Hello, I will be doing that as part of my due diligence testing before trying the link again. It will be a few weeks unfortunately.
Based on previous comments by a user here I don’t think phase is critical.
Back from other tasks.
Just hooked up the antennae for a local test. I plugged in the rooftop antenna and was in the house (almost immediately below it so at least 90 deg off the directional axis and at least 15ft below it) checking the settings on the other before taking it out for a short range test, when I noticed they synced. To be clear the rooftop antenna is pointing away from the house. I was surprised they connected since I didn’t expect the signal to propagate directly below the antenna - maybe the signal was bouncing off trees, which are about 25-30ft away?
What’s left? Oh yes, checking the noise floor in my area. I suspect there is little given the lack of residents between the antennae. Tomorrow if the weather permits as I need to run the ethernet cable from the antenna into the house.
Ran a spectrum analysis for 40sec from one end; no link yet so I can’t run a remote analysis. When I set up the other antenna if I have any issue I will run an analysis from there too.
No link (sync) yet.
I confirmed azimuths with an online calculator (that uses lat/long) and the link path with a satellite photo on the ground since compass headings are poor near metals (like my metal roof or the metal building at the other end). They should be close at this point, but still wondering how many degrees off-azimuth can they be and still link up?
I calculated the tilt angles with another online calculator that uses elevation differences and distance to triangulate them. Then use a phone app to measure tilt, which is a potential source of error of course. But they should be close. Again, I am unsure of just how sensitive the antennae are to misalignment.
I did some checking with Link Planner and modified the “runway” at both ends by moving the forest closer to the antennae and noticed their algorithm is quite sensitive to this variable. Online sources said 900MHz frequencies need 20-25ft to develop their waveform before obstructions can interfere (aka runway), but Link Planner seems to want much, much more space. Feedback appreciated.
I also finally noticed that there are powerlines near the path at one end too. Kicked myself for not noticing earlier; probably visualizing up the hill/trying to get the azimuth correct and not focusing on what’s right in front of my nose. I’ll check the noise floor at this end next time I’m there, but I’m wondering just how far out of the path do they need to be for a decent link?
1 Like
-
On the master and slave sides, make sure TX power is cranked up all the way and take note of the center frequency and channel width you’re using on the Master.
-
On the slave side, go to Tools -> Aiming, and set to ‘Single frequency only’ and then set it to the center that you have the master set to and Enable aiming mode.
-
Adjust antennas until you are able to see the maximum amount of energy. Unfortunately this aiming tool is only available on the Slave side.
-
Once that’s dialed in, turn off aiming mode, and on the Slave, uncheck all your channels and channel widths EXCEPT for the ones you’re using on the master… e.g. Master is using 10MHz channel width @ 915 center, then make sure those are the only things checked on the Slave. This will help reduce registration time and make it easier for your to optimize and complete your alignment.
-
On the Master side, watch the Logs -> BHS Registration Failures for indications that your Slave is trying and/or failing to connect. If you see frequent BH ReRange errors, or other issues, try using a smaller channel width and/or different channel until you can get it to successfully register.
-
Once the radios are registered, and your satisfied with the RF optimization, then on the Slave enabled all the channels and channel widths you might want to play with.
-
At this point you can carefully play with different center channels and channel widths to see if you can squeeze more throughput out of the link. Each time you try a new combination, run a few link tests and note the results.
-
Set your link to the combination that you’ve found gives you the best and most reliable throughput.
3 Likes
Is difficult to tell from you photos perspective but the antenna looks very low to he ground.
Consider gaining some height on that end. The ground is not your friend. When it comes to your RF pattern think Blimp not pencil.
It’s amazing what can be done with chain link fence top rail add a few turn-buckles and a few other odd and ends.
If I’ve got this straight, the beam width of the900mhz PTP radios is a radius of around 20ft. If I’m mistaken please correct me. Does that mean ideally the antenna should be at least 20ft off the ground? If so then I think I’m close at one end and good at the other.
Thanks for the antenna mount tips. I could probably source such supplies locally and not pay for likely expensive antenna mounting hardware.
@David_Thomson “ Does that mean ideally the antenna should be at least 20ft off the ground? “
The near by trees, in your photo, are a densely packed mass that is likely killing your link.
900MHz needs a bit of height and distance before it encounters obstruction so that it can develop a usable Fresnel Zone and please take note I said usable not perfect.
Get the antenna above that mass or find someone with a can of gas and a good chainsaw or better yet a bulldozer.
If your entire path is obstructed with a large mass such as that shown in the photo then you will need more height on both ends.
Trust the math, test the theory but until you do the work you will never see results.
Did you ever run a LinkPlanner simulation for this path?
I’ve spent some time thinning the trees to help penetration I can see thru them now. I have the 25ft of “runway” between the yagi and the trees for the signal to propagate.
The Link Planner software estimates a usable, ok link given the design I provided and their own algorithms for forest density, etc. I think part of my problem was the topo with the antenna height so I’ve modified that to get the fresnel zone off the ground (partially before).
Lighting forest fires is not a good joke, but I get your point. I’ve been using a chainsaw. 900mhz is supposed to function with obstructions - not solid, but I’m not going to give it a clear ROW after spending 6x over a decent 2400mhz setup.
Let me clear something up. I was not suggesting you or anyone burn anything notice “and a”.
Let’s imagine for a moment you and pitch black room and a flashlight. If you put your hand over the reflector and turn it on you will still be in the dark. But as you increase the distance between your hand and the reflector more of what is on the other side of your hand will be illuminated.
Unlike your hand you cannot move the mass so you need to raise the antenna.
1 Like
My mistake on the gas bit - don’t forget the premix.
Unfortunately raising the antenna over the trees is not feasible - they are over 50ft high, and the antenna tower needed for that is not in the budget. But I can thin the trees. I have thinned the trees and on one end about half of the fresnel zone would go thru a thinned portion of the stand that is quite open. I could do the same on the other end, and if I could get some sort of link I could justify the time to thin the beam path to improve bandwidth.
But yesterday’s test ran into a new complication. Don’t see this stuff from the ground… we moved the antenna spot to the far end of one building, further away, to “raise the antenna height” and improve the outcome on Link Planner. But once I was up there to test the link I noticed the antenna will have to be Xft above the roof to have a clear fresnel zone. The building owner doesn’t like the idea of a ~20ft antenna mounted to the side/crown of his building (its old, wood framed, etc).
Trust the math. Test the theory. Do the work.
From your response it seems you are running out of options. You say you can’t go up any higher so that leaves you with going thru.
It it were me and I really needed the link I would cut everything, or what I am allowed, down that was in the path. Remember to plant two tress for every one you cut down.
Trees, bushes and undergrowth continue to grow so if the link is marginal now it will only get worse with time.
There are no guarantees that your efforts will pay off. The only thing certain is there is no chance the results will improve if you do not Do The Work. .
1 Like
My advice at this point - is to pack up and leave.
It took years for us to figure that out, and we would do whatever we could to appease the client’s design or budget or aesthetics limitations imposed on us. Time and time again, the client would call back and say “it’s not fast enough” or “it sucks” or “come fix it”…
…and the FIX was always this. “MOUNT THE ANTENNAS WHERE THEY NEED TO BE MOUNTED, IN ORDER FOR THE LINK TO WORK CORRECTLY.”
So now - when a client says “I don’t want that on my roof”, or “You need to mount it on the other side of the house where I can’t see it”, or “Only use a short mast - not the tall mast”… if there isn’t another technical solution WHICH ALSO PROVIDES US ENOUGH LOS AND SIGNALS AND SPEEDS NEEDED… then we explain that, and pack up and leave. 
There are almost always options: A bracketed tubular tower, or a wood power pole, or a repeater off a barn or Quonset which is tall enough - or trenching a cable to a location where it is clearer (with MicroTik GPER cables can go a LONG ways), or a 50’ guyed telescoping mast on the roof… or something that will work properly. But if none of those options work for the client or their budget or their sense of aesthetics, then my advice is don’t do half-assed / kinda functional internet for them. Give them the options which will work - and then pack up and leave. 
1 Like
But… for what it’s worth… if this does link on the bench, and if it does work thu other trees where you can test it in a controlled environment where you’re confident of the aiming… so if you’ve confirmed on bench and LOS and NLOS that your gear is setup correctly… then I’d be suspect of this…
900 Mhz shoots through a surprising number of trees, but not much dirt. 
So, that would be a suspected culprit, and I’d probably want to go there (to the top of that bump) with the SM & mast, and a battery pack, and a tablet/phone - and see if I can link from the ground there.
I’m just saying that it doesn’t take too much to be off on the height estimates to go from a ‘plausible’ link, to ‘no chance in hell’ link. That can literally be 5’ or 10’ difference in height, if you’re shooting thu dirt. 
AND - if you’re wondering if more height at the client side is really a solution – look into renting a manlift and see. I read back and I understand there’s not a lot of customer budget… but a towable manlift with the SM and battery and tablet in the manlift bucket… that can very quickly tell you how much height is the solution.
3 Likes
One more thing… I’d ditch the UBNT antennas, they are crap IMHO… get some KP performance 17dBi dual pole antennas. They are the gold standard for 900MHz IMHO. If it doesn’t work with two of those KP’s then it ain’t gonna work period.
4 Likes
Unfortunately I am the customer. The building owner is the neighbor nearest me with broadband that’s willing to let me setup an account (under his business account, which is very nice) and put a wireless bridge antenna on his building.
What’s this about a repeater? Could I use two bridges, one to get the internet from his building to another structure on his property with a better path to my house? I have a 2400mhz PTP setup that I could use for the first bridge (hundreds of feet), then the 900mhz setup for the 1/3mi thru the forest but less topo.
Any way to buy just two? I don’t really want a 5 pack for $650 for another proof of concept test.
Bench test - golden. I have one on the roof pointed towards the neighbor, and the other in my office two stories below (inside), and they link without even aiming them towards each other (Maybe this stresses Eric’s comment that UBNT antennae suck?).
I need some sort of 120v battery setup so I can take the antenna midway to test a partial bridge - isolate the problem stretch perhaps. Can I use a crummy generator? I bet there’s a ton of THD off of it so I was worried about the radio.
Aiming - ahem. Well, that’s been a process. Finished? Perhaps not. Magnetic compass (with declination adjustment for here since my coordinates are absolute north referenced) is unreliable since both my house and the other building are metal roofed. Phone compasses are garbage for things like this in my experience. So I was using a print of the actual bridge path as a line feature on a satellite image to visualize where it would be relative to features on the ground.
I can’t do that with the tilt (6 degrees given the elevation difference and bridge length, calculated online) and both adjustments are pretty sensitive to single degrees eh? I think 1 degree off means no link but my geometry fu sucks. Probably my only way to do this without LOS is to keep going part way and confirming the link/aim to confirm its on the planned path.
My neighbor has a lift - and is letting me use it. I can try higher just for proof of concept, but first I should confirm the aim and partial distance bridging.