes/s. I have no reading of RSSI and such because MadWifi reports strange numbers with these cards.

The throughput seems a little low. I'm testing in an urban setting so It might be due to interference. I guess I have to do some tests in a better environment.

[malk315]

I thought I would post my experience with XR-9 radios (not SR-9, but still 900mhz and atheros based etc) to help give everyone a baseline.

I'm using 2 x Alix 3c2 with ECN #027 applied to help w/ high powered radios in mini-box outdoor WRAP/Alix enclosures.
Each has XR-9 with 12" MMCX to N-female pigtail and the following antennas:

http://www.hyperlinktech.com/item.aspx?id=1933

Software is Linux Kernel 2.6.26.3 (vanilla from kernel.org) built with Madwifi.org 0.9.4.1 from the 0.9.4 branch:

http://madwifi.org/browser/madwifi/branches/madwifi-0.9.4

I think I tried channels 3 and 4 (922 and 917 mhz) and I didn't tell it a specific channel width (5 or 10 mhz) to use as I don't know how/if that can be done w/ stock madwifi driver and NOT the ubnt radio driver. See the madwifi branch above to determine the HAL in use.

The link was about the same elevation w/ antennas roughly pointed 1000 feet apart where 300 feet of it was dense (80 foot + tall pine and oak trees) plus two single family homes of about 2300 sq ft. -- I couldn't "see" each end of the link -- so I set one up in my yard and the other at the cul-de-sac of my neighborhood but used google maps to determine the "bearing" to aim the yagis roughly.

I was getting about -61 to -65 receiving link (can't remember RSSI - sorry) and it was using about 18 megabit speed. I was able to do something like 10 megabits of throughput or maybe even a bit better at times. I was very impressed at how well these did with the major foliage here in somewhat rural Massachusetts. 2.4 or 5 ghz had no chance with this setup.

I took the same gear into Cambridge, Mass right on the charles river and tried to do a clear line of sight link about 800 feet apart over the water from shore to shore on a diagonal (same elevation) and it was marginal at -75 to -78 link. Tried all channels - ping would work, but I was amazed how poor the link was. I don't know if it was the water from the river -- but it was clear line of sight! I attribute it to a lot of 900 mhz noise in the city -- I don't have a spectrum analyzer to check anything out.

The same link in cambridge with 2 x XR-5's, same software, alix board etc. and a 12 Dbi omni on one end and backfire dish on the other end would yield a -48 (symmetrical on both ends) and strong RSSI (can't remember again, sorry!) with 48 - 54 mbps rate and throughput of 26+ megabit. 5.8 ghz LOS in Cambridge, MA by the charles river works fantastic. I'm confident I could have lengthened that link w/ LOS by a bunch of distance and perhaps tuned for long distance and done very well.

So out in the country w/ some foliage, I would consider the XR-9 where some NLOS is involved and there's not a lot of 900 mhz noise, but in the City for me anyway, the XR-9 is not a good option.

The radios I have experience with are XR-5, XR-9, SR-2, and CM-9. I like them all and pick different ones depending on what I'm doing.

My favorite antennas are Pacwireless omnis (5.8 ghz 12 dbi and 2.4 ghz 9 dbi) along w/ Backfires resold by hyperlink or the pacwireless backfire (17 dbi I think) and I've had good luck spraying 2.4 ghz with 90-120 degree sector panel Superpass antennas (out of canada) with 14 or so dbi of gain.

I typically use LMR-400 x 4 foot pre-made with N-male connectors and pigtails with the thickest braided wire through 6 ghz that I can find. I've been told MMCX is better for 5+ ghz and U.fl is OK for 2.4 and lower frequency stuff. I've been pretty happy with everything.

I also understand PC Engines has released the Alix 2dX and 3dX updated boards where I've confirmed with Pascal Dornier that one of the changes is to deal better with higher powered radios -- he said the updates for the 2dx and 2dx boards go beyond what was done for ECN #027 on a 2cX and 3cX board to make the D series boards even more rugged for high powered stuff. Next time we buy any new hardware, we're going to go w/ D series Alix boards although we've had no problems with Alix 3c2 boards equipped with XR-5 in the top slot and CM-9 in the bottom slot, and also an XR-9 in the top slot and XR-5 in the bottom slot although I don't think I turned up the XR-5 when testing the XR-9's at my house and in Cambridge.

Hope this info helps folks messing around with this gear.

--
Eric Malkowski
BVWireless, LLC
Northbridge, MA

[UBNT-Mike.Ford]

Hey Guys,

Thanks for the input. Also I would rarely recommend 900Mhz anywhere near a city as it is a crowded band and very easy to interfere with.

Thanks,

Mike

[rconaway]

We just ran some 900MHz antenna tests in Phoenix and North Las Vegas. Both tests were run with Hyperlinktech 900MHz antennas, 6dBi and 11dBi omnis to a Hyperlinktech 7dBi magnetic antenna. Around our office and in North Las Vegas, the useable distance effectively doubled with the 11dBi antenna and the S/N ratio dropped almost 6dBi in both cases. Both areas have a lot of interference and we are running 1.75MHz channels, but the difference was impressive. Our worst case test in the back of our office with the antennas 10' off the ground to the bottom of the antenna with no grounding, surrounded by semi's parked on the street, warehouses, trees, and industrial buildings lining the road, we got 1 mile with -78dBm to the car with 30dBm of power at both ends. The 6dBi antenna got about 1/2 mile in exactly the same direction.

[DieselPower]

That sounds about right for low dbi antenna city conditions. I have two towers, one 460 feet tall, and the other 120 feet tall. They are 38 miles apart, I can NOT see the flashing light on either of them from the other tower. So this is non LOS. Both are access points, but I can connect either one and get about 1.2 Mb/s across the link. That almost 40 miles. I am using 15 dbi omni antenneas and 2Ghz-10Mhz band with routerOS. TX power is normal 28DBi.

Now, just barely over a hill, 1/2 mile from the tower, there are a few houses I cannot connect. I assume that is because of the 7 degree beam width of the antenna plus the hill obstruction. They are about 700 feet below the transmitter at only 1/2 mile. Most of my customers are non LOS, but at greater distances. Most of them are 2 to 8 miles from the AP. There are a few that get down to 300 Kb/s but most run the queues to the max at 768Kb/s all the time. I'm using 15 Dbi pac wireless wire grid antennas on the CPE end.

I really have to watch myself though, because I'm softhearted and will try my best to connect anybody. That has got me in trouble several times! Usually you will connect someone, leave, knowing they are very happy because it's way better than dialup, only to find out they can only connect about 30% of the time!

Anyway, Just thought I'd throw that out there.

[rconaway]

900MHz is tough but in a mobile environment in a city, it's about our only option. The best equipment for that is a frequency hopping radio with GPS mapping to know when to hand off but that's the high end. We just tested some more 900MHz radios with a 5.5MHz channel and they had a about a 2.75Mbps throughput.

[WHT]

DieselPower...

How did that 92 KW FM station interference work out?

http://forum.ubnt.com/forum/viewtopic.php?p=9715&highlight=#9715