nRF9160 increased power usage after calling getaddrinfo

[DeltaXP]

Setup:

• Custom board limited to only necessary components
• mfw_nrf9160_1.1.1
• fw-nrfconnect-nrf revision: v1.2.0
• Verizon network/sim
• PSM mode
• UDP client
• Keithley DMM7510 7 1/2 digit multimeter for current measurement

I have a simple test setup with nRF9160 connected to Verizon network, successfully enabling PSM, and programmed to periodically send a UDP packet. I am trying to measure the low power capability of the nRF9160. With a static IP configured for remote address I am able to achieve an Idle/PSM connected state of about 4.6uA. If I perform a DNS query with getaddrinfo, the idel/psm connected current increases to about 36uA. The following graph (x-axis: sample cnt, y-axis: mA) captures 1 sec of the idle current in both configurations, static ip (blue) and getaddrinfo (orange):

As you can see there is a larger current spike that must be periodic that causes the increase in current consumption. I am not sure where this is coming from as I haven't traced the implementation of getaddrinfo, it could be in application/zephyr stack, bsdlib, or modem.

Overall, I am very happy with be able to achieve <5uA idle state PSM connected! Any insight into the DNS query issue would be helpful.

[DeltaXP]

I captured 10 seconds of measured idle current in both configurations, static ip (blue) and getaddrinfo (orange):

The large current spikes are definitely periodic about every (1.25 sec) and account for the increase power consumption. I also ran Segger Ozone with J-Trace to see if there are any instructions running on the application processor. The only instruction load was from the Zephyr kernel idle thread that executed periodically about every 20 seconds with only a few instructions. This seems to indicate that this issue completely resides within the modem firmware.

Would it be helpful to run a modem trace?

[Simon]

DeltaXP said:

Would it be helpful to run a modem trace?

 A modem trace would be nice

Best regards,

Simon