nRF9160 DK sleep current too high after opening and closing a TCP socket

[Thomas]

Hi,

Similar to https://devzone.nordicsemi.com/f/nordic-q-a/45544/nrf9160-dk-psm-current-too-high-after-opening-and-closing-a-tcp-socket/260026#260026, however they said that issue has already been fixed and to make a new post.

After opening, connecting and closing a TCP socket and powering down the modem (either AT+CFUN=4 or  AT+CFUN=0), the device wakes every 1.2 seconds which raises the sleep current by an average of 20uA...

Below is a capture of the device sleeping (~6uA), then registering, opening a socket to Google, closing the socket and powering off, then waking every second. I found that it stops waking every second after a few minutes?

I'm using version 1.2 of the modem. I've also included a simple sample as well:

1234.tcpOpenClose.zip

Note: The sample use RTT as the debug interface not the UART.

[Sigurd]

Hi,

Thanks for reporting this. I will check the behavior with the Nordic modem firmware team.

[Thomas]

Sigurd?

[Thomas]

Thanks for the update Hakon

[Sigurd]

Hi Thomas,

The nRF9160 closes the TCP connection but server doesn't send TCP FIN. The modem therefore keeps polling incoming data.
The TCP server should have finalized closing of TCP connection by sending TCP FIN.
We will look into ways to optimize this in the modem in future modem fw releases.

[Thomas]

Thank you Sigurd,

Surely if the modem is no longer registered or powered off it should stop?

Kind regards,

Thomas

[Thomas]

Could this not lead to socket issues? There are only a limited number of sockets and if the modem does not free them up properly they will quickly run out? 

[Sigurd]

Hi Thomas,

Thomas said:

Surely if the modem is no longer registered or powered off it should stop?

Yes, I believe so. The modem team is studying how to improve the power efficiency in TCP, but it may take some time. At the moment the problem is that some TCP timers are left running even though they should be stopped.

Thomas said:

Could this not lead to socket issues? There are only a limited number of sockets and if the modem does not free them up properly they will quickly run out?

No, socket is closed correctly, it is only TCP protocol inside modem that keeps waking up because of missing TCP FIN from server side.

[Thomas]

Okay that is good to know, Thanks Sigurd.

Do you have any indication when a fix might be released?

[Sigurd]

Thomas said:

Do you have any indication when a fix might be released?

No timeline for this yet. Maybe in a couple of months.

[Thomas]

Hi Sigurd, 

I see this has not been fixed in version 1.2.1 of the the modem firmware. Any further update on this? Will it be in 1.2.2?

Kind regards,

Thomas