First B24 Link - Signal and Capacity Fluctuation Questions


#1

Hi, We’re just testing our first B24 link and hoping to put into production this week. I have been noticing several fluctuations in watching the web interfaces and wondering if I’m seeing interference or possibly have an alignment issue. Any comments/questions welcome. The link is about 1.2 KM long, one radio on a roof mount at valley bottom where we connect to fiber, one about fifty feet up on a tower with clean line of site where we do our distribution from. Main signal reading is about -57 on each side. I’m running at 2x40MHz channels as the throughput on the 2x80MHz seemed worse.

Observations:

  1. The upload direction of the link’s capacity and speed is showing much better than the download’s:
    PHY Tx/Rx (Mbps): 480 / 800
    MAC Tx/Rx (Mbps): 288 / 480
  2. The odd time I see one of the MIMO streams drop off
  3. I see the link length estimate jump to 6000km once in a while
  4. Normally the PER stays below 2% but the odd time on the download side it jumps to 5 or 10% briefly
  5. EVR seems OK - typically above -25 although some of the streams drop to below -20 occasionally
  6. The SNR of the streams is 25-30 on the tower (download) side and 30 on the valley side.

On the tower itself there is a lot of our 5.8GHz Cambium gear and some radio antennas for the local fire department and trucking company. We are in a small town a fair ways out in the country. Looking at the spectrum analyzer for the link we see lower than -90 for noise.

Numbers looked a bit better yesterday when it was a clearer day, today is a bit cloudier. I was pulling about 430Mbps download and 480Mbps upload when directly connected to the router at the tower yesterday. This matched up about with the MAC rate at the time so I think the link is working OK.

I’ve watched the aiming videos and I think I have the alignment good. Right now there is minimal traffic going through the link but I’m running out of time on testing, getting lots of pressure to bring it live as currently we are not adding new subscribers at this tower.

So wondering if these are just normal signal variations at this distance and I shouldn’t worry too much about watching the real time data, just trend things over time and go from there? Or would any of this possibly indicate an alignment or interference problem? Or does the link need to be under more of a load for accurate stats?

Thanks for your time all.
-Nick


#2

PER should not be that high if there isn’t any noise and you have a good moment. I would expect some sort of noise to be causing the issue, especially with you mentioning that the link is worse on the Tower end and less so on the other. But if the spectrum analyzer is not seeing anything maybe not, do you have the B24 connected to the Mimosa Cloud?

I have not specifically used the B24, but I have worked with other 24 GHz gear. Aiming can be a huge PITA, you are working with a very narrow beam and sidelobes are very tricky.

I would try lowering the power on both sides and redoing the alignment, swing through a full 40 degrees horizontally and vertically to verify you are not on a sidelobe. Lowering the power will make sidelobes less powerful/probably not connect at all. (On Friday I was doing an install with one of our techs and he could only get -67 on a PTP link that should have been in the low 50’s, he was insistent that the alignment was good and he had checked, even when I showed him a nearby link that was pointing to the same tower and was in the mid 50’s. It was foggy out so he couldn’t see the tower so, I went up and initially almost believed him as I started going through the horizontal scan, but I kept going and caught the mainlobe and dropped it down to -51 with hardly any work.)

After that, you might want to have a chat with Mimosa support, there could be something up with your Tower radio. They will need remote access to it, just as a heads up. Before you contact them double check to make sure all the magical check boxes are marked off. (Grounded, no random crap blocking the antenna and other silly things that support people have to ask “just to make sure”)


#3

Thanks for the response, I hadn’t thought of lowering the power to test aligning to rule out the sidelobes, interesting idea. I’ll maybe try and pop up again and do the big swing of the radio but when I mounted it I did have a clear day and can see the other side so hopefully not too far off. But definitely from playing around this weekend I can see how touchy the 24 GHz can be to aim so I’m not ruling out alignment. I think I have most of the usual boxes checked for installation and yes they are cloud connected, no problem to give support access. And the historic cloud graph doesn’t catch any PER spikes that high, it looks fine so that may or may not be an indication of anything. Looking back now on the PHY graph and the upload stays right around the 800 mark while the download size is more toward the 400-500 mark with much more fluctuation. We haven’t settled into full winter yet, have a bit of a warmer spell right now, maybe 5 below 0C and lots of frost out so that might explain some of it. But the different readings on each direction worry me a bit. Thanks again!


#4

Hi @Nick8,

What is your B24 target signal level? Also are you using the current 2.5.0 firmware that came out last week?

The new 2.5.0 firmware has improvements for Auto TDMA mode if you’re running Auto mode (by default Auto mode is on).

I’m out of the country this week otherwise I’d take a look at this. Please visit with Mimosa support so they can look at your radios through the cloud.


#5

Thanks Dustin. I will load the new firmware later today, missed that release. Target on the tower side is -57.4 and I’m sitting a hair above that at -57.1. On the valley side I’m below target. Target is -54.4 and I’m at -56.4ish. If that’s enough of a difference to mean an alignment problem I’ll head back up. But I’ll check in with Mimosa support as well. Thanks again.


#6

Are there any significant differences between your chains? (like +2 dB difference)


#7

Just did a quick check and on one side (valley side) I do see a little variation, up to 2DB difference at times it looks like. Little tighter on the tower side but coming close to 1.5DB difference at times. Is the desired goal to have them as close together as possible? Thanks!


#8

Always.

If they are symmetrical then all the other parts of the signal get a lot cleaner.

That said, under 2 dB is pretty decent. Probably better for an antenna the size of the B24, but as I said before I have not worked with the B24 itself, just a lot of other Mimosa equipment and other 24 GHz equipment.

Is anything interesting popping up in your logs? Never hurts to check.


#9

Well nothing in the logs that stands out. I just loaded the 2.5.0 firmware and post reboot I am seeing some much bigger differences between the chains - I assume this means it’s fluctuating some. image Hopefully the picture comes through OK. This is tower side. Sorry, first Mimosa radio link plus first 24GHz link so am still learning through this setup. But I’m guessing this means I need to tighten up the alignment?


#10

@Nick8,

I would check alignment first, while you are there a quick check of the hardware and to make sure that nothing is wrong with the link.

Unfortunately, with PTP links you may need to visit each location more then once to make sure that both alignments are good. Or have someone else on the other side of the link you can do the alignment with, just make sure only one of you are making adjustments at a time.


#11

Pol 2 looks to be off on alignment, or it could be getting a bit of a Fresnel zone obstruction/reflection from something nearby.

Please make sure that you aren’t clipping the tower, a roof, or even a tree. These truly require line of sight. It doesn’t look very bad, but it will cause performance degradation.


#12

Thanks all. I’ll play around with the alignment some more as soon as I can get up there, think it’s really frosty out tomorrow so might have to wait a couple days. Have a 5.8 PtP shot in identical location on roof and tower providing the distribution bandwidth now so in theory the path should be clear for a 24 GHz shot, so I’m leaning more toward alignment vs Fresnel interference. I did work with support yesterday and we did catch a small bit of interference on the spectrum analyzer at one point so possible we have a bit of coming and going interference that could be a factor as well. I’ll report back after I do a little fine tuning.


#13

Just an update, I did fine tune things this weekend and definitely better, well under the target signal levels on both sides, tighter grouping of the signal strength differences in the chains. Just for fun I switched it back to 2x80 for channel size for a bit and weird things like cloud connectivity data gaps started happening and at one point it just stopped passing traffic so possibly fighting some interference, not sure, or perhaps further fine tuning of the alignment would help with that too? Switched it back to 2x40 and things have been ticking along pretty good since - sticking at 8-9 MCS rates in general. This should give us the tower link a bump up in capacity. Thanks for the assistance everyone, much appreciated.


#14

No problem @Nick8.

Glad to be of help.

I agree there seems to be some sort of interference, a good way to test it would be to move around in the spectrum with your 40 MHz channels and see if you get modulation drops.

You might want to have a chat with Mimosa support about the cloud connectivity issues. That shouldn’t ever happen.


#15

And just a follow up, after a few days running at 2x40 cloud data is updated 100% solid. So maybe a bit of interference with a bit of a signal drop going to 2x80 at the distance I’m working at just won’t be perfect where this link is. I can pull a linux distro at full speed so things looking good, going to start moving customer traffic over.