How to handle a LTE Connectivity loss

  • Dear Pycom-Community,

    yesterday I had an issue with my GPy. I send every 3 minutes Data via LTE-M. At some point, the GPy couldn't send data anymore, while it still told me, that my GPy is attached to the network and that a data connection is still there. Every time, before I send data, I check the connectivity:

    print("Is LTE attached:")
    print("Is LTE connected:")

    I've checked the site of the company, who provided the IoT SIM-Card and I could see, that my GPy was only attached to the network, but that there actually was no data connection.

    I guess a LTE connectivity loss happened, mentioned here:

    In the section under LTE connectivity loss, you can read the following:
    If the connectivity is lost this will in general not be reflected when you check lte.isconnected().

    Probably that happened for me. I contacted the SIM-Card provider though and they want to check, if something happened at there side. I will update the thread with their answer.

    I've waited like 40 minutes for the modem to get a data connection again, but it didn't happen. Resets via the Reset Pin or the Reset button didn't help with the issue at all. After 40 minutes I did a power cycle, completely cutting off the power and afterwards, the GPy got a connection again and could transmit data, which is kinda concerning.

    I thought about using lte.deinit() in this case and make a completely fresh start. Would this probably solve my issue? Any suggestion how to handle this situation to get a data connection going again?

    Are there any plans to update lte.isconnected() to spot a LTE Connectivity loss every time it happens to react to that problem? @Gijs

    Kind regards,


  • @Gjis
    Sorry for bothering you again. I just wanted to make sure, that an alternative also possibly works, which kjim suggested (Thanks again). Wouldn't lte.deinit also work in the case, that I suspect a missing data connection, which the GPy is not aware of?

    This page says the following:

    Disables LTE modem completely. This reduces the power consumption to the minimum. Call this before entering deepsleep.

    • detach : detach from network.

    • reset : reset LTE modem.

    LTE.deinit should still be useable, even if there is a data connection, right? That would make a power cycling obsolete and therefore a MOSFET, but it's always great to have other options at hand too. :)

    I would like to test it myself, but it is hard to reproduce this issue by force, because it depends on hicc ups in the network, which I cannot control. The constellation, where everything goes wrong is probably rare.

  • Global Moderator

    That could work. The high-side driver schematic indeed looks like something you'll need!

  • @Gijs said in How to handle a LTE Connectivity loss:

    I'd also use a small 100nF capacitor over the Gate-Drain to protect against switching transients

    I guess a resistor between the Arduino Pin and the MOSFET Gate would be also fine?

  • @robert-hh
    Originally, I've bought some adapters like that, because I've bought a SOT23 Watchdog. Lucky me :) .

    Thanks again everybody for the Input! It really helps!

  • @rcolistete said in How to handle a LTE Connectivity loss:


    Agreed. I have them in my boxes and soldered them. There are also SOT23 -> TO92 adapters available. Just a tiny PCB.

  • @rcolistete
    Nice, with this space it should be possible to solder them comfortably on some copper pads on my boards with a standard soldering station.

  • @robert-hh They are SMD but not too small, SOT23, with almost 2 mm of space between pins.

  • @rcolistete
    Thank you for your suggestion! Indeed, I'm looking for through hole p-channel mosfet.

  • @rcolistete These are indeed interesting parts, especially for low voltage applications. But @SciWax looked for a device in a through-hole package, not a SMD one. And it is really strange that devices like the one you mentioned seem available only in SMD packages.

  • What about the DMG3415U low power p-channel MOSFET ? Or the new one, DMP2045U. They cost about US$0.05/each if you buy 100-200 units.

  • @SciWax I just gave the IRFR5305 and the IRFD9024 a test. At 5V GS voltage the On resistance was 0.1 and 0.3 Ohm. At 3.3 Vgs it was 12 and 120 Ohm, so less good. I also do not know, why the P-Fets with the low Gate Threshold seem only available in SMD packages.

    You can also consider a moderate power PNP transistor like BD231, 2N5194 or BDX54 or the like. The have a higher Base current, but a low Emitter-Base threshold. The CE saturation voltage of 0.4 to 0.8 V does not matter in your case.

    P.S.: The IRFR5305 is a real power monster, given its to92 sized package, with 31A max current. Looking at the package, the connecting pins seem to be the limiting factor.

  • Thank you! I really appreciate it. I will have a look into the datasheets of the mosfets you've mentioned! And thanks for putting together the additional information!

  • @SciWax OK. Having had the same question, what do I have in my part boxes:
    IRFR5305: Gate Threshold 2-4V, RdsOn 0.065Ohm, small package
    IRFR9Z34N. Gate Threshol 2-4V, RdsOn 0.1 Ohm, TO220 Package
    IRFD9024: Gate source Thresold 2-4V, RonOn 0.1 Ohm, 4 Pin DIL Package
    Devices with higher RdsON:
    ZVP2106A: Gate Threshold 1.5-3.5V, RdsON 5Ohm, TO92 Package

  • Still looking for a logic level p-channel mosfet. :/

    I've looked on digikey for a through hole one (I've put for instance "irl p-channel through hole mosfet" in the search bar). I only found obsolete ones, like the IRL5602L or IRLIB9343. Can anybody else recommend me a logic level p-channel mosfet, I can control with a 5V Arduino to control the VIN-Pin on my GPy? I think I'm using the wrong keywords, because I have a hard time finding anything with "IRL p-channel".

    Sorry for bothering you, robert, but you are really knowledgeable about this stuff. Do you have an idea?

  • Global Moderator

    I can indeed support that I would recommend the usage of a P-channel MOSFET there (generally, you would use a P-channel to switch on the VCC side, and N-channel to switch on the GND side. Note that the gate voltage operates in opposite)
    Pull the gate up to VCC to stop the flow of current, and pull it down to GND to connect the two. I'd also use a small 100nF capacitor over the Gate-Drain to protect against switching transients

  • I'm looking for the right MOSFET right now. I still have a IRL 540N lying around, but after looking for MOSFETs in this forum here, I found alot of people, who mentioned P Channel MOSFETs, which I probably need, when I want to use the VIN Pin to power off the GPy. Any suggestions for the right logic-level MOSFET for the GPy I can control with my 5V Arduino? I should also probably put a resistor in front of the Gate to not hurt the Arduino Pin?

    Kind regards,


  • @Gijs Thanks for the suggestions at this late time.
    My Arduino counts the times a Code 200 wasn't received by the GPy. During a 3 minute intervall, the GPy has 9 chances for sending the latest 3 minute average and receiving the Code 200 in return. At this point, I think I will just do it like that: If the GPy fails 9 times, I can be probably sure, that something fishy is going on with the network. At this point, I will just power cycle the GPy via the MOSFET. After that powercycle, if lte.isconnected = True, the Arduino is allowed to send my measurement strings to the GPy again. I will just use another Pin for the state of lte.isconnected, for instance P11. The Arduino will read this Pin.

  • Global Moderator

    Right, you are correct. Though the last time I used uping, it concurred with a memory leak somewhere and have not tried it since. AT commands indeed do not work, though you could opt to suspend the ppp session (lte.pppsuspend()) and use the AT+PING command, after which you lte.pppresume(). On the other hand, that could result in additional complications. Another solution would be to use socket.getaddrinfo(), which would query the DNS server to resolve a hostname. A downside here, is that I've heard of some caching done with these requests, so you might want to be careful there either (I do not think I have experienced this firsthand).

    A last resort, would be to do a tcp request to a server and receive its response, that way, you know for sure there is a connection (if the socket does not error)

Log in to reply

Pycom on Twitter