Using nrf9160 for cellular connections in cars

[jonny_v]

Hi!

We are testing nRF9160 in an application were we monitor sensor in certain modules on trucks / trailers. Our test application is based on the Serial LTE sample application, using modem FW 1.3.1.

When testing on our lab everything works fine, but as soon as we test "on the road" we tend to get disconnected / very poor reception - effectively stopping our logging to be logged to cloud. Our theory is that the nRF9160 use too long time to change to closer base stations as we drive.

Do you have any hints or tips to what we can do to improve performance in such an application? Any helpful commands/settings/literature for example.

Note:

We use LTE-M1 mode, as that seems to be the recommended for mobile applications in general. Though, I notice that reported signal strength is typically better in the nRF9160 with NB-IoT.

[Albrecht Markus Schellenberger]

Hello Andreas,

We are testing nRF9160 in an application were we monitor sensor in certain modules on trucks / trailers.

Where in the truck/trailer is your board placed?

When testing on our lab everything works fine, but as soon as we test "on the road" we tend to get disconnected / very poor reception - effectively stopping our logging to be logged to cloud. Our theory is that the nRF9160 use too long time to change to closer base stations as we drive.

Might there be any obstacles between the antenna and the sky, that might disturb and thus affect signal quality?

Do you have any hints or tips to what we can do to improve performance in such an application? Any helpful commands/settings/literature for example.

Have you done reference testing with e.g. a nRF9160 DK?

Regards,

Markus

[jonny_v]

We have tested with two types on antennas with ~1m cable to the board.

* Whip-type antenna placed on top of the car "roof"

* Foil antenna fixed to windshield with adhesive

Both worked more or less equally well. When we aren't moving the signal is OK, typically around -90 dBm. So there shouldn't be a problem with obstacles etc.

Our version of the dev.kit is too old to run the same FW version as we are currently using (we upgraded FW in our design because it should work better with respect to detecting and switching of cells). But that was a good suggestion, I'll try setting up a test with the DK.