My 4G setup/project

This a discussion place for anything.
swwifty
Posts: 390
Joined: Thu Aug 30, 2018 8:21 pm
Location: NE GA Mountains

My 4G setup/project

Post by swwifty » Tue Nov 20, 2018 7:47 pm

I wanted to do a complete post on my setup with context, and technical details on what I did. I'll start by giving some context on the why.

Last year I moved to the Northeast GA mountains. I work as a Systems Engineer, and work full time remote. Thus, as part of this move, fast and reliable internet was a must. I was fortunate that the home and location we picked has good reliable internet (Dual bonded DSL that gets me 21 mbps down / 1.25 mbps up ). Most of the county I live in does not. When we moved, I switched from T-Mobile phone service, to Verizon, because T-mobile at the time did not cover this area. I knew I could use tethering as a backup (it was in fact faster than my DSL at about 30mbps down / 15 mbps up, on my phone) for my primary wired internet connection, but never thought of using it as a full time secondary connection.

Fast forward to August of this past year. One day (for reasons I cannot recall) I started researching using a 4G connection as a primary internet connection. I stumbled upon a provider (who will go unnamed for a variety of reasons) that sold service with AT&T, and provided a Mofi 4500 router. Not knowing much about LTE and the associated hardware, I decided to take the plunge as they offered a 30 day no questions asked return (at the time I didn't even know if AT&T would work in my area for sure, as I only had phone service with Verizon.). I got the modem and it did in fact work (well enough that I continued to pursue making this a permanent solution). I started digging into it and learning, and after a while, I discovered the ltefix store. I quickly realized I could build a much better version router/modem with my own pieced together hardware. I had a raspberry pi 2 that wasn't doing much, so I opted to put together a setup with that. This led me down a serious rabbit hole of 4G hardware, learning about how LTE works, and doing tons of experiments in different locations with different antennas to learn as much as I could through experience. I won't go into all of that here, as you can find alot of those details in thread here: http://ltehacks.com/viewtopic.php?f=24& ... f90f2e4259 which reviews flat panel antennas sold by ltefix store.

Now, let's get into some of the technical details of what I settled on:

Main hardware:

Raspberry Pi 3 (Running latest version of GoldenOrb aka rOOter)
Sierra MC7455 Modem
2 - 17dbi 1900mhz (band2) Yagi Antennas

Supporting Accessories:

USB Enclosure for modem
PoE splitter
Gas Lightning Surge Protectors
Mini Circuits NHP-600+ High Pass Filter
10ft LMR 400 cable
4ft LMR 195 cable
10 foot antenna mast and gable mount
40+ feet of outdoor 4 pair ethernet cable (Serious thanks to this ethernet cable from my co-worker who gave this to me for free.)
Outdoor Waterproof enclosure

I ended up mounting the antennas on the south side of my home (the direction the tower is from me, aimed them with a compass cause I don't have LOS even with no leaves on the trees). I put the PI and all the needed parts on the outside of the house, so the antenna cables could be as short as possible. Otherwise, I would have had to have 40+ feet of cables, and some serious signal loss through them, so I opted for a outdoor setup. After some seriously extensive testing, I opted for now, to use the 17dbi yagis which are designed only for band 2. Where I live, I can get signal on band 12 (700mhz), and band 2 (1900mhz). I found through extensive testing, that band 2 was much faster, even though the signal strength was much weaker from it. Band 2 is 20mhz wide and Band 12 is 10mhz. I also suspect that band 12 is much more congested where I live, because there is very few cell towers, and users on the outer edge of the cell (like me) are most likely using band 12 as their primary cell, and band 2 as their secondary if they are lucky enough to get any signal. Hence, this is why I opted for this setup. When I was using antennas that gave me the ability to use carrier aggregation I found that they weren't much faster at all, then just band 2 alone. Time will tell, but I have the flat panels still in the event I need to fall back to them.

Time for some pictures.
6500562927016073818.jpg
Yagis on the top, below is a Ubiquiti Nanostation that I moved outside as part of this project. It's used to connect my Mom's home (about 500 feet away from my home) to my own network, so she can freeload internet off of me, hehe. On a side note, moving it outside, has made the signal much more stable. I previously had it in a window in my office. Her's is mounted on the outside backside of her house high up near the roof line. I'm also going to add a guy wire setup to this mast. It's fairly stable right now, since the yagis are fairly light, but in the future if I use the flat panels for some reason I don't want them up there unsupported. They are significantly heavier than the yagis. Plus, I want to be able to add more antennas up there for my Software defined radio possibly in the future.
2018-11-18 15.52.35.jpg
Waterproof box for Pi/Modem. On the left you can see the gas surge protectors and the high pass filters. On the right is a ubiquiti grounding device for the ethernet. I have yet to finish grounding the antenna cables, and antenna mast to the ground block, and then taking that to the ground. (I just finished all of this last Friday). Hoping to complete the grounding this weekend, or next.
2018-11-18 15.46.35.jpg
Inside the waterproof enclosure. Modem on the left, Pi in the middle, PoE splitter on the far right. I used velcro to attach all of this (some very strong stuff), so I can remove it, if I need to work on it and not have to remove the entire box from the wall, etc.

Finally, the all important speed test.
finished project speed test.png
Also, an RF profile of the terrain I have to pass over to make a connection. This tower is 5 miles south of me. Tower is on the far left in this profile, my house is on the right.
Screen Shot 2018-11-20 at 8.40.00 PM.png
Edit: forgot to include a picture of signal strength info
Screen Shot 2018-11-21 at 6.59.48 AM.png
You can see the signal strength isn't all that great on band 2, but it's enough to get me 85mbps down and 15mbps up on non-peak hours.


All of this 4G setup, is connected to a firewall I have running PFsense. This firewall then load balances the traffic between my 4G connection and my DSL connection. It will appropriately remove a gateway if one of the connections goes down. This accomplishes exactly what I was trying to do. I ultimately wanted a backup connection, but didn't want to pay for something I only used when my main connection was down. Now I have the best of both worlds!

Also as a part of all this, I worked through totally revamping my internal network. I purchased also a managed ubiquiti switch, so I could setup vlans to isolate traffic for various internal parts of the network (guest wifi network, etc.) There was a lot of work done to accomplish this, and I'm very satisfied to have 90% of it completed now.

I'm sure at the moment I'm leaving out some details (I'll update this post if I think of anything else I forgot), but please feel free to ask any questions.

swwifty
Posts: 390
Joined: Thu Aug 30, 2018 8:21 pm
Location: NE GA Mountains

Re: My 4G setup/project

Post by swwifty » Sun Nov 25, 2018 6:06 pm

Today, I decided to re-do the water proofing on my coaxial cable. I had originally, only wrapped them with some coaxial seal, but realized that wasn't correct method. It rained the first time this past Friday evening, and one of my antennas (the main one) suddenly had 10dbm (10 times less!!) less power received for RSRP signal. I couldn't figure out why my main antenna would have so much less power received than the other diversity antenna. The only thing I could figure is some moisture or water got into my connection on the main antenna. I did some research on how coaxial seal is normally installed and realized I didn't do it with the ideal method. The ideal method is wrapping first in electrical tape, then coaxial seal, then finish with electrical tape. When I took the connections apart, I didn't detect any water in the connectors, so time will tell if that was the issue (or I had an environment / RF issue, or something else).

During this work, I opted to do a quick comparison of using one antenna to two antennas. I wanted to see what level of MIMO i was getting in this final location (I previously did some mimo tests in other locations).

Signal stats with both antennas attached to the modem:
Screen Shot 2018-11-25 at 4.21.31 PM.png
Screen Shot 2018-11-25 at 4.21.31 PM.png (59.62 KiB) Viewed 2494 times
Speed test with both antennas (this at 3pm in the afternoon on Sunday, aka peak hours)
Screen Shot 2018-11-25 at 4.21.25 PM.png
Signal stats with only one antenna attached (the main one, which is the highest one on the antenna mast as well)
Screen Shot 2018-11-25 at 4.23.11 PM.png
Screen Shot 2018-11-25 at 4.23.11 PM.png (61.13 KiB) Viewed 2494 times
Speed test with only that one antenna attached:
Screen Shot 2018-11-25 at 4.24.05 PM.png
As you can see, I'm clearly getting spatial multiplexing (aka two data streams, or what we commonly call MIMO) which is giving me double the data rates in download. Uplink is the same, as uplink does not use spatial multiplexing, and only transmits on the main antenna.

And last but not least, I'm starting to detect a lot of other new cells, that are on the same frequency. I think they are cells on the same physical tower (just different sector antennas). I was detecting a few others (other than the one I was connected to obviously) when all the leaves were still on the trees, but I think I'm detecting a lot more now than the leaves are mostly all on the ground. I also think this is why I see my RSSI values fluctuate between about -78 to -91 dbm. If I recall correctly, RSSI is calculated including interference from other cells.
Screen Shot 2018-11-25 at 4.29.52 PM.png
Screen Shot 2018-11-25 at 4.29.52 PM.png (60.28 KiB) Viewed 2494 times

User avatar
JimHelms
Site Admin
Posts: 708
Joined: Tue Dec 19, 2017 8:59 pm
Location: DFW Texas
Has thanked: 45 times
Been thanked: 47 times
Contact:

Re: My 4G setup/project

Post by JimHelms » Sun Nov 25, 2018 8:23 pm

I commend you on your installation. Neat plumbing job as well.

To weatherproof outdoor connections, I use dielectric grease inside all connectors before screwing them together. Once together, I seal the outside of the connectors with F4 self-fusing Silicone tape.
These users thanked the author JimHelms for the post:
Dart67 (Wed May 01, 2019 12:34 pm)

swwifty
Posts: 390
Joined: Thu Aug 30, 2018 8:21 pm
Location: NE GA Mountains

Re: My 4G setup/project

Post by swwifty » Sun Nov 25, 2018 8:27 pm

JimHelms wrote:
Sun Nov 25, 2018 8:23 pm
I commend you on your installation. Neat plumbing job as well.

To weatherproof outdoor connections, I use dielectric grease inside all connectors before screwing them together. Once together, I seal the outside of the connectors with F4 self-fusing Silicone tape.
Thanks Jim! I thought today about putting some plumbers tape on the threads, but I figured what I did was good enough. I learned the technique online from guys who do cell tower installs, so I figured its probably sufficient :)

swwifty
Posts: 390
Joined: Thu Aug 30, 2018 8:21 pm
Location: NE GA Mountains

Re: My 4G setup/project

Post by swwifty » Mon Nov 26, 2018 12:33 pm

I just had an interesting "a ha!" moment. Yesterday, after I re-did the water proofing on my coaxial cables, it started raining. I've been monitoring closely to make sure my signal stats stay the same between antennas. Most of last night it rained, and this morning, it cleared up and is now fairly windy with gusts up to about 20mph I'd guess.

I notice this morning that the RSRP values on the main antenna (which is setup in a vertical orientation for vertical polarization) is flucating again between about -112dbm and 120dbm, while the secondary antenna (which is horiziontally oriented for horizontal polarization) is flucuating between about -107dbm and -112dbm. I actually don't think this issue, is related to what I think was a water ingress issue this past weekend. When that happened my signal strength weakened to about -124dbm, and stayed there for almost a day.

A screen shot of the signal stats I'm referring to. The RxM is the main antenna (which is the highest one in the pic below) stats, and RxD is the secondary antenna (which is the lower one in the pic below)
Screen Shot 2018-11-26 at 12.57.29 PM.png
Screen Shot 2018-11-26 at 12.57.29 PM.png (55.8 KiB) Viewed 2483 times
2018-11-26 12.41.02.jpg

A little back story on how I got to this theory. I have two Ubiquiti Nanostations that I use to connect my Moms house that is about 500 feet away from mine (for internet and network, etc.) I've learned a lot about wifi technologies as well, throught out my adventures into 4G. The nanostations have two antennas as well for MIMO. They have a vertical polarization antenna, and a horizontal one. I've noticed there are differences in signal strength between these polarizations, and in doing research, I learned that vertical polarizations are effected most commonly by vertically oriented objects like trees, and horizontal polarizations are effected mostly by horizontally oriented objects (like roofs for example).

Fast forward to today. I think what is going on is the vertically polarization antenna at the top (my main one) is being effected by the trees swaying in the wind. I don't have direct line of sight (even if the trees at my house were all cut down). There is a few ridges that are in between my tower (as depicted above) that are slightly in the way, along with the trees.

Now the question is whether I should change the top antenna to horizontal polarization, or just leave them as is, so I have polarization diversity. Also, below is a screenshot that shows my view to the tower. This picture is about 75 feet higher than the roof line of my house, so I can see the tower. I took this picture to help me orient my antennas further, and give me some reference points, as I can't see it directly on my roof.
2018-11-26 12.16.13.jpg

swwifty
Posts: 390
Joined: Thu Aug 30, 2018 8:21 pm
Location: NE GA Mountains

Re: My 4G setup/project

Post by swwifty » Mon Nov 26, 2018 1:41 pm

Also, this note from wikipedia on polarization and reflections. It might be my metal roof causing this issue too.
Screen Shot 2018-11-26 at 2.40.09 PM.png

swwifty
Posts: 390
Joined: Thu Aug 30, 2018 8:21 pm
Location: NE GA Mountains

Re: My 4G setup/project

Post by swwifty » Mon Nov 26, 2018 3:39 pm

so, after some more experimentation, I don't think this is a polarization issue. I swapped the connectors on my modem, to see if the problem stuck to the same antenna, and it did. I also took the top antenna which was oriented vertically for vertical polarization, and turned it horizontal. This made no difference.

The only thing I can figure is the antenna got water damaged, or was defective all along and I didn't realize it. I have a hard time believing there's a 10dbm signal difference between these two antennas when they are only mounted 24 inches apart (2 wavelengths at 1.9ghz)
Screen Shot 2018-11-26 at 4.06.30 PM.png
Screen Shot 2018-11-26 at 4.06.30 PM.png (55.29 KiB) Viewed 2477 times
I guess I'll see if I can get it RMA'd and go from there.

In other news, swapping the antennas, gave me an even faster new high score for upload speed of 22mbps!

User avatar
JimHelms
Site Admin
Posts: 708
Joined: Tue Dec 19, 2017 8:59 pm
Location: DFW Texas
Contact:

Re: My 4G setup/project

Post by JimHelms » Mon Nov 26, 2018 6:09 pm

Good analysis and information on signal reflections.

Since you have swapped the point of connection at both the modem and the antennas, this would sort of rule out it being a coax cable or connector.

It would appear that the antenna may be the issue.

swwifty
Posts: 390
Joined: Thu Aug 30, 2018 8:21 pm
Location: NE GA Mountains

Re: My 4G setup/project

Post by swwifty » Mon Nov 26, 2018 6:15 pm

JimHelms wrote:
Mon Nov 26, 2018 6:09 pm
Good analysis and information on signal reflections.

Since you have swapped the point of connection at both the modem and the antennas, this would sort of rule out it being a coax cable or connector.

It would appear that the antenna may be the issue.
Well, the one thing I haven't done is swapped the antennas, so they are using different cables between the antennas and the modem itself.

I'll need to swap the antennas between the cables, to confirm where the issue lies.

Keep in mind, I have that high pass filter, and gas tube lightning ground in line for both antennas as well. I've now been seeing a difference as high as 15dbm for RSRP on the antennas. Clearly something isn't right. I would expect maybe 3-5dbm, at most?

I"ll try to swap them in the next day or two, and see what happens. If it is water ingress damage, tough lesson to learn that's for sure. It would surprise me though cause when I disconnected the "bad" antennas cable to redo the connector, there was no hint of water in the connection at all.

swwifty
Posts: 390
Joined: Thu Aug 30, 2018 8:21 pm
Location: NE GA Mountains

Re: My 4G setup/project

Post by swwifty » Thu Nov 29, 2018 8:49 am

A few updates, since I last posted.

1. I went up on the roof to swap the antennas, to test if it was a bad antenna or bad cable, or both, that was causing my poor signal on one of my antennas. In the process of removing all the electrical tape, and coaxial seal, I must have broken the pigtail N connector on the "Bad" antenna, so I will never know 100% for sure, but I put the coaxial cable that went to it, onto the other antenna, and it still had the same signal strength. I suspect the issue the whole time was something with the antenna the whole time. I suspect water somehow got into it, cause it was never the same after the first rain it received while outside.

2. This past Monday and Tuesday it was very cold and windy. I immediately noticed that the signal stats on my "good" antenna, got significantly better (like 10dbm better). I don't know 100% why this was, but it's interesting. I did some reading online about foliage antennuation (which actually has lots of studies done on it) and apparently, wind can have a big impact on signal strength. It would seem in my case, it was favorable. I also noticed the nanostations I use for my short 500ft link, also had a good bit better signal strength (5-7dbm). The signal on the 4G connection, and the nanostations has returned to "normal". Although, the 4G connection is still about 3dbm stronger.

Signal stats when it was real windy. I was also able to get a upload speed of 27mb/sec when the signal was this strong! (RxM is the good antenna, RxD was the "Bad" antenna)
Screen Shot 2018-11-27 at 9.56.00 PM.png
Screen Shot 2018-11-27 at 9.56.00 PM.png (60.47 KiB) Viewed 2464 times
The signal strength now (and holding steady, note RxD has no antenna attached, I took the bad one down. Compare these stats with the ones I originally posted, I've picked up 3dbm since I put these up outside permanently, which is a doubling in signal strength!)
Screen Shot 2018-11-29 at 9.45.31 AM.png
Screen Shot 2018-11-29 at 9.45.31 AM.png (51.29 KiB) Viewed 2464 times
3. Lastly, it dawned on me that my PI 3 is currently limited to a 100mb/sec ethernet connection, due to the PoE splitter. The PI 3 does have a 1G ethernet connection, but the PoE splitter limits that. The ethernet interface also shares a bus with the USB ports, which obviously isn't great for speed either. I wanted to see if my 4G speeds were topping out at my max ethernet speeds, so I did a iperf test between my 4G PI, and my internal laptop (this is through my firewall). The results aren't entirely surprising, but its pretty clear my limit is the ethernet connection. In previous tests, I've been hovering around 85mbps max download. In all my other tests, before this was permanently mounted with the PoE splitter, I saw the same limits, so I'm not 100% sure where the limit was, but I suspect I was hitting this before I realized it.

Code: Select all

[ ID] Interval           Transfer     Bitrate
[  5]   0.00-10.00  sec   102 MBytes  85.4 Mbits/sec                  sender
[  5]   0.00-10.03  sec   102 MBytes  84.9 Mbits/sec                  receiver

Post Reply
  • Similar Topics
    Replies
    Views
    Last post