Wish me luck...

[token]

I suppose this is just me babbling about what I'm trying to do, but sharing my story.

I had a pretty decent thing with a cheap but effective setup. It has absolutely gone off the rails with this tmobile/sprint merger.. I'm over a thousand dollars in now and I have no idea if I can recreate the magic, but I'm going to try.

I have a Sprint tower 1500 feet away, and it was great. They activated 3xCA b41 and I was backhaul limited, 100 down, 10 up. No complaints. It was absolutely solid and consistent.

Then TMobile took away 2 b41 bands.. then a storm came through and the remaining b41 band became slow and awful. Then it disappeared altogether. Then I got the new "unified network" sim and I'm getting video throttled.. it's just been gut punch after gut punch.

Now where things get expensive. When that b41 radio was damaged, I upgraded to an RM502Q-AE with the intent of 4x4 MIMO making up most of the difference.. and it did, I got back to 50mbps.. but then they shut it down anyways.

I purchased 4 of the ltefix log periodics and put them in the attic. My old $20 panel worked better in the attic for my local tower, but something else happened.. my tower started dropping me and sticking me on a b71/n71 tower 15 miles away (b71 at +2db SINR and N71 at -18 SINR)... Which the old crappy panel never really was able to get, so that's interesting. I also started pulling b26 off the back, which is useless since it's 3mhz and also a horrible signal, but also something I never quite had happen before. It's double asphalt shingles on my roof, but I'm pointing through the end of the house so I don't know how that affects it.. I do know the amount of interference in-house is absurd though.

So here's where I am currently.. yet another ltefix order, got the 4x4 mimo panel with poe/enclosure.. I'm going to take a shot at that tower 15 miles away.

I'm curious about if changing to a properly spaced panel vs haphazardly strewn log periodics, getting rid of most line loss, *and* getting the antenna outside will realize enough gain to make this actually work.

Highest I could get my n71 in the attic was -10db SINR, yet it's a 10mhz channel and it was trading blows with my local tower for download speed, 5mhz b25, 20mbps.. but upload was abysmal (about 0.5-1.5mbps vs 12mbps) so I couldn't stick with it.. signal *strength* on b71/n71 in the attic is -110, I've gotten it as low as -105.. and yes, I know strength is mostly meaningless.. but I'm thinking, there's enough *strength* of the signal even in the attic, RF hell interference-wise, that if I get this thing outside, mostly away from the noise and with less obstruction/reflections from metal ducts and such... There's enough usable signal here that this might actually work.

Thing is, TMobile pre-sprint was unusable on a cell phone in the house. No bars, no network lock. I had to walk outside a few feet to even see bars. This was on b71.. which is supposed to be good at penetrating buildings.. but even still, there is no substitute for being outside and free of interference, as I go from no service to 2-3 bars on a basic cell phone about 15 feet away in my driveway..

Is it unreasonable to think the difference between the attic and the roof is going to be the difference maker in pulling in this extremely marginal signal to a useful level?

I'm going to be installing this on Tuesday (fingers crossed, weather depending) and obviously the goal is to pull in something at least reasonably close to what I had. 50mbps is my target, obviously I'll take more if I can get it.. but I will be reporting back on my results.

Also, while that tower 15 miles away is the one I seem to have good luck with in my attic, there's another 10 miles away, similar terrain characteristics, and I can pull the b2 on that on my car modem not even a mile from here at 100 feet less elevation, so *maybe* that's an option?

Either way, I'm hoping this (admittedly expensive for my very limited budget) 4x4 mimo panel antenna is going to be the difference maker, along with being outside.. there are 4 places, all just outside of my local area, with towers.. so this antenna has 4 shots at making me giddy.

I'll certainly be back to let you all know how it goes, along with my thoughts and impressions on this mammoth monster mimo antenna... Though I'm rambling, I can certainly use some inputs or thoughts, similar experiences, etc.. please, I don't want this to be a meandering monologue.. feel free to jump in.

[token]

The cat5e to the roof needed replaced.. as soon as I started working with the existing line, it crumbled. Yay, unexpected attic run! but I got it all installed.. and...

Well, RSRP is improved on that N71/B71 signal. from about -110 to -100. Much stronger signal.

SNR is worse. Whereas -12 to -16 on 5G/+3-+4 on LTE was doable in the attic, best I could pull on the roof is -18 5G/0 LTE. I spent well over an hour inching it around to locate a useful signal. That's the best I've pulled in.

I did manage to run a speed test on the random B26 I ran across. I couldn't get any throughout at all on B26 before, but now I get a few megabits from it. Sadly, it's a 3mhz channel and definitely not desirable.. but I guess yet another emergency fallback is welcome

Despite the much worse SNR on N71, upload speed is hugely improved. Best I could get was 2mbps in the attic, it's now 8-10mbps, roughly the same as the download speed. While these are well below where I wanted to be, I will acknowledge that I'm impressed -20db SNR is giving useful speeds at all.. my ping is regularly in the teens on N71.

RSRQ -15
SNR -20db
RSRP -99

These numbers seem kinda mismatched, to be honest.

The best signal/speed on the distant N71 tower is, oddly, pointing right at my local tower, 500m away, which is -66 RSRP on LTE. It's not the correct direction, off by about 40 degrees... So while it's likely I'm getting some overload from the local tower (connecting to it, pointing right at it, the SNR is lower than pointing away from it.. 15-20 pointing right at it, 22-25 pointing away by about 90 degrees).. so for some reason, the distant tower seems to like it there, even though it's not technically ideal..

I will try raising/lowering it a bit, but I don't foresee much improvement likely, to be honest.

Disappointing.. can't blame the antenna, but obviously hoped for it to overcome the challenge of this not-ideal signal situation. Back to the drawing board, I guess.

[mtl26637]

I had similar situation with Tmo setup. Had B66/n71 locked in with RM500Q-AE for quite some time and started having issues with it losing signal and booting me off and losing connection arouns xmas of last year. Never figured it out so went with a different provider for a few months before messing with it again. Hooked up Tmo setup again and had same issues. Even bought another RM502Q-AE to play with hoping it was something funny with the firmware or something. I did a hard reset on the 502 and all the sudden it latched on to B66 primary band and 200mhz n41 and I about fell over. I did have issues with the setup wanting to jump back to B66/n71 all the time so I locked bands to B66/n41 and haven't looked back since. BTW, I also did a hard reset on the 500Q I have and same thing it started picking up n41 band also. Something was wacked with the settings in both modems but started working fabulously after hard resetting both. I'm guessing through those months where I was having issues they were upgrading the tower? I couldn't be happier with my speeds at this point being 300-600mbps on each.

Also, your SINR values could be playing tricks on you. When the modem starts aggregating on additional bands/frequencies your SINR values will get worse so when your setup is outside it may be aggregating more bands or additional MIMO than when in the attic is why you are seeing worse SINR and better speeds.

As far as antennas go, I would look at what bands/frequencies you are after and then try to find the best fit antenna for that specific frequency on each modem port. I'd rather have 4 single antennas fine tuned to a specific frequency band hooked to each modem port rather than some type of 4x4 mimo enclosure with 4 ports. The "all-in-one" MIMO antenna setups usually will have lower gains across the whole spectrum and not a good fit unless you are close to the tower, IMO.

[mtl26637]

One other thing, I keep forgetting that the signal stats that show for Quectel 5G modems in rooter are incorrect. To calculate the SINR for 4G on Quectel, rooter has to take the measured SINR number from the AT command and basically double it and subtract 20 to get the real SINR value (no idea why Quectel did this). Well come to find out, the SINR value reported for 5G signal in the AT command is the actual SINR value and does NOT need to be manipulated in the above manor. Rooter does the above calculations on both the 4G and 5G signals so the reported 5G SINR value is incorrect.

To fix the issue you need to edit the "/usr/lib/rooter/common/quecteldata.sh" file. I usually manually edit this file as a quick fix but don't recommend if you don't know much about how. It looks something like this:

SINR=$((($(echo $SINRR) * 2) -20))" dB"
to
SINR=$((($(echo $SINRR) * 1) -0))" dB"

I've meant to let the rooter guys know about this for some time but keep forgetting. The above basically removes the doubling of the value and subtracting 20 without breaking the code flow (I'm no coder). This is more thank likely why you are seeing such low 5G SINR values for sure.

[token]

Hey, thanks for the heads up on the snr reading discrepancies. I will adjust accordingly.

As for the reading playing tricks, unfortunately there is no aggregation. It's just a single 10mhz n71 channel with a terrible signal.. what did you do to hard reset it?

I do have 4 of the log periodics that are sold by the store responsible for this forum. I went with the panel because there's the acceptance factor aspect.. the single panel is "less ugly" and I can put it outside.. honestly, if I was just trying to max out my n71 and could get away with it, I would've gotten 4x 91XGs and let it rip... Or even the 2x2 panel with 2 of the log periodics would've been better, but compromises...

Anyways, I went back on the roof for a few hours to further experiment. I locked to b41 and spun it around, and was able to pull in a stray signal from I didn't know where. -123db RSRP, very marginal. 2db snr is the best I've been able to get out of it, which on b71 is slow but usable. I'd love to try to aggregate it with b25 for upstream and just focus on getting b41 downstream, might actually be usable, but enabling b25 immediately switches to my local tower. Just b41 gives me basically no throughout and pings around 1000-1500ms (with about 65% dropped packets).

So I looked on cellmapper and was able to identify the tower, and get the relevant data to try to lock to that tower and *try* to see what it'll give me. Since I'm pretty much right under a b25 tower, I don't anticipate that I'll get much from the tower that's 5 miles away.. I'm honestly surprised that b41 is even in play, quite surprising.. I'm hoping that if I can aggregate b25 for upstream that the distant tower won't be as affected by me being right under this other tower, and b41 being sent from a tower will get to me easier than I can talk to it.. that's an ideal world situation, I don't anticipate it working out in such a way, but I'll give it a try tomorrow. It'll be yet another nice day for rooftop shenanigans.

The two directions I get the strongest signal pointing the panel are not at all in the direction of the tower.. RF is so fickle sometimes -_-

[mtl26637]

As for the reading playing tricks, unfortunately there is no aggregation. It's just a single 10mhz n71 channel with a terrible signal.. what did you do to hard reset it?

Code: Select all

AT+QPRTPARA=3	< Reset Modem to Factory Default

I have read several times to not use this command many times. I have seen where Quectel rep. recommended NOT using it and suggested an alternative. I do not remember the alternative but I've ran this one on both modems and things were fine. Use at your own risk.

It is already aggregating the 4G signal and the 5G signal. What you don't see is which of the 4 ports are dedicated to the 4G signal and which to the 5G signal and which are diversity or MIMO for one or the other or both. I guess what I was trying to say is a better antenna can add another MIMO channel rather than just being there for diversity on a weaker antenna and the stats would show "weaker" numbers, however, it will run faster since adding the additional path. This happens quite often from what I've seen, stats are all over the place as the modem decides wtf it wants to do and is constantly changing MIMO / Diversity paths all based on instantaneous SINR / CQI values.

Speaking of CQI, rather than focus on power/intensity/noise factors, you might look closer at the reported CQI. This number helps me get a better idea of any interference from adjacent cells. Probably the most important when you get right down to it. It is reported in the serving cell AT command on the Quectels. It ranges from 0-15 I believe. Anywhere from 11-15 is normal and under 10 your starting to have some issues. This directly plays a role in how the modem calculates its speeds. You will see the serving cell AT command ran many times in the debug menu.

I do have 4 of the log periodics that are sold by the store responsible for this forum. I went with the panel because there's the acceptance factor aspect.. the single panel is "less ugly" and I can put it outside.. honestly, if I was just trying to max out my n71 and could get away with it, I would've gotten 4x 91XGs and let it rip... Or even the 2x2 panel with 2 of the log periodics would've been better, but compromises...

Don't get me wrong, I have one too. It's just not being used. It's perfect for those situations where getting the signal isn't a problem. Much cleaner look for sure.

Anyways, I went back on the roof for a few hours to further experiment. I locked to b41 and spun it around, and was able to pull in a stray signal from I didn't know where. -123db RSRP, very marginal. 2db snr is the best I've been able to get out of it, which on b71 is slow but usable. I'd love to try to aggregate it with b25 for upstream and just focus on getting b41 downstream, might actually be usable, but enabling b25 immediately switches to my local tower. Just b41 gives me basically no throughout and pings around 1000-1500ms (with about 65% dropped packets).

So I looked on cellmapper and was able to identify the tower, and get the relevant data to try to lock to that tower and *try* to see what it'll give me. Since I'm pretty much right under a b25 tower, I don't anticipate that I'll get much from the tower that's 5 miles away.. I'm honestly surprised that b41 is even in play, quite surprising.. I'm hoping that if I can aggregate b25 for upstream that the distant tower won't be as affected by me being right under this other tower, and b41 being sent from a tower will get to me easier than I can talk to it.. that's an ideal world situation, I don't anticipate it working out in such a way, but I'll give it a try tomorrow. It'll be yet another nice day for rooftop shenanigans.

The two directions I get the strongest signal pointing the panel are not at all in the direction of the tower.. RF is so fickle sometimes -_-

Do you have 5G phone? Another option is to drive to each tower and record what cells/bands your phone is connecting to at each site. Then you will be able to know which to go after and have a better idea of max speeds you would be able to obtain. I would try hard to find a tower within reach that is offering out 5G n41 signal. Thats where your gonna find best speeds with Tmo. Sounds like you have a handle on it though, just a lot of testing and trial and error really.

[token]

Thanks for the info. I tried it and.... I can no longer get 5g to connect, lol. To be fair, my sim came locked to a t9 so I was surprised 5g worked anyways.. I didn't really experiment too much because it's rare to have enough time where taking the internet down doesn't get me a stream of angry complaints from other residents :/

Oh well.. the weather turned again so i can't really go up there and play with it for now, so I'll see what happens in a few days and see if maybe I can get it to lock onto something yet again.

Thanks for your assistance thus far. I do believe I'm just a bit too far out and with too many trees for luck to be on my side.

Edit: yet another freeze warning so still waiting...

[token]

Well, today was decent enough to spend an hour on the roof. Things took an interesting turn. Frustrating and annoying, but different and promising.

If I disable B25, B12 and b71, it's now latching onto B2 or B66 as the primary and aggregating with n71 and B2/66.. RSRP around -120 and SNR of -5ish. Download speed approximately 15-20mbps though. It *will not* aggregate with n71/b71/B12 as primary. I'm thinking it believes the signal is too low... Which I suppose it kinda is.. but I really wish it would let me use the useful primary channel and give me the weak ones as additional.

I'm rather shocked that B2/B66 even try now, though.. I do believe the reset helped with that. As for 5g not reconnecting, I forgot to reapply "the magic" so that was on me...

My download speed is better with that horrible B2/B66 primary than it is with n71, and b71 just will not aggregate with those.

I don't understand why it's so resistant to aggregate. It's a very poor signal on B66/B2, but 20mbps is fine.. it's roughly what I'm already getting on my local tower, and I'm guessing with a competent b71/n71/b12 primary channel and B2/B66 secondaries, it'd beat it, have comparable upload speed and fix the latency/packet loss.. but it just will not let it happen.. the modem says it's capable, so is the signal just too weak and making the tower kick it off? Is the modem the one not allowing it? So close yet still so far...

I suppose I'm veering further out of antenna talk on this at this point, I'm happy to move it elsewhere if anyone thinks I should.

[mtl26637]

I've seen the same. From all of my trials, it will randomly connect to B71, B12 etc for the 4G primary, but it will latch on for a bit and get booted off as the tower does not like it for whatever reason. I'm fairly confident that its the tower that is the culprit. B2/n71, B66/n71 or B66/n41 are Tmo's goto combinations and anything else as 'primary' it doesn't seem to like. I don't think you will find anything stable other than those combinations for that particular tower. Only stronger clean signals will you see any additional aggregation such as B66/n71 + B71, but those are limited and I've only seen a select few combinations with an added 3rd 4G band.

If I was to take it any further my next step would be to find out if n41 is available from that tower. That is where you are going to find the fast speeds and quite honestly the stability. I have almost no interference on my n41 signal (SINR values ranging around 15-25). Even though n71 is a much lower frequency (600Mhz) my SINR values weren't near as good as n41. Plus the fact the the typical channel bandwidth's I've seen for n71=20Mhz and n41=100 to 200Mhz so you can see why n41 is so much faster. If that tower has n71 I bet its only a matter of minimal time before it gets n41 if not already. If no n41 then I'd focus on B66/n71 signal if you are sticking to that tower.

[token]

According to the T-Mobile map, the nearest n41 tower is 8 and a half miles to the northeast (random UC tower on its own by a truck stop, it stands out on the map), which isn't exactly ideal.. but certainly not the worst. The hill doesn't drop as quickly in that direction, it rides the ridge for about a mile and attenuates it too much.

The tower over there is absolutely *stacked* though. B2, B12, B66, b/n41, b/n71.. unfortunately even b/n71 is just a touch out of reach from that tower.. *everything* good is just a touch out of reach. That b41 tower to the south west, the n71/b66/B2 to the south east, the n71/b66/B2/B12/b41 to the north, the fat B2 straight west, a bunch of other ones along a fairly major state route straight east.. all in play with a clear shot over the trees, but alas, I can't get over them. Thick woods of massive 70-80ft trees.

If I could get over these giant trees, I'd have an embarrassment of choices since the terrain here is relatively favorable.. but the thick dense woods around me murders my chances of success. I was content with an acceptable n71 from that south-east tower, which was acceptable in the attic, but is slightly worse speed outside.

I will be moving the antenna/router setup. The "acceptance factor" threshold was breached with this antenna.. so I'm taking down the entire mast and going to mount the antenna on the house itself somewhere. I have enough cat5e to run the line wherever it needs to go, but I'll likely spend a day trying different spots to hopefully find a magical spot.

I thought the roof was gonna be great, but with the complaining and no actual tangible improvement to our internet, I can't justify it.. so it's coming back down (along with my hd8200u, which I meant to do last year to be honest) and I'll try different spots.. so the hope isn't dead.

I'll say this much.. related to the hd8200u, I bought "the best" antenna and put it up high for a specific tv market 55 miles away with multiple hills in the way. It failed to reliably pull it in. I spent 10 years experimenting trying to get that market, and you know what? I have all the OTAs from that market now. The expensive giant antenna on the roof was beaten by a homemade 8 bay bowtie *inside* of my barn sitting on the floor of the chicken coop. On both roofs in different spots? Unstable. Outside? Unstable signal. Inside, it's as marginal as I can possibly get, but it doesn't lose lock even on my weakest channel (on hdhomerun signal meter, decode threshold is 50.. I have 48-49 on one channel.. it stays there and decodes. Bizarre, but I'm not going to complain). I went around with a signal meter on my phone and went all around my property trying to find a spot. This is the spot. It's solid. Inside a building, on the ground.. makes no sense, but RF is weird.

So just because the spot I thought the LTE antenna would work best isn't working as expected doesn't mean anything. It just means I need to keep looking for a good spot to put it. I'll find the spot it wants to be. And knowing my luck, my local tower will get upgraded once I find the sweet spot for a different tower, but I'll be fine having a speedy backup when the power goes out.. which happens plenty enough to factor it in.

Sorry, I'm quite wordy -_-

[mtl26637]

I think you are on the right path, just a bunch of trial and error. If your after n71 you might try to use some of those old tv antennas since n71 is ~600Mhz and most of the LTE antennas weren't designed for that frequency range. I found a couple yagi tvws antenna that work extremely well for n71 so maybe you could repurpose the extras you have? That band would be the best band for making it through foilage too. The primary 4G band (typically B66 from what I've seen) is the tricky one since most of the large carriers use this frequency and can interfere with Tmo's slice of the band. Vzw has a slice setting at 2140Mhz which sets right next to Tmo's slice at 2150Mhz and when they come from the same general direction it can potentially cause some significant interference.

Have you tried locking down the 4G/5G bands? I typically try to avoid band locking if possible but sometimes if the signals are coming from towers in the same general direction and it keeps latching on to the wrong tower you can lock it down via the frequency and PCI to get to a specific tower.

[token]

I've done some locking. I was having speed issues on my local tower yesterday so I locked to another side.. unfortunately it didn't seem to help, but I've got the process down.

Generally I have everything but tmo/sprint bands disabled because I generally do a lot of experimenting and it's easier to put it back to 2:4:5:12:25:26:41:66:71 than it is to copy/pasta or memorize the long full list.. Usually disabling 25 is enough to get started in exploring the cells in my area.

Since the antenna will be getting moved around in the next few days, I'll certainly report back on anything interesting that happens with my experimentation.

And yeah, good solid n71 was the original goal.. and may still be my fallback, but getting it to latch onto a b66, b2, and a separate b41 in my very first attempt at placement has me intrigued.. and I'll likely try to see if I can take a real shot at these. Finding a place with a reasonable b66/B2/n71 would be fantastic.. or even, dare I dream, n41? Yeah, I don't have high hopes for that one, but 35mhz of bandwidth would still be better than the 5mhz I'm playing with now.. n71 SA is 10mhz, so even locking that one down could be a significant improvement.. I'd be content with either of these, but obviously salivate at the b66/n71/b2 or even n41 potential. 8 and a half miles.. just right on that threshold of "maybe", enough to mess with my head -_-

[mtl26637]

Yes, let us know how it pans out. Fyi, the ports on the RM50xQ-AE don't play nice when it comes to Tmo's bands either. This Quectel modem has the 4 ports. Ports 0, 1, 2 and 3. Ports 1 and 2 are the "low" band ports for frequencies of 600Mhz and up so those 2 ports are the only ports that you can pull in band 71. Ports 0 and 3 are for frequencies from 1400Mhz and up and would be your B66 for your 4G primary frequency. In my opinion this is why this modem doesn't do well with B12/n71 as band 12 is a "low" band which means the primary 4G signal would be locked in on ports 1 and 2. Since n71 is low band also then it would also have to come in on ports 1 and 2 which doesn't ultimately work out to well. So depending on what bands/frequencies your after its worth looking at the modems port mappings to know which antennas go to what ports as this will play a pretty big role in what you ultimately end up with.

[token]

Definitely not antenna related at this point but I wanted to note this since I brought it up earlier..

"extended temperature range" seems to be why I have so much latency on these further towers. On the quectels, 60c is considered "normal temperature range", 60-75 is considered "extended". From what I've been reading, at these higher temperatures, the modem goes into a state where it's trying to maintain a connection but throttling itself to avoid overheating.

I noticed as I was connecting to higher frequency/further towers that my qfe_wtr_pa0 shoots up into the 70c range almost instantly, and the others follow shortly afterwards.

When I reconnect to my local tower, the latency is still present until qfe_wtr_pa0 drops down under about 65c. I've tried this multiple times and it seems pretty consistent.

I've ordered a heat sink and small 40mm USB fan and will see how that affects things. It seems as though this is the bigger issue to my currently poor results (I did move the modem around a bit, never could get the latency consistently low on these distant high bands. Even when it seemed like I find a spot, as soon as I went to upload anything, it shot up and never came back down)

Yeah, a better signal would let the modem use lower power and not get as hot, but ample modem cooling is the easier and less temperamental solution.. so I'll report back on this.

The good news is TMobile snuck a b66/B2/B12/n71/b71 antenna up on my local tower. No b/n41 yet and the panel isn't live yet, but we're on the right track in multiple ways now. Finally...

Edit: last night the modem was at 26c. I connected to b66/n71/B2 combo and my latency was under 100ms. Temperature started going up little by little but latency was still low. Ran a speed test, temperature shot up, latency never recovered until I went back to my home tower. Definitely seeing a pattern here...