Getting GPS fix with nRF9160-DK and Thingy:91

[bojan]

Hello guys.

Here with me I have both nRF9160-DK, hardware version 0.8.3 and Thingy:91 boards.
I am on the nRF Connect SDK tag 1.2.0 and have nrf9160 SiP modem firmware updated to 1.1.1 version.
When trying only gps example that I found under samples/nrf9160/gps I was unable to get any GPS fix even when using external antenna (TW4020 from Talysman):

Scanning [-] [1;1H[2JTracking: 0 Using: 0 Unhealthy: 0
Seconds since last fix 1742

Scanning [\] [1;1H[2JTracking: 0 Using: 0 Unhealthy: 0
Seconds since last fix 1743

I know there is some issue with the iBasis SIM roaming cards and local carriers (I am located in France, by the way) that does not allow entering PSM mode. Is this issue occuring in GPS-only example as well? To the best of my understanding, LTE is not active in GPS example.

I don't see any switch on 0.8.3 version of nRF9160-DK board for a selection between on-board/external GPS antenna, am I missing it? Do I need to do something more in addition to connecting external GPS antenna to appropriate connector?

I even tried with some Orange SIM card thinking it is a local card but this card is, to the best of my understanding, registered as a roaming card under other carrier (Bouygues):

Where would be the right place to ask for appropriate SIM cards?

Finally, I was happy to watch your webinar about SDK v1.2 and new exciting GPS solutions (SUPL, A-GPS...). Would that things make our GPS connection easier or we would still need to resolve the issues with local carriers (entering PSM mode...)?

Thanks for your time and efforts.

Sincerely,

Bojan.

[Pete Skeggs]

I'm not sure the detailed differences between 9160DK 0.8.3 and 0.8.5, but, version 0.8.5 has a tiny SW12 near the GPS antenna connector.  Switch it in the CON position to use the connector, or the ANT position to use the on-board antenna.

[bojan]

I was unable to see any switch near GPS antenna in 0.8.3 version. I bought the board from Mouser few weeks ago so I wonder why the latest version is not delivered to me!?

[bojan]

Karl Ylvisaker,

I have TW4020 external antenna from Tallysman as well as MXHS83QE1000 adapter cable from Murata. I see the mentions of bias tee for external GPS antennas that are active.

Do I need Bias tee for Thingy:91 board and TW4020 external antenna? If affirmative, how can I mount it on Thingy:91 board?

Regards,

Bojan.

[bojan]

Karl Ylvisaker, TW4020 seems to be an active antenna with Bias Tee needed. In many of your posts (e.g. here or here) you say all we need is an adapter (like the MXHS83QE3000) and external antenna.

Anyway, according to this post, there is a chance we get GPS fix if we use GPS in GPS/GNNS standalone mode with LTE mode turned off. Is it enough to apply 

"AT\%XSYSTEMMODE=0,0,1,0"

 command to select GNSS support only or we need to explicitly turn off LTE mode somehow?

Sincerely,

Bojan.

[Karl Ylvisaker]

Hi Bojan,

bojan said:

Do I need Bias tee for Thingy:91 board and TW4020 external antenna? If affirmative, how can I mount it on Thingy:91 board?

Yes, like you mention in your latest reply - all active antennas need Bias Tee. When connecting the external antenna, the bias tee needs to be connected between the SWF connector and the external antenna cable. The SWF connector is marked "J2" on the hardware illustration on this page.

bojan said:

Anyway, according to this post, there is a chance we get GPS fix if we use GPS in GPS/GNNS standalone mode with LTE mode turned off.

 The post you reference is in regards to having LTE and GPS run concurrently, which requires PSM(but you may do the procedure Sigurd describes, if you need to use both). Unfortunately, activating standalone GPS mode does not improve GPS antenna performance, and the external antenna is still needed.

bojan said:

 command to select GNSS support only or we need to explicitly turn off LTE mode somehow?

By running the AT-command you suggest, you disable LTE and enable only GPS, which is correct. Remember to also use the Set command to activate your GPS module, in addition to enabling it.

Best regards,
Karl

[Karl Ylvisaker]

Hi again Bojan,

I have checked with some of my colleagues about this issue, and it seems that using an active external GPS antenna with the Thingy:91 at times can be very frustrating.
The main reason for this seems to stem from the added loss by the required external components.
Are you currently developing primarily on the nRF9160-DK, or the Thingy:91?

Best regards,
Karl

[bojan]

Hello, Karl Ylvisaker.

We are prototyping with Thingy:91.

I tried with Thingy:91 on-board antenna and GNSS/GPS mode and was able to get the fix in about 110 seconds by putting the Thingy outside my window. The accuracy of the location is quite satisfying. With external antennas, SUPL, A-GPS and the things you announced in your latest webinar about SDK v1.2, I guess this will be a success story!

This about external active antenna and Thingy:91 is quite important information. You are obviously internally aware of that weakness but there is no will from your side to spread that info to the community in a clear way. By reading your posts (e.g. this one), I first though all I need is adapter cable and external antenna. Then, in the meantime, after hours of error and trial, I realized I need a Bias tee as well because the antenna is active. I ordered the Bias tee and now it seems that there is some issue with Thingy:91 and active external GPS antennas. Quite frustrating!

We started with our product development almost two months ago and at that point I convinced the team to go with Nordic solution based on my prior Nordic experiences (mainly around nRF52 BLE modules). During the past 2 months, we spent 90% of the time in setting up the environment and we still feel the pain! There is a nRF52840 module on the Thingy board but it mainly serves as nRF9160 controller and debugger. It is quite difficult to keep that debugging feature inside nRF52840 and add some BLE services on the top of that. I still desperately wait for your feedback on that topic (this thread).

nRF9160-DK boards with some hardware faults (0.8.3 and older) are still on the market. I contacted Mouser, as you suggested, to find out why did they send me the old version and they simply said they are selling from their stocks in a FIFO manner. It is not their fault, they are just selling what you provided them.

I start feeling bad and guilty because I convinced people to go that way. Let me be clear here, I don't have any doubt you are working hard in the background to provide us a high quality development environment. Neither I have doubt your products are great (even the best on the market)! But my managers are not technical people, they just see we are eating time and budget at the moment.

More useful examples, more stable development environment with easy debugging possibilities, more responsive technical support, spreading the info in a clear way... all that is needed if you want to keep your image on the market!

Until the day I begin feeling proud and managers start tapping my shoulders on the corridors, I am sending you my best regards! Let's all grow together!

Cheers!

Bojan.

[Karl Ylvisaker]

Hello Bojan,

Sorry for my late reply.

bojan said:

We are prototyping with Thingy:91.

With the troubles you describe, you may have an easier time prototyping your GPS features on your nRF9160-DK, instead of on the Thingy:91.

bojan said:

With external antennas, SUPL, A-GPS and the things you announced in your latest webinar about SDK v1.2, I guess this will be a success story!

Yes, this is an errata we are working diligently to alleviate, and the new version will improve performance.
I am glad you enjoyed the webinar!

bojan said:

You are obviously internally aware of that weakness but there is no will from your side to spread that info to the community in a clear way.

The errata on the nRF9160 has been known for some time, as it is documented here and here.
I agree with you that we should increase our efforts to communicate clearly the workaround for this errata.
Especially the parts concerning the additional required components to achieve satisfactory GPS performance.
I am very sorry to hear that your experience with this product has been frustrating, and internally we have opened an issue regarding the current state of the external antenna communication. I will let you know as soon as it has been resolved(even though you have already gone through this, know that it will be addressed).

bojan said:

I convinced the team to go with Nordic solution based on my prior Nordic experiences (mainly around nRF52 BLE modules).

I am glad to know that your previous experience with us has been more pleasant, and that you would recommend our products to your team! 

bojan said:

More useful examples, more stable development environment with easy debugging possibilities, more responsive technical support, spreading the info in a clear way... all that is needed if you want to keep your image on the market!

Thank you Bojan, for all your feedback on this issue - I have documented it, and forwarded it to the appropriate people.

Best regards to you as well!

Karl

 

[Karl Ylvisaker]

Hello again, Bojan

This might not be relevant for you anymore, but I thought I should provide an update on the matter since I promised that I would in my previous reply.
In summary, to clarify the GPS antenna issue of the Thingy:91 v1.0.0 we published an Errata for the issue.
This errata has also been addressed in later releases of the Thingy:91 (after v1.0.0), where GPS performance is increased. There is however a known issue with using the GPS of the Thingy:91 while the USB is connected, as this degrades performance. For best performance, it is also recommended to make use of AGPS.

Thanks again for the feedback you provided on this!

Best regards,
Karl

[bojan]

Hello, Karl Ylvisaker!

Thank you very much for the update! 

I should start a new project very soon where connectivity to the Cloud and getting GPS coordinates will be some of the requirements.

Some people within the team are inclined towards using NL-SW-LTE-QBG95 modem but I myself would prefer we go with nRF9160.

The device we are gonna design will use external antennas for both cellular and GPS. I did not use nRF9160 for quite a while. Can you tell me what's the current state of GPS part of the modem:

- Do you support assisted GPS that will enable us to get a cold fix within 10seconds?

- Is there any known issue remaining around GPS?

Thanks in advance.

Sincerely,

Bojan.

[Karl Ylvisaker]

Hello again, Bojan

bojan said:

Thank you very much for the update!

No problem at all, Bojan - I figured it was better late than never!

bojan said:

Some people within the team are inclined towards using NL-SW-LTE-QBG95 modem but I myself would prefer we go with nRF9160.

I am happy to hear that Bojan, and I hope that the fixes we have implemented since last we spoke provides you with the functionality your upcoming project requires!

bojan said:

- Do you support assisted GPS that will enable us to get a cold fix within 10seconds?

When you use A-GPS you will not actually get a cold start, but rather a warm or hot start which most likely will have a < 10 s TTFF.

It is hard for me to make any exact estimates on the TTFF with A-GPS since A-GPS also depends on which country you are in I would therefore strongly recommend that you test this yourself on your end, with a recent DK updated with the most recent modem firmware. This is what will give you the most accurate answer.

bojan said:

- Is there any known issue remaining around GPS?

No, there are no known GPS performance issues other than the degraded GPS performance for the Thingy:91 while the USB is connected.

Best regards,
Karl

[bojan]

Hello, Karl Ylvisaker!

Thanks for the encouraging updates!

We would most probably start with the development kit and that's why I would like to find out more about the following:

- Are there any limitations concerning the GPS part on the nRF9160 development kit?

- Can we use an external GPS antenna with the devkit without any problem? Do we need a bias Tee for passive antennas?

Finally, what is the latest hardware version of nRF9160 devkit, and can we explicitly request it when ordering from your distributors? It can happen that they deliver an older version of the devkit because they have it in their internal stocks.

Cheers,

Bojan.