Page 1 of 1

Tower Based LTE Project

Posted: Sat May 02, 2020 9:32 pm
by pstephens
Inspired by swwifty's epic thread I'm planning to build a tower based LTE setup to improve my internet connectivity.

Some history:
  • I've tried fixed wireless in the distant past, but, without line of site to the tower, results were mediocre.
  • More recently I've used DSL. But the quality of the copper is poor. It works well part of the time but has packet loss and high latency the other part of the time. Probably noisy corroded copper interconnects. I've had the AT&T tech out multiple times to no avail.
  • I've also tried an AT&T hotspot (ZTE MF279), also with poor results. Added an external 20 ft antenna (but didn't understand MIMO at the time, so only one antenna). Then tried a booster. All variants have had noise based latency and packet loss problems.
  • Most recently got a MOFI 4500 SIM4 (also in ignorance) to obtain better low level access to the modem. This has been instructive, but I'm still having the same problems.
  • But! Driving around the neighborhood with the MOFI has shown that better connectivity is possible. Unfortunately I live in a small valley. Thus...
The Tower. I already have a Universal Towers 21-70 70 foot tower on order. This tower is over specified but I've had trouble determining the wind loads for the proposed antennas (see below) so I'm erring on the side of caution. A beefier tower should give more margins in a wind storm.

Raising the Tower. The tower is hinged by removing 4 of 6 bolts at the base. Universal's instructions are to walk the tower up by hand, but I'm skeptical this will work for a 70 ft tower. Also I would like to lower and raise the tower single-handed and so I'm planning to configure a winching system. I've whipped up a [url=http://www.diligentsoftware.com/test/tower.html]web page[/url] to model the loads on the tower at various geometries and tower angles. Turns out that the highest tension is experienced while the tower is flat on the ground. Moving the gin pole (where the pulley is attached to) closer to the base of the tower helps. Increasing the gin pole height reduces tension but increases gin pole torque. Anyway, current plan is to locate the tower close to the back of my garage which will give me the equivalent of a gin pole 11 ft tall offset from the hinge about 9 ft. Maximum tensions experienced are around 1400 lbs and how to safely attach the cable to the tower is "to be determined".

Antennas. Currently planning on using two Bolton Technical's UltraGain 26 Directional Antennas in a 2x2 MIMO configuration. I'm not sure if I need this much gain, but I'm willing to over specify to ensure project success. The count of line of site cellular towers (from the perspective of the top of tower) is around 5 for T-Mobile (my current SIM) ranging in distance from 3.5 to 10 miles. The tower with the most bands available is also the closest but also more populated and so maybe more congested. We'll see how things look once everything is set up.

Pointing. These antennas are highly directional; between 7 and 20 degrees for a half powered signal depending on frequency. There is no good way to hand point the antennas from 70 feet below, so a rotator is required. I'm considering the Alfa Spid RAK rotator but the stock controller looks dated. An upgraded RT-21 controller from Green Heron improves the situation with a built in power supply, better acceleration/deceleration control, and maybe 1/10 degree of precision. Not 100% sure about the precision as this is still an open research topic.

Feedline. The LTE modem will be stowed in an equipment box at the base of the tower so requires about a 70 ft feedline run. Originally I was considering relatively standard LMR400. But LMR600 looks even better (and more expensive). And for about the same price as LMR600 I can also get Andrew Heliax LDF4-50A. How do these compare? For a 70' run and an 1800Mhz signal: LMR400 has 3.99 dB, LMR600 2.59 dB, and LDF4-50A has 2.15 dB of attenuation. Is the performance gain (52% better attenuation, or 60% efficiency vs 40% efficiency) worth the cost? I don't know. But if a 50' LMR400 run is acceptable than a 75' LDF4 run should also be acceptable.

Some additional feedline notes:
  • Frequency range is important. It would be nice to upgrade to Sub 6Ghz 5G someday. LDF4 can do 8.8Ghz while LMR600 can only do 5.8Ghz and LMR400 can do 6Ghz.
  • Alas, millimeter band 5G seems out of "reach" without moving the modem closer to the antennas. I haven't found any coaxial feedline that can do > 24Ghz frequencies for a significant distance. (Side note: I keep seeing mention that 5G millimeter band only has a range of several hundred meters. But I find this hard to believe as satellites hundreds of miles up often use Ka band to communicate with the ground.)
  • More flexible coax is needed through the rotator. LMR-400-UF UltraFlex might work here as it has a braided core.
  • Lots of N connectors will be needed as they have a built in weather seal.
Lightning. There will be at least 3 grounding rods for each of the three legs of the tower. I might have a 4th rod dedicated to the equipment box. All grounding will be bonded together, including with the residential ground. I'm still not sure whether I'll also want to put a lightning rod at the top and/or a feather duster to dissipate charge. This is an ongoing research project.

Modem. I'm going to continue with the MOFI 4500 SIM4 for now. It's possible a higher category modem would improve carrier aggregation and available bands. For T-Mobile I'll be missing B66 and B71, but very few modems support B71 yet. I'm starting with a T-Mobile SIM, but ATT might be a good option to try once the dust settles. I would be wildly thrilled if it appears that I'm bumping into MOFI's 100Mbps ethernet limitation. A future upgrade might be a CAT12 or CAT20 modem paired with a Raspberry PI 4. Or maybe 5G modems will become affordable one day?

Power for the modem will be provided from within the residence via PoE rated at a sufficient amperage.

Summary. I'm trying to retire as much risk as possible through research and design. I'm tired of naive solutions that disappoint. I'll try and keep this thread up to date as the project progresses.

Opinions & Ideas welcome!

edit: formatting

Re: Tower Based LTE Project

Posted: Sat May 02, 2020 10:35 pm
by gscheb
Hello,
What is your final goal? What potential speeds are you wanting to get out of this?

Re: Tower Based LTE Project

Posted: Mon May 04, 2020 11:00 am
by pstephens
gscheb wrote: Sat May 02, 2020 10:35 pm What is your final goal? What potential speeds are you wanting to get out of this?
My original reply seems to have been deleted. Maybe forums were restored from backups?

Original gist: Reliability over bandwidth. Low latency over bandwidth. Good bandwidth would be icing on the cake.

edit: spelling

Re: Tower Based LTE Project

Posted: Tue May 05, 2020 2:31 pm
by Sambb198
I have the same exact issue as you. Would love to see how it turns out - maybe do the same.
Currently have two yagi antennas installed to the same Mofi modem.
I have a question about your setup. How do you plan to connect the 4 antennas to the Mofi Modem?

Re: Tower Based LTE Project

Posted: Tue May 05, 2020 7:56 pm
by pstephens
Sambb198 wrote: Tue May 05, 2020 2:31 pm I have a question about your setup. How do you plan to connect the 4 antennas to the Mofi Modem?
I actually only plan on using 2x2 MiMo. So only two antennas will be necessary for my build.

For a 4x4 MiMo setup I would need to make several upgrades:
  • Modem. Acquire a modem with 4x4 MiMo. The Sierra Wireless EM7455 only supports two antennas, so I would need to upgrade to a Quectel EM20-G or similar.
  • Feedline changes
    • Either add 2 more feedline coax cables from the top of the tower down to the planned enclosure at the base. 140 more feet of cable, so a bit more money.
    • Or move the enclosure to the top of the tower. This would actually improve antenna signal a bit and reduce cost, but I would need to lower the tower every time I wanted to touch the hardware.
  • Antenna changes
    • Either acquire a 4x4 MiMo antenna with 4 antenna ports. Haven't seen anything like this with much gain. Here's an 8dBi 4x4 MiMo antenna for example.
    • Or acquire a pair of 2x2 antennas and physically separate them by around 4-6 wavelengths which is about 5-8' for 750Mhz, Or so I've heard.
    • Or acquire two more 1x1 antennas and cross polarize them in a pair but also physically separate from the existing pair. But my mast is only 8' (probably 6' above the thrust bearing) so it might be hard to place the Ultragain 26 antennas properly.
4x4 MiMo would be an interesting project for later. A minimal experiment would be to try the lower gain 4x4 MiMo antenna and temporarily move the equipment box to the top and see what happens. Ethernet + PoE would run down the tower rather than coax.

Re: Tower Based LTE Project

Posted: Tue May 12, 2020 10:01 pm
by Didneywhorl
I'm currently playing with the 4x4 mimo flat panel. Its a fantastic unit. I'm not far from my tower, but it technically is better tuned than my pair of dual mimo narrow band 15dbi antennas. I have it connected to a weatherproof enclosure via 4 5' LMR200 cables. More testing to come.

Re: Tower Based LTE Project

Posted: Sat May 16, 2020 2:25 pm
by pstephens
Update:

Got the big hole dug for the tower footing. Ended up being a little deeper than needed so will need to backfill and compact a bit.

Next steps:
  • Design a proper grounding system to safely dissipate the inevitable lightning strike. Seems to be a lot of guesswork without employing significant engineering. The big unknown for me is soil resistance for my location which has significant quantities of limestone. Multiple ground rods radiating from the tower base, all bounded together, seems to be the solution. But how many rods?
  • Ufer Ground. I can get some additional grounding performance using an Ufer Ground which is basically utilizing the concrete footing as a ground.
  • Forming the footing. I dug the hole larger than needed with the anticipation of hitting solid limestone while installing 4 ground rods on the outside corners. Turns out that I didn't encounter rock until about 4 feet down. Instead of a larger hole, I should have dug the hole of the appropriate size (5' x 5' x 6' deep) and drove ground rods separately. Anyway, now designing forms to handle concrete pressure at 6' deep. Here's a good reference: Concrete Forming Design/Construction Guide.