Noob needs help setting MVNO's APN for RM520N-GLAA to override SIM's APN

Many Quetel modem users experience no problems whatsoever. These days MBN usually takes care of setting the APN(s) to usable values.

Sometimes there are provider issues, particularly in the US. While I’m aware of many of these, I have no direct experience. I live in Australia.

Power supply is a fairly common point of failure with today’s powerful modems. It manifests itself in poorer signal conditions, where the modem has to ramp up its transmit signal, loading up the power amplifier.

1 Like

Agreed, snowgum.

Also, improving the antennae should mean the radio could work less hard and draw less power.

I got the new TE Connectivity antennae today and promptly installed them.

They do make a huge difference in stability, though the stability is not 100% perfect yet.

I am now able to very slowly complete downloads of larger OS or application updates, unlike before, when such large downloads would stall early on in the download and then time out. I can also do more downloads in parallel. Videos and audio streams no longer get stuck. It would fool me into thinking it’s a perfectly stable connection except that I see occasional short disconnects when watching the router’s status or running a continuous ping.

Latency/jitter are much better. I need those for gaming and video calls/screen sharing. Big win in that regard.

I still can never get better than 21Mbps, a problem with the MCUZone MR5210 mentioned or hinted at by others on this forum and on the OpenWrt forum. The real test of top speed will be taking the entire setup and driving to where there are a clear 5 bars of 5G signal–while my service is prioritized–then repeatedly measuring top speed.

I think switching from the USB power supply (5.0V at 1.5A = 7.5W) to the DC barrel jack (12V at 1A = 12W) made the biggest difference so far in both reliability and speed.

Next up is receiving the 12V at 3.0A = 36W power supply.

I emailed MCUZone to ask about how much power is actually provided to the modem module at 3.7V. They responded that they do provide the 3.0A to the modem module, but I’m still suspicious about it.

Chances are high that the power supply replacement will be the last change I make with this MCUZone box. If it is still not reliable or fast, I’ll start looking for a replacement, though the current alternatives are slim for M.2 to Ethernet/USB boards.

It’s been about 48 hours of running with the new antennae, and I have been watching the stats throughout the day.

The connection failed overnight and did not self-recover. I had to power cycle the MCUZone box to get it connected again.

The new antennae have helped somewhat but not enough in terms of these stats:

RSRP is Poor or Fair with new antennae, up from always Poor with old antennae.
RSSI has always been Excellent with both old and new antennae, but from the unreliable and slow real-world connection on both of my smartphones AND the Quectel RM520N-GLAA, I don’t think RSSI means much. I have always had poor-quality phone calls or data in this neighborhood.
RSRQ is usually Poor, sometimes Fair, very rarely Good with the new antennae. With the old antennae it is almost always Poor, very rarely Fair. This is, as I understand it, a measure of signal interference.
SINR is Good or Excellent with the new antennae and Fair to Good with the old antennae.

So, all of the specs did clearly improve with the new antennae, but the two that probably matter most, RSRP and RSRQ, were so bad that even though they improved, they are still not in the Good range often enough to help.

I also noticed that every day from midmorning till about 7pm seems to see slow throughput and more frequent disconnects. I suspect that call density or interference is at fault, because 7pm till midnight the performance is notably better. I also never seem to be able to get 5G during the afternoons, but by late evening 5G is available.

Oh, and the 20Mbps cap is probably due to the poor signal, since it clearly says either 10M, 15M, or 20M in the ul_bandwidth/dl_bandwidth stats just above the signal quality stats. Given that these stats come from the modem itself, it is likely not the MCUZone box capping the throughput.

Insufficient power is still an issue until I get the 12V 3.0A power supply; I suspect the afternoon disconnects are due to the radio working hard to deal with interference, then running out of power. Not sure why it cut off last night for the whole night, though.

I am stuck waiting on the new power supply to arrive and probably won’t have any different results than now until then.

Forgot to mention that it has become clear that some of the error messages such as “SIM not registered” and such are not due to the modem but the Gl-Inet travel router’s firmware. I ran the AT commands you recommended and other ones from the manual, and the modem was much more clear about its status than the Gl-Inet router’s web interface.

While it was true that if I had problems with the connection, there would be errors in the Gl-Inet web interface, there were also times when there would be errors in the Gl-Inet web interface while the modem was successfully connected and passing data. Gl-Inet probably needs to improve the way in which they query the modem in a future firmware release.

That is amount of spectrum in mhz, not Mbps.

1 Like

Thanks, rooted!

Today I learned something new! I guess this means the possibility that the MCUZone MR5210P is capping speeds is back on the table.

It’s still probably the low signal quality first, given that all my smartphones also get poor signal and slow and unreliable data speeds. I’ll get the 3.0A power supply this Monday. Hopefully that helps. If not, I will try driving to where signal is much better to test. If that goes badly, it probably is the MCUZone box capping the speed. If that goes well, I’ll buy a 5G signal booster for my home. The signal booster would have the additional benefit of improving the smartphones’ call reliability.

This is the USB enclosure I use for all my M.2 modems:

I would have posted it earlier, but that web site was down.

I have around 10 different modems in them, including 5G units from Quectel and SIMCom.

Thanks, snowgum!

I hadn’t seen that one before (was only looking at those with Ethernet).
It does say it outputs 3.8V at 3.0A to the M.2 device–that is the minimum continuous current needed according to Quectel. Sure would be nice if it output 3.8V at 5.0A like the RM520N-GL’s Hardware Design Guide says, but at least it meets the minimum, which should mean that it won’t cut out (exactly what it says further up on the page: “Prevent voltage drop when high data transfer .”

This should work well and is priced right.

The new, overkill 36W power supply arrived today. I got a DVE DSA-36PFH-12 that is commonly sold with name-brand wifi routers.

We shall see whether stability and/or speed improves. The outages last Friday and Saturday may have been related to two back-to-back solar storms that struck the earth on Thursday, Aug 1, lasting till the end of this week, Aug 10 or 11.

So far so good with the overkill 36W power supply. It feels more stable, though I can’t really prove it.
Performance is still poor, but now, I can watch the travel router report the stats pulled from the modem, and it shows that the 5G signal is awful and that the modem is very frequently falling out of 5G (on band 71) into 4G LTE (on band 2). Previously, the modem would lose the connection so badly that it would take a while to connect to any band and show any stats. This is probably the benefit the bigger power supply offers: the ability to transition in and out of 5G without totally losing the connection.

At this point, I can’t blame the MCUZone box for the slow performance because the 5G signal is just so poor. Next test would be to drive to where the 5G signal is known to be good, but I realized that I’d need access to AC power, which is pretty rare in coffee shops (and other places where you can sit down and set up a laptop) nowadays.

I’ll just assume the MCUZone box is not the limiter of top speed and focus on buying a signal booster with internal and external antennae next. It would have the additional benefit of boosting the poor signal for all our smartphones too.

Thanks to snowgum and rooted! Although my setup is not perfect yet, it’s clear what the next steps to take are: get a good 5G signal, then maybe buy a different M.2 board/box.

Follow up after about a week: only a few losses of service, but these were short, and they seem to be related to network congestion at peak usage times.

I also saw the unit hit 81Mbps overnight when downloading Steam game updates (it records the peak), so the MCUZone unit can hit that high a download throughput at least.

I have two outstanding questions:

a) Does transitioning out of 5G into 4G always cause a temporary pause in data transfer? It looks like it does, but I’m curious about the theory/design behind it.

b) Does poor signal increase the likelihood of experiencing degraded service in high congestion?
Put another way, if the provider sees two devices and has to choose between prioritizing one over the other, does the signal quality of each device weigh into the decision, and does the device with poorer signal quality get deprioritized? Put in the most practical terms, would buying a signal booster help with the apparent congestion problem? I expect no, though some signal booster marketing language says yes.

Thanks in advance for any and all help!

Those are interesting questions, best answered by a network engineer.

On the topic of priority, some cells seen by the modem will likely have a higher priority than others (in terms of whether they should be selected).

These priorities are often visible in the AT+QENG="neighbourcell" results:

AT+QENG=“neighbourcell”
+QENG: “neighbourcell intra”,“LTE”,100,69,-12,-116,-82,0,23,6,30,22,255
+QENG: “neighbourcell intra”,“LTE”,100,47,-10,-114,-95,0,26,6,30,22,255
+QENG: “neighbourcell intra”,“LTE”,100,46,-19,-123,-95,0,16,6,30,22,255
+QENG: “neighbourcell inter”,“LTE”,3148,189,-12,-120,-100,0,23,4,22,26
+QENG: “neighbourcell inter”,“LTE”,3148,496,-18,-127,-100,0,23,4,22,26
+QENG: “neighbourcell inter”,“LTE”,1250,411,-11,-111,-88,0,26,5,18,20
+QENG: “neighbourcell inter”,“LTE”,1250,83,-16,-113,-88,0,26,5,18,20
+QENG: “neighbourcell inter”,“LTE”,1250,298,-20,-118,-88,0,26,5,18,20
+QENG: “neighbourcell inter”,“LTE”,1250,167,-18,-115,-87,0,26,5,18,20
+QENG: “neighbourcell inter”,“LTE”,9410,202,-13,-95,-73,0,16,5,16,20
+QENG: “neighbourcell inter”,“LTE”,9410,136,-17,-99,-73,0,16,5,16,20
OK

A larger number (highlighted) is a higher priority.

EARFCN 100 is on B1, 3148 is on B7, 1250 is on B3 and 9410 is on B28.

Signal strength and quality are also major factors in selecting cells.

1 Like

Thanks, snowgum!

Seems I have to do some reading first before I can understand your post (that’s a good thing). I’ll look into my neighbourcell output after I learn enough about what it means.

I did a small experiment with my iPhone with only 4G LTE capability.

I walked about a half mile to the nearest McDonald’s, running internet speed tests along the way.

McDs: 3/4 bars, 31.86 down, 2.59 up

park: 4/4 bars, 58.37 down, 25.95 up <— benefit of being outdoors and away from other signals’ interference (nobody else was around with their cell phones).

home: 1/4 bars, 5.46 down, 0.28 up

Presumably these were in the same cell, though it is possible that I crossed from one cell to the next.

Ironically, McDs is in the direction of the strongest antenna tower and should have the best signal by far, but being indoors and next to other patrons with their cell phones in their pockets seems to have made a difference.

Given this info, my buying and installing a signal booster that has an external antenna should solve my poor data speeds.

Everything was going great until last night, when the modem started acting up such that my travel router (GL-inet Slate AX) couldn’t properly talk with the RM520N-GL somehow. It was able to pull the signal stats, which looked okay, but it couldn’t tether to the modem over USB. Normally, the travel router sees the MCUZone box+Quectel modem as an Android device in USB-tethering mode, but it was not seeing it as a tetherable device (even though I could send AT commands and it could pull signal stats).

Took a few soft then hard restarts of both the modem and the travel router, and it’s still not stable.

I’m pretty sure power is no longer any kind of issue (unless the MCUZone box doesn’t supply enough power to the Quectel modem even though i have a 36W power supply). In this case, although signal is and always will be poor in my home, I don’t think the poor signal was the cause either, because it looked more like the modem was malfunctioning.

Another possibility is that the carrier, US Mobile, is giving me trouble. I doubt this, but I will test it by moving their SIM to my smartphone.

Looks like it’s time to get a different box than my MCUZone. Rework.network has sold out of their USB+Ethernet box but still has the V7 USB-only box. I will probably buy the more-expensive Gl-Inet GL-M2 even though it’s not got the best features just to get the single-vendor firmware support on the travel router so that it issues the correct AT commands to the modem.

I am now testing a cellular signal booster (a SureCall Fusion2Go meant for cars) at home. It boosts bands 2 and 66 but not 71 (there are no consumer-grade band 71 boosters yet).

I chose the smaller Fusion2Go over the more powerful Fusion4Home because I didn’t want to have to run cables to make sure there is over 50ft of separation between the indoor and outdoor antennae.

Anyway, so far, in the midst of a supremely-frustrating three days of data drops every few minutes, the signal booster seems to be working to improve stability. It’s too soon to say for sure, but I have enjoyed an hour of no data drops since setting up the unit to test. I put the interior antenna very near the MCUZone box.

When I move my iPhone towards the interior antenna, I can see the signal go from 1 bar up to 4 bars (if held 1 inch away from the antenna). It seems to be doing its job of boosting the signal without triggering a feedback loop.

I’ll report back after a week or so about stability of the data connection. Then perhaps I might try running without the signal booster to see whether it really made the difference.

The signal booster does help but there are still reliability issues. I still see the MCUZone+Quectel lose its connection and seemingly restart on its own. At this time, these are the primary suspects:

a) MCUZone box may be too unreliable; buy a rework box or GL-M2 dev board

b) could be some problem with the US Mobile MVNO and how it handles data. I started doing research on this and did run across complaints about poor reliability with their data, with someone saying their APN is less reliable than T-Mobile’s APN. I can test this by A|B testing my Ting SIM (Ting is also a T-Mobile MVNO) versus my US Mobile SIM–this should diagnostically isolate US Mobile as a potential cause. I am also going to try US Mobile’s Dark Star (AT&T MVNO) since this is a three-month trial of US Mobile as a whole.

c) There is always the possibility that my Quectel RM520N-GLAA modem itself is damaged somehow. It’s a long shot and further down the diagnostic tree, but I may someday buy a new modem to A|B test that too.

Follow up.
I learned something interesting: the signal booster can cause more problems. If the MCUZone box with its high-gain antennae is placed too close to the signal booster, I get far more crashes of the RM520N-GLAA modem.

It’s probably due to distortion or a feedback loop. My iPhone does better with the signal booster around, and it does even better when downloading data if I move the iPhone closer to the signal booster. I suspect this different behavior is because smartphones probably don’t have high-gain antennae inside them, but I don’t really know.

Anyway, I could simply move the MCUZone box further away from the signal booster, but then I lose its benefit. I turned off the signal booster and went back to the typical crashing-every-few-minutes behavior.

rework.network’s M.2 box came back in stock, and I ordered one right away. I will report back after replacing the MCUZone box, transferring the Quectel RM520N-GLAA into the rework.network box. If that whole setup doesn’t work reliably, the next step would be to buy a new Quectel modem.

Probably the final follow-up on this thread (I hope this helps someone else).

I got the rework.network M.2 to Ethernet box today and transferred the Quectel RM520N-GLAA over to it. Right away, it was clear the internet connection was more stable.

Poor signal is still an issue here, but I have finally knocked down the biggest problem: the unreliable MCUZone MR5210P box.

Not so fast! :stuck_out_tongue:

With those statements in conjunction, at least me you got successfully confused. Didn’t you intend to refrain from purchasing that module due to its assumed too low power delivery specification? At that point however, something doesn’t quite add up as it doesn’t make any sense that especially this board shall be known for its stability if the power supply is insufficient at the same time. So either the power demand of the Quectel modems isn’t that high or that rework.network board (you’re probably referring to the repeatedly sold-out “Dual-Q 5G2PHY” there) delivers more power after all.