Weird LTE-M PSM behaviour

[ante]

Hi,

I am trying to get PSM to work using a nRF9160-DK running Assets Tracker (nrf tag v1.2.0) with iBasis connecting to Telia Sonera in Sweden. I have added/modified this config:

CONFIG_POWER_OPTIMIZATION_ENABLE=y
CONFIG_LTE_PSM_REQ_RPTAU="10100110"
CONFIG_LTE_LOCK_PLMN=y
CONFIG_LTE_LOCK_PLMN_STRING="24001"
CONFIG_BSD_LIBRARY_TRACE_ENABLED=y

I.e. PSM is enabled, requested TAU period=6 min if I am not mistaking. On the board I have set Switch 2 to GND thus activating PSM.

When I start the application with this setup it seems that the device actually enters PSM mode as I can see the periodic ~1.2 sec spikes disappears. Expecting a period of 6 min it instead looks like there is no communication until ~20 min:

If I check the negotiated values using %XMONITOR I get:

%XMONITOR: 5,"Telia","Telia","24001","0027",7,20,"0188D616",,,,,"","00000000","11100000"

So my questions are:

1. Since Periodic-TAU value is "11100000", it seems it is disabled even though I see a big difference in current consumption when flipping Switch 2 to GND. Could you please explain whats happening here?
2. Normally active-time value is set to "11100000" (disabled) when PSM is deactivated but what does "00000000" mean?
3. Do you have any idea of what happens after the 20 min? If anything I would expect one short spike, instead it looks like several LTE registration attempts.

I have tried setting up PSM using AT commands manually, but it behaves the same way. It is very hard to figure out what is happening when there is very little feedback from the device. I have attached a modem trace log that I guess contains some more details. Are there any chance of get hand of a tool that can parse the log to get some additional feedback of what's happening between modem and base station?

trace-2020-04-23T09-19-54.093Z.bin

[iBasis Tech Support]

Hi,

As you mentioned, this result means that PSM mode is not enabled. The recommendation will be to increase the timers because each MNO has different timers configured. Usually, PSM works with timer higher than 4 hours, so please use this values:

AT+CPSMS=1,"","","00111000","01001000"

 

Also, to verify PSM, execute these commands:

 

AT+CEREG=5

AT+CEREG?

 

1

%XMONITOR: 5,"Telia","Telia","24001","0027",7,20,"0188D616",,,,,"","00000000","11100000"

 

Kind regards,

Iuliana

[ante]

Hi again Luliana!

yes I have read that different operators have different minimum PSM periods, but I want to get a better understanding of what is happening since, to me, I think the nRF91 is acting unexpectedly. If I try to activate PSM with an operator that does not support it this result in %XMONITOR and +CEREG returning a periodic TAU and active time values "11100000", and the devices then continues communicating like before. This sounds reasonable to me. When I try a PSM that, I guess is too low, I would expect that the device either would enter PSM and return the negotiated PSM values with CEREG/XMONITOR or that it would not enter PSM and return "11100000" for both timers and continue communicating like before - instead I get the weird behavior I mention above.

To test the suggested values you sent me I switched to the ATClient sample application and used manual AT-commands. After executing:

AT+CPSMS=1,"","","00111000","01001000"

I read back the values:

AT+CEREG?
+CEREG: 5,5,"0027","018A0116",7,,,"01001000","00111001"
OK

AT%XMONITOR
%XMONITOR: 5,"Telia","Telia","24001","0027",7,20,"018A0116",211,6300,59,28,"","01001000","00111001"
OK

As far as I have read on the forum here this would suggest that PSM is now active. However, by measuring the current it very much looks like it is not the case. This is the current consumption from the time I send AT+CPSMS:

So is there a way to get more detailed information on what is happening between base station and modem? Maybe I am missing something, but as far as I see it, only looking at CEREG/XMONITOR is not a reliable way of checking if PSM is active/working.

Kind regards

/Ante

[ante]

Ok, so I did miss something. The active time was set to 48 min, if I am not mistaking, and I didn't do power measurements for that long. By decreasing the active time significantly I could verify that the nRF did actually enter PSM mode.