nRF9160DK v0.9.0 GPS Issue

[gscultho]

I am having issues running GPS and LTE concurrently on the nRF9160DK and am wondering if the rootcause is an incompatibility between modem FW v1.2 and hardware v0.9.0. I am using:

-ncs v1.3

-nRF9160DK hardware v0.9.0

When I flash modem firmware v1.2 and run any ncs sample that uses A-GPS (ie asset_tracker), everything works as expected. LTE enters PSM mode, and GPS can run concurrently. But when I run the ncs sample "gps" without using A-GPS, the GPS will track ~10 satellites, never using any and never achieving a fix. This issue is resolved by reverting to modem firmware v1.1.3 or older, and the "gps" sample then operates as expected. So, since the only difference between these two cases of "gps" sample working vs not working is the modem firmware version, I believe this points to an issue in modem firmware v1.2 and hardware v0.9.0.

I would use modem FW v1.1.3 for development then if not for issues with this firmware version:

-PSM mode is most often denied by the network, whereas it is always accepted by the network using modem FW v1.2. I am using iBasis sim roaming on Verizon network, and have made sure that TAU is large enough to be granted PSM by the network. So, the GPS often cannot run under this modem FW version as well since PSM mode is not granted. 

-LTE network connection takes minutes using v1.1.3 as opposed to seconds using v1.2.

I am wondering if Nordic has seen this issue previously or if it can be recreated and the cause of this issue confirmed. The issue can be created on my end using the "gps" sample fresh from ncs v1.3 without modifying any configurations or source code. 

Thank you

[Didrik Rokhaug]

Hi.

I am a bit confused about what you are writing. If you are using the "gps" sample unmodified, the nRF9160 will not use PSM, but simply not start the LTE stack. If you modify the sample to use A-GPS, it will still not use PSM, but instead start and stop the LTE stack when fetching the A-GPS data.

 

-LTE network connection takes minutes using v1.1.3 as opposed to seconds using v1.2.

 Does this happen every time you use v1.1.3, or only the first time after v1.1.3 has been flashed to the modem?

One way to speed up the attach sequence is to issue AT+CFUN=0 after attaching. That will store the current network parameters in flash, which can speed up subsequent attaches.

I am not aware of any compatibility issues between mfw v1.2.0 and DK v0.9.0. That combination was tested as part of the release of mfw v1.2.0, and I am not aware of any problems related to the GPS that was uncovered during that testing.

Does it work if you use NCS v1.2.0?

Do you have other DK versions that you can test with?

What GPS antenna are you using?

I do not have any v0.9.0 DKs available at the moment, but the GPS sample works as expected on a v0.8.5 DK with an external patch antenna.

Best regards,

Didrik

[gscultho]

Hi Didrik, thanks for the response. 

Allow me to clarify. Any sample project I have tried that uses A-GPS to start the GPS works as expected using modem FW v1.2. So, "asset_tracker" and "nrf_cloud_agps" both work as expected and are able to achieve both LTE and GPS operation using modem FW v1.2. However, when trying to use the GPS without A-GPS ("gps" sample, for example) with modem FW v1.2, a GPS fix is never achieved. About 10 satellites are tracked, but zero used and no fix. The "gps" sample begins to work correctly simply by changing modem FW from v1.2 to any previous version. 

I have just tried to run the "gps" sample with ncs v1.2 and modem FW v1.2, and the problem persists. The GPS tracks many satellites but achieves no fix with modem FW v1.2 flashed. As soon as I flash modem FW v1.1.3 or any other and reset the board, the GPS achieves a fix in under 30 seconds. 

To directly answer your questions:

- The slow connection time occurs every time with modem FW v1.1.3, and PSM is usually denied as well (it is always accepted with FW v1.2).

- The problem persists with ncs v1.2.

- My team has another DK that I can try but they are all v0.9.0.

- I am using the included Molex antenna. SW12 is switched to "CON".

I have not tried the command AT+CFUN=0 but will do so and see if it improves performance when using modem FW v1.1.3. 

Thank you for your support.

[Didrik Rokhaug]

Hi, and sorry for the late answer.

Could you take a modem trace?

That would let us see what is happening on the modem side.

I have seen similar behavior when LTE was running in the background (no PSM or eDRX, but no user data sent or received either). This only allows the GPS to run for very short periods of time between the Paging Time Windows, which didn't seem to be enough to allow a fix. The positioning of the device didn't help much either, as it was indoors with poor view of the sky. But, as the GPS sample does not activate LTE (unless using AGPS), I don't think that is the problem in your case.

Best regards,

Didrik

[gscultho]

Hi Didrik,

I tried a new board (same hardware version) and did not see the same issue, so I assume it was specific to that one board. I can still take modem trace if desired, but this issue is resolved on my end.

[Didrik Rokhaug]

Great to hear.

If the problem is limited to just that one DK, I don't think a modem trace is necessary at this moment.

Best regards,

Didrik