nrf9160 e-i-DRX nb-iot much higher power consumption than expected

[papatel]

Using modem firmware versions 1.3.5 and 1.3.2 we see exceptionally high average current consumption when measuring eDRX cycles on Jio's nb-iot network in India (Band 5).  We are using the PPK2 and the 9160DK version 1_1_0.  We've tried using the PPK as an ammeter and a source meter with the same results.

The network confirms our eDRX interval of 20.48s and PTW of 2.56s

We believe our power measurement setup is good because our floor current between eDRX cycles is ~40uA.  This is consistent with documentation and our SIM still being on.

According to the power calculator, we should see ~168uA average current.  Instead we see ~584uA, see plot below.  We've also included a closeup of the eDRX profile.

Supply Voltage is 3.7V.  We've taken many other measurements and our setup is always ~400uA higher average current for DRX and eDRX cycling.  We really need an explanation quick or our high volume product is dead in the water.

One possible explanation is iDRX PDCCH repetitions.  I've left those at the default setting of 16 on the calculator.  When we query the radio using AT+CEINFO? the reptition count is set to 1 but we have no idea if that command works as expected.

Thank you!

[helsing]

Hi Patrick, sorry for the delay.

One possible explanation is iDRX PDCCH repetitions.

That might be possible, although the duration is quite long.

Would you be able to record and provide us with a modem trace?

[papatel]

Hello Helsing,

We've moved onto other testing for now but will try to circle back soon with a modem trace.  If possible, I'd really like the Nordic engineering team to confirm NB-IoT eDRX energy consumption for common settings against the calculator with the latest firmware.  If this has already been done, please share what information you can.  Our goal here to make sure we are achieving expected eDRX power consumption.

Other module manufacturers often put a power consumption number for a specific eDRX configuration and we've been able to verify against the published numbers.  We need to do the same with the 9160.  These modules don't achieve what the Nordic calculator claims but they do achieve much less than our measured 9160 consumption given the same eDRX configuration on the same network with the same exact SIM card.

There are a lot of factors at play here regarding our module selection and yes, power consumption is important.  A few uA here and there won't make or break our design but 400uA certainly will!

I don't see anything special about Jio's network in India.  I suspect the results are easily replicated on any NB-IoT network that supports the 20s interval and the 2.5s PTW...

Thank you!

[helsing]

papatel said:

We've moved onto other testing for now but will try to circle back soon with a modem trace.

When ready, the modem trace will help us understand the current consumption better. I am in any case discussing the issue with our modem engineers and will report back any findings.

papatel said:

If possible, I'd really like the Nordic engineering team to confirm NB-IoT eDRX energy consumption for common settings against the calculator with the latest firmware.

I have requested some updated measurements to confirm that the levels displayed in OPP are still achievable.

[helsing]

Hi Patrick, sorry for the delay.

Around 200uA is expected. The cause for the high current consumption could be that there seem to be more active time than expected. In order to improve the current consumption one should seek an explanation for the active time. (A modem trace would help).

papatel said:

If possible, I'd really like the Nordic engineering team to confirm NB-IoT eDRX energy consumption for common settings against the calculator with the latest firmware.

In our internal testing a similar scenario show about 173uA average, with 2.56s PTW, (327s eDRX with 2.56PTW) but there is only paging monitoring and virtual SIM used so the floor current is around 5uA. The floor current being x8 higher in your case translates to +15% current consumption or +30uA. With a

20.48s eDRX period our modem team calculated that the described scenario would give 200uA average current consumption if a real SIM was used.

If the close-up is from the PTW window, it looks like the UE might receive some data from some paging occasion and performs some more RX e.g. paging data. It is difficult to tell what is going on so a modem trace is necessary to gain more understanding.

Below is how 2.56s PTW looks like in our tests against a CMW environment. The image is quite different. Having 168uA average over eDRX cycle is optimistic but possible with optimal channel conditions and a good hardware sample.

The distance between markers is 1.489s, Average: 1.263mA.

The image above is captured when running an unreleased MFW, however, the current consumption using MFW 1.3.5 should be in the same order of magnitude.

Here is another measurement, not for 20.48s eDRX but for 10.24s eDRX. The test should not differ from a test case using 20.48s when looking only at the PTW part. The current consumption includes the SIM current.

In this case there is nothing to be received and no neighbor channels to be measured. The PO event above takes 1.38s in total. This also differs from your image and points towards something being received in your case. There seem to be still some activity/payload to be received from network and therefore you are not seeing the expected figures.