03/09/2015
AF5X on 21 mile link with very heavy interference still achieves 200Mb aggregate
Description

AF5X_Pic.JPG

 

We have a 21 mile link to a tower site where a competitor has 16 Cambium APs blanketing the entire UNII-2 (DFS) and UNII-3 bands. This leaves us almost nothing to use there as far as frequencies go for backhauls, except for the new UNII-1 band between 5150 and 5250. Right now our choices for using that band are either RocketM radios using the latest 5.6 firmware, or the new AirFiber 5X radios.

 

Here's what the situation looks like in the UNII-3 bands:

HudsonBrit-8-24-14.png

 

This AirView scan was done at the site on an AirFiber5, so only the nearest APs show up (due to the high F/B rejection and side lobe supression of the AF5 antennas). As you can see, the whole 5.8GHz band is compromised, especially considering that the expected signal levels at 21 miles are in the -67 dBm range – there's really no space for us to use there. Plus, the competitor there has an AP on 5840 which just barely show up here (the antenna is behind and above the AF5) which kills the use of that. We did try to fit into the spaces left using the AF5, and actually got it to work OK, but they complained that we were wiping out two of their AP sectors, so we had to stop using the 5.8 band entirely.

Here's what the DFS UNII-2 band looks like:

HudsonFireAPlink2Mid.png

 

Even if the EIRP limits didn't make a 21 mile shot pretty impossible here, the entire band is unusable – 16APs on one site will do that to you...

And here's what the site looks like from another site 15 miles away – this is from a Rocket at the other site, and it's amazingly noisy due to all the radios at the first site.

FirestoneHudCliMid.png

 

The two links point SouthWest and WestNorthWest, about 30 degrees apart. So we need to keep interference between the two shots to a minimum as well. We initially did this with a pair of AF5s, and it worked, but with those radios we could only use DFS channels, and at 21 miles it just wasn't cutting it with all the local noise.

Brit-FireMap.png

We did get backup links to work with RocketMs at UNII-1, but we couldn't get more than 60Mb reliably across the long link, even at 30MHz bandwidth which just isn't enough to serve all the clients at the far end (2 shots total of 36 miles). So enter the AF5X

 

The AF5X (as most know by now) is a half-duplex only connectorized radio just slightly larger than the RocketM, using the design and technology from the AirFiber radios. This gives it much better channel utilization than any other non-AF UBNT radio, including the AC radio line, especially at distance and in high noise environments. It is capable of 256QAM on a 50MHz channel across the UNII1,2 and 3 bands. And since it works with existing and new UBNT dish antennas, the link design can take advantage of the higher gain available from larger dishes. The EIRP restrictions still limit the Tx conducted power available, but there's no limit on Rx gain, and as far as the link is concerned, gain is gain.

 

One current issue with the AF5X is that since it uses an outboard GPS antenna (a good thing) the GPS antenna connector gets in the way of the downward-pointing connector on standard RD30 and RD34 dishes:

AF5XGPSconnInt-2.jpg

 

As you can see in the above picture, the GPS antenna connector and the dish connector are actually in the same place. It is possible to get the radio to snap in with a lot of work, but you can also just snap the lower two prongs into the mount slots – the picture below shows the back of the radio.

AF5Xrear.jpg

 

You can see the 4 rectangular prongs which snap into the dish mount. If you only use the bottom two, the GPS connector fits OK, and the rear dish cover protects the radio and keeps if from falling away from the radio.

 

But there's a better way, which can also improve the link itself – slant polarization.

The AF5's integrated antennas use orthagonal polarities for their two chains, but instead of using horizontal and vertical, the two chains are at 45 and 135 degrees. They are still orthagonal (90 degrees from each other) but by using slant instead of H and V two things happen: First, there's around 3 dB of isolation from the surrounding H and V signals (AP sectors, other MIMO links, etc.), which is helpful. Second, reflections in the environment (which tends to have more horizontal or vertical surfaces in it) are much less of an issue, so multipath is reduced.

And with the AF5X in particular, rotating the feed 45 degrees moves the GPS connector into the space between the two dish connectors:

AF5XfeedRot45-2.jpg

 

The above picture shows a standard feed on a RD30 rotated – note the level is no longer level. UBNT has built a replacement feed launch piece (the rear part of the feed with the connectors on it) which is built with the level offset 45 degrees to match the rotation. Plus, if you try just rotating the standard feed, only one hole lines up for the screws. The picture just shows what the result looks like – I don't suggest you do this yourself. Although ClaudeSS has a very good How-To here on re-drilling the holes in the feed to make this work better.

 

Our 21 mile link uses an RD30 at one end, and an RD34 at the other. We may change out the 30 for a 34 in future, but right now the link is working very well -

AF5XBritHudson.png

 

Even with the limitations in power due to the crazy Out of Band Emissions FCC regs, we're getting 120Mb one way and 80Mb the other simultaneously on a 40MHz channel. Not bad at 21 miles in really bad interference.

Before too long (as soon as the snow stops...) we plan to put up a second AF5X link to the far site, and run both of them on the same channel at this site to see just how well the GPS sync works for co-location. I'll write that up when we do.

Jim

AF5X on 21 mile link with very heavy interference still achieves 200Mb aggregate

by on ‎03-09-2015 07:33 AM

AF5X_Pic.JPG

 

We have a 21 mile link to a tower site where a competitor has 16 Cambium APs blanketing the entire UNII-2 (DFS) and UNII-3 bands. This leaves us almost nothing to use there as far as frequencies go for backhauls, except for the new UNII-1 band between 5150 and 5250. Right now our choices for using that band are either RocketM radios using the latest 5.6 firmware, or the new AirFiber 5X radios.

 

Here's what the situation looks like in the UNII-3 bands:

HudsonBrit-8-24-14.png

 

This AirView scan was done at the site on an AirFiber5, so only the nearest APs show up (due to the high F/B rejection and side lobe supression of the AF5 antennas). As you can see, the whole 5.8GHz band is compromised, especially considering that the expected signal levels at 21 miles are in the -67 dBm range – there's really no space for us to use there. Plus, the competitor there has an AP on 5840 which just barely show up here (the antenna is behind and above the AF5) which kills the use of that. We did try to fit into the spaces left using the AF5, and actually got it to work OK, but they complained that we were wiping out two of their AP sectors, so we had to stop using the 5.8 band entirely.

Here's what the DFS UNII-2 band looks like:

HudsonFireAPlink2Mid.png

 

Even if the EIRP limits didn't make a 21 mile shot pretty impossible here, the entire band is unusable – 16APs on one site will do that to you...

And here's what the site looks like from another site 15 miles away – this is from a Rocket at the other site, and it's amazingly noisy due to all the radios at the first site.

FirestoneHudCliMid.png

 

The two links point SouthWest and WestNorthWest, about 30 degrees apart. So we need to keep interference between the two shots to a minimum as well. We initially did this with a pair of AF5s, and it worked, but with those radios we could only use DFS channels, and at 21 miles it just wasn't cutting it with all the local noise.

Brit-FireMap.png

We did get backup links to work with RocketMs at UNII-1, but we couldn't get more than 60Mb reliably across the long link, even at 30MHz bandwidth which just isn't enough to serve all the clients at the far end (2 shots total of 36 miles). So enter the AF5X

 

The AF5X (as most know by now) is a half-duplex only connectorized radio just slightly larger than the RocketM, using the design and technology from the AirFiber radios. This gives it much better channel utilization than any other non-AF UBNT radio, including the AC radio line, especially at distance and in high noise environments. It is capable of 256QAM on a 50MHz channel across the UNII1,2 and 3 bands. And since it works with existing and new UBNT dish antennas, the link design can take advantage of the higher gain available from larger dishes. The EIRP restrictions still limit the Tx conducted power available, but there's no limit on Rx gain, and as far as the link is concerned, gain is gain.

 

One current issue with the AF5X is that since it uses an outboard GPS antenna (a good thing) the GPS antenna connector gets in the way of the downward-pointing connector on standard RD30 and RD34 dishes:

AF5XGPSconnInt-2.jpg

 

As you can see in the above picture, the GPS antenna connector and the dish connector are actually in the same place. It is possible to get the radio to snap in with a lot of work, but you can also just snap the lower two prongs into the mount slots – the picture below shows the back of the radio.

AF5Xrear.jpg

 

You can see the 4 rectangular prongs which snap into the dish mount. If you only use the bottom two, the GPS connector fits OK, and the rear dish cover protects the radio and keeps if from falling away from the radio.

 

But there's a better way, which can also improve the link itself – slant polarization.

The AF5's integrated antennas use orthagonal polarities for their two chains, but instead of using horizontal and vertical, the two chains are at 45 and 135 degrees. They are still orthagonal (90 degrees from each other) but by using slant instead of H and V two things happen: First, there's around 3 dB of isolation from the surrounding H and V signals (AP sectors, other MIMO links, etc.), which is helpful. Second, reflections in the environment (which tends to have more horizontal or vertical surfaces in it) are much less of an issue, so multipath is reduced.

And with the AF5X in particular, rotating the feed 45 degrees moves the GPS connector into the space between the two dish connectors:

AF5XfeedRot45-2.jpg

 

The above picture shows a standard feed on a RD30 rotated – note the level is no longer level. UBNT has built a replacement feed launch piece (the rear part of the feed with the connectors on it) which is built with the level offset 45 degrees to match the rotation. Plus, if you try just rotating the standard feed, only one hole lines up for the screws. The picture just shows what the result looks like – I don't suggest you do this yourself. Although ClaudeSS has a very good How-To here on re-drilling the holes in the feed to make this work better.

 

Our 21 mile link uses an RD30 at one end, and an RD34 at the other. We may change out the 30 for a 34 in future, but right now the link is working very well -

AF5XBritHudson.png

 

Even with the limitations in power due to the crazy Out of Band Emissions FCC regs, we're getting 120Mb one way and 80Mb the other simultaneously on a 40MHz channel. Not bad at 21 miles in really bad interference.

Before too long (as soon as the snow stops...) we plan to put up a second AF5X link to the far site, and run both of them on the same channel at this site to see just how well the GPS sync works for co-location. I'll write that up when we do.

Jim

" How can anyone trust Scientists? If new evidence comes along, they change their minds! " Politician's joke (sort of...)
"Humans are allergic to change..They love to say, ‘We’ve always done it this way.’ I try to fight that. "Admiral Grace Hopper, USN, Computer Scientist
":It's not Rocket Science! - Oh wait, Actually it is... "NASA bumper sticker
":The biggest problem in tech I see right now is that most users don't want to do things that are hard. That doesn't bode well for the industry or the society.": (me. actually ;-)
Comments
by Ubiquiti Employee
on ‎03-09-2015 07:59 AM

Nice write-up @eejimm 

by
on ‎03-09-2015 08:03 AM

What kind of latency do you see across this link? Is is the 1-2ms hopefully even under some load?

by
‎03-09-2015 08:15 AM - edited ‎03-09-2015 08:19 AM

Ping times stay at 1.5-1.7ms.  measured between the routers on either side of the radio.   So the one-way latency is half that or better.

Jim

by
on ‎03-09-2015 08:48 AM

Looks great Jim!  My 14.1 miler just keeps humming right along.  Glad I've got a half dozen more 5Xs on order!

by
on ‎03-09-2015 10:02 AM

Anyone knows why the GPS antenna uses gold plated connectors, and the RF outputs, jumpers, and the new feed cavity seem to be nickel-plated?

by
on ‎03-09-2015 10:04 AM

Thanks for the great story!

 

"the crazy Out of Band Emissions FCC regs" you mentioned, is that only for UNII-1 band? Or all the 3 bands in 5Ghz?

 

Last summer, many people were petitioning and protesting the out-of-band emission rules, I didn't seem to hear any result after that. Did FCC reconsider it? Or it became final?

by
on ‎03-09-2015 11:38 AM

....if only I could get my hands on 6-8 AF5X radios.....

by
on ‎03-09-2015 11:43 AM

Jim, just a curious question here on the idea of switching out the RD30 for the RD34.  I don't have any RD34 or AF5X so I'm not sure how this all works out.

 

Increasing the gain increases the on-channel noise pickup (already known noisy environment), so the SNR doesn't improve.  So, the 4dB of gain is only useful for the remote side because of the TX gain, but you are already at EIRP with the RD30 so the gain is lost in lowering radio output.  Is the lower output going to clean up the signal enough for the virtual wash in SNR?

by
on ‎03-09-2015 12:34 PM

@rebelwireless  potentially.

it could also reduce the noise because of the tighter beam.

by
‎03-09-2015 12:50 PM - edited ‎03-09-2015 12:55 PM

Remember the gain comes at the beamwidth of the dish, not from 360 degrees around it.   So there's no increase in the noise that's off-axis to the front of the dish.   The higher gain will also reject more signals from the sides, and gain is gain - yes you have to turn down the conducted power for EIRP, but you still get the +4 dB gain on the recieve end, so their's still a net 4dB gain increase even before lowering the noise.   And lowering the conducted power is often necessary to get the higher modulation rates to work.   So a bigger dish is always a win (unless it's too big for the structure it's on, which is another issue completely - and this tower is a monster that would probably survive a drone strike...)

Jim