Enable eDRX seems not work

[Tim Chao]

Hi,

I am trying to test eDRX function on 9160DK with modem_shell.

The command returns ok.

Modem FW version:      mfw_nrf9160_1.3.1
Device ID:             nrf-351358811475208
Operator full name:   ""
Operator short name:  ""
Operator PLMN:        "46692"
Current cell id:       28846103 (0x01B82817)
Current phy cell id:   214
Current band:          3
Current TAC:           12700 (0x319C)
Current rsrp:          27: -114dBm
Current snr:           23: -1dB
Mobile network time and date: 22/07/14,06:06:48+32
PDP context info 1:
  CID:                0
  PDN ID:             0
  PDP context active: yes
  PDP type:           IPV4V6
  APN:                internet.iot
  IPv4 MTU:           1500
  IPv6 MTU:           1500
  IPv4 address:       10.173.94.11
  IPv6 address:       2001:b400:e232:2d7c:7625:b63d:bce6:8597
  IPv4 DNS address:   168.95.1.1, 168.95.192.1
  IPv6 DNS address:   2001:b000:168::1, 2001:b000:168::2
RRC mode: Idle
LTE cell changed: ID: 28662797 (0x01B55C0D), PCI 71, Tracking area: 12700 (0x319C), Band: 3, RSRP: 21 (-120dBm), SNR: 15 (-9dB)
mosh:~$ link edrx -x 0010 -e -m
eDRX parameter update: eDRX: 163.839996, PTW: 2.560000
eDRX enabled
RRC mode: Connected
eDRX parameter update: eDRX: 20.480000, PTW: 2.560000
RRC mode: Idle
LTE cell changed: ID: 28846103 (0x01B82817), PCI 214, Tracking area: 12700 (0x319C), Band: 3, RSRP: 26 (-115dBm), SNR: 23 (-1dB)

But when I check the PPK output, it seems that the modem stay in DRX state and never enter sleep.

And here is the trace log.

trace-2022-07-14T06-06-46.613Z.bin

Please help to check what I did wrong.

Thanks.

[Achim Kraus]

Peaks of 50mA every 2.56s may also be caused by sticking in RRC Connected.

The UICC requires in my experience about 20-30µA in idle, so far away from that 50mA.

Though the question is 9 month old, I'm not sure, if you want to invest more time for investigations.

If you're interested:

How long a devices stay in RRC Connected is defined by the network operator and may be shorted by using RAI. Without RAI you need to wait for the "RRC Active Timer" until the device switches to RRC Idle and you eDRX settings get active.

For details, see GSMA Energy Efficiency for Mobile IoT or Nordic - LTE-M vs NB-IoT Field Test: How Distance Affects Power Consumption.

Currently some network operators are rolling out Rel 14 LTE-M and with that you have a chance of using AS RAI with LTE-M.

[Tim Chao]

I'm sure the latest LTE_LC_EVT_RRC_UPDATE event I saw is RRC Idle, but not sure the reported state is right or not...

As I know, the RAI is not suitable for TCP connection (please correct me if wrong). Because I am using MQTT protocol for data transmit (some customers even want HTTP...), I guess RAI is not the solution for me.

BTW, dose current "RRC Active Timer" is readable?

[Achim Kraus]

PSM, eDRX and RAI are based in "network data messages", at least as far as I know.

TCP uses more "network data messages" than just a 1:1 of your application data, because e.g. a TCP inject TCP-ACK requires a "network data messages" as well. The best way to see what happens is to use a trace and check the IP-message. You will see then "near by" the resulting "network data messages".

Anyway, if your device reports "RRC Idle", then it may have taken a little longer, but it reaches the RRC idle state. What I see in the trace:

2.3s XMONITOR reports LTE-M and eDRX "1001" => 163s

20s RRC Connection Release

22.6s XMONITOR reports still LTE-M and eDRX "1001" => 163s

55s RRC Connection Request

55s Request eDRX for LTE-M to "0010" => 20s

57s Modem reports eDRX for LTE-M updated the value reported by the network to "0010" => 20s (as requested).

55s RRC Connection Released

no new connection for the next 60s.

How are the times in the trace related with the times in the power chart? (I use some additional GPIO to related the times in the PPK II (digital input there) with the modem.)

So, I'm not sure, what we actually see in the power chart. I would guess, eDRX is negotiated and working, the modem reaches idle, but something keeps the modem from entering eDRX. That requires then a analyze of the trace by Nordic's modem experts to see what happens in that time. But I guess, it will be clearer if the times of the traces can be related to the times of the charts.

[Tim Chao]

Looks like Nodic fixed something in mfw.

I try to reproduce this issue with my NRF9160DK, the issue still exists.

But after I upgrade the mfw from 1.3.1 -> 1.3.4, it report "+CEDRXP: 0" (disabled) when I perform "link edrx" command.

So, I try do downgrade the mfw back to 1.3.1 again, it reports ok and cannot enter sleep mode.

[00:00:17.639,770] <dbg> lte_lc: at_handler_cedrxp: +CEDRXP notification, +CEDRXP: 4,"0010","0000","0001

[Achim Kraus]

I started my eDRX experiments with the 1.3.3.  I guess, with that version the reported network timers are the right ones. If eDRX is supported or not seems to be also depending on the used SIM card. With one I don't get eDRX, but with the other one I get it, same network provider, same base station. If the device operates in a network, which supports eDRX, the device doesn't report entering the sleeping mode. But it reduces the power when entering the RRC idle mode.

That chart is from a device operating in a network supporting LTE-M, eDRX, but no AS-RAI. In the quite phase (RRC idle) it consumes every 20s about 2.5mC (peaks up to 60mA). In the time between this peaks you see the quiescent current of the device including the SIM. In my case, this is about 25µA higher then in power sleeping mode, what is caused by that SIM card. So, if you add all together, eDRX takes 25µA (90mC per hour) and 2.5mC per receive peak (with 20s that's about 450mC), it's 540mC more than PSM. Maybe once softsim is more general available, then it only depends on the interval.