Thingy91 GPS without AGPS never acquires lock?

RE: Thingy91 GPS without AGPS never acquires lock?

Albrecht Markus Schellenberger — Thu, 29 Jul 2021 06:41:36 GMT

Hello Arnout,

[quote user="Arnout"](Btw for some bizar reaon, I cant reply to your exact post (buttons are just not there), so the page chronology looks really weird, but ok)[/quote]

that is strange, but no worries :-)

[quote user="Arnout"]

I have this with both:

- roaming sim card from "whereby" => uses belgium orange network

- roaming sim from "olivia wireless" => uses belgium proximus network

[/quote]

It is not always given that these so called virtual network operators have PSM included in their negotiation with the main network operators. On the other hand, it also can happen that if the device doesn’t exactly request the PSM values supported by the operator, it can be rejected PSM anyway.

So to start with, I would recommend you to check with both Whereby and Olivia Wireless if they support PSM and what the device should request.

Regards,

Markus

RE: Thingy91 GPS without AGPS never acquires lock?

Arnout — Wed, 28 Jul 2021 10:12:26 GMT

Hi,

(Btw for some bizar reaon, I cant reply to your exact post (buttons are just not there), so the page chronology looks really weird, but ok)

[quote userid="100109" url="~/f/nordic-q-a/77059/thingy91-gps-without-agps-never-acquires-lock/322172#322172"]Just to make sure that I got this right. You requested this after the GPS operation was enabled, right? [/quote]

Yes, that is true.

[quote userid="100109" url="~/f/nordic-q-a/77059/thingy91-gps-without-agps-never-acquires-lock/322172#322172"]What SIM card/network provider are you operating with?[/quote]

I have this with both:

- roaming sim card from "whereby" => uses belgium orange network

- roaming sim from "olivia wireless" => uses belgium proximus network

Kind regards,

Arnout

RE: Thingy91 GPS without AGPS never acquires lock?

Albrecht Markus Schellenberger — Wed, 28 Jul 2021 08:55:22 GMT

Hello Arnout,

[quote user="Arnout"]I guess this means the network does not allow the special PSM features?[/quote]

Yes, exactly. It seems like your network provider does not allow your device entering PSM.

[quote user="Arnout"]I did the AT%XMONITOR command, but seems to keep saying -1 . [/quote]

Just to make sure that I got this right. You requested this after the GPS operation was enabled, right?

[quote user="Arnout"]How big a problem is this? Does this really interfere with the GPS modem? Is it then advised to shut down the cellular connection first fully before attempting the GPS lock? [/quote]

This is actually quite a big problem. See, the LTE stack is always prioritised and as long as it does not switch to idle (which it only does when entering PSM), the GPS socket will not have any possibility to run (long enough). What SIM card/network provider are you operating with?

Regards,

Markus

RE: Thingy91 GPS without AGPS never acquires lock?

Arnout — Tue, 27 Jul 2021 12:09:22 GMT

Hi,

I did the AT%XMONITOR command, but seems to keep saying -1 .

%XMONITOR: 5,"","","20601","0C3D",7,20,"02BF6B05",4,6300,42,24,"","11100000","11100000","01001010"
OK

I guess this means the network does not allow the special PSM features?

How big a problem is this? Does this really interfere with the GPS modem? Is it then advised to shut down the cellular connection first fully before attempting the GPS lock?

I can see that sometimes the state goes from unblocked-to-blocked. I was assuming this was when the cellular modem took over and did some communication again (and thus GPS had to stop). Given that apparently PSM is not enabled, and I can still get locks within 30s, what exactly is the benifit of "really" doing PSM?

Thanks for your clarification on this!

Kind regards

Arnout

RE: Thingy91 GPS without AGPS never acquires lock?

Albrecht Markus Schellenberger — Tue, 27 Jul 2021 12:04:16 GMT

Hello Arnout,

[quote user="Arnout"]Is there more for me to check (e.g. network parameters) regarding the PSM?[/quote]

08:14:25.884 [00:00:07.223,632] #[0m<inf> gps_control: Enabling PSM#[0m
08:14:25.884 [00:00:07.230,194] #[0m<inf> gps_control: PSM enabled#[0m

Sometime after PSM has been enabled, you could cross-check with the AT%XMONITOR command that the network has allowed you PSM and the values to be used. This should look somewhat similary to this (example taken from the Asset Tracker v2):

AT%XMONITOR
%XMONITOR: 5,"","","24201","7AA9",7,20,"01444000",,,,,"","00000011","11000001","01011001"
OK

[quote user="Arnout"](This is for asset tracker v1. I haven't yet tried to look into the v2 again, but I recall it enabling PSM as well?)[/quote]

PSM should be enabled by default, yes.

[quote user="Arnout"]

For now, I'm planning to work in a mode (new pcb design) where the entire nrf9160 is shut down externally for long periods of time, and woken up sporadicaly . It then has to fully reregister to the network, and also get the GPS lock, and send everything to the cloud. In that mode, getting the gps fix asap is key, and the possible "hang" is of course pretty undesirable.

Do you guys have some guideline/tutorial on best practices with using the GPS in such a usecase?

[/quote]

A GPS cold start is possible within around 30 seconds TTFF, under preferable circumstances. Additionally, Low accuary mode can be enabled which makes it possible to get a fix using 3 instead of 4 satellites. But in general, I would recommend you use either

A-GPS to receive satellite data via a cloud service (the device is reporting (multi-) cell location to the cloud). This should go faster compared to a GPS cold start and thus will lower the power consumption.

P-GPS to load more data which is saved in the device memory.

Regards,

Markus

RE: Thingy91 GPS without AGPS never acquires lock?

Arnout — Mon, 26 Jul 2021 09:03:54 GMT

Hi Markus & Markus,

I switched to asset_tracker(_v1), with nordic's AGPS, and a fresh thingy91 to restart some GPS tests. (I've been taking BLE UART logs of the thingy while riding to/from work. The logs combined with the nordic cloud map give me a dataset to work with for getting a feel on how "stable"/fast the GPS locking is. I've ran the vanilla asset tracker demo as well as some modified code that just tries to get a lock or give up in 2 mins, and reboot. (To test "clean" gps performance))

In fully open sky, after boot, I often manage to get a lock after 30-80s. (Never sooner than 30s)
Subsequent locks -can- be much faster (even within a few seconds), but sometimes also seem to just "hang". (e.g. I can be waiting for minutes on a lock. Just give up and make it reboot, and then after 30s it gets the lock). There's little consistency in the behavior, but my feeling is that, with an open sky, if it does not get a lock within +- 90s, its better to just give up and retry. (Note that the periodic LTE data actually also serves as a "reset" for this it seems. After being "unblocked" it sometimes locks very quickly, while there seemed to be no progress before)

Regarding your PSM question: As far as I understand the code explicitly enables PSM before doing the GNSS measurement. An example log file:

devzone.nordicsemi.com/.../tmplog.TXT

08:14:21.229 [00:00:02.509,643] [0m<inf> asset_tracker: Network registration status: Connected - roaming[0m
08:14:21.230 [00:00:02.518,188] [0m<inf> asset_tracker: PSM parameter update: TAU: -1, Active time: -1[0m

...

08:14:25.778 [00:00:07.126,434] [0m<inf> asset_tracker: Starting GPS[0m
08:14:25.859 [00:00:07.198,120] [0m<dbg> nrf_cloud_transport.nct_cc_send: mqtt_publish: id = 1 opcode = 1 len = 641[0m
08:14:25.860 [00:00:07.210,357] [0m<dbg> nrf_cloud_transport.nct_cc_send: mqtt_publish: id = 2 opcode = 1 len = 57[0m
08:14:25.884 [00:00:07.223,632] [0m<inf> gps_control: Enabling PSM[0m
08:14:25.884 [00:00:07.230,194] [0m<inf> gps_control: PSM enabled[0m
08:14:25.884 [00:00:07.236,114] [0m<dbg> nrf9160_gps.enable_gps: GPS mode is enabled[0m
08:14:25.936 [00:00:07.248,168] [0m<dbg> nrf9160_gps.start: GPS operational[0m

(This is for asset tracker v1. I haven't yet tried to look into the v2 again, but I recall it enabling PSM as well?)

Is there more for me to check (e.g. network parameters) regarding the PSM?

Do you guys have some guideline/tutorial on best practices with using the GPS in such a usecase? I checked that the asset tracker v1 does use "full power" GPS. Somehow it does loose the info it had before a reboot (takes 30s VS 3s after a previous lock -- some state is not maintained. (Question is how relevant the state would be after like 1 day though)). PSM should be enabled. Other stuff to take into account?

Will keep you posted,

Arnout

RE: Thingy91 GPS without AGPS never acquires lock?

Markus Tacker (he/him) — Thu, 22 Jul 2021 10:35:17 GMT

I don't recommend to use it without A-GPS, but ... the asset_tracker_v2 has a GPS timeout setting, make sure that it is long enough. Especially if you are not getting PSM from the network, GPS search can be interrupted constantly. Set it to something long, like 600 seconds.

RE: Thingy91 GPS without AGPS never acquires lock?

Albrecht Markus Schellenberger — Thu, 22 Jul 2021 10:29:53 GMT

Thanks a lot for your feedback, Arnout! May I ask if your Thingy:91 is registered to the network with PSM?

Regards,

Markus

RE: Thingy91 GPS without AGPS never acquires lock?

Arnout — Thu, 22 Jul 2021 07:09:10 GMT

FYI, Tried really in an open field the other day. Same thing: Saw 4 satellites within 30 secs, but then fails to indicate them as "using". After 10 mins I gave up.

I'll try to run the asset_tracker (v1) code again, as this clearly worked better, and try and see what the difference is

RE: Thingy91 GPS without AGPS never acquires lock?

Albrecht Markus Schellenberger — Tue, 13 Jul 2021 13:19:08 GMT

Hello Arnout,

Yes, mfw 1.2.3 should be fine.

Regards,

Markus

RE: Thingy91 GPS without AGPS never acquires lock?

Arnout — Tue, 13 Jul 2021 07:01:28 GMT

Hi Markus,

Regarding modem FW, I am running mfw_nrf9160_1.2.3. (The thingy hw version is 1.4.0, and has the NRF B0 HW revision btw).

If I have the chance, I will try the really open-sky thing, but it is impractical for testing from the office of course:)

Kind regards,

Arnout

RE: Thingy91 GPS without AGPS never acquires lock?

Albrecht Markus Schellenberger — Tue, 13 Jul 2021 06:54:34 GMT

Thanks a lot for the detailed feedback, Arnout!

The device should issue a fix whenever it has acquired at least 4 satellites and deems that the accuracy of the fix will be acceptable. So maybe the cellular activity prevents the device from reaching the above mentioned accuracy?

However, on your picture I still can see that the sky is not really open and that there are objects which could interfere with the device. I wonder if you have the possibility to make a reference measurement in a park or similar, where you really have an open sky, and check if that makes any difference?

By the way, what modem firmware are you using? If unsure, you can use the Request revision identification +CGMR AT command.

Regards,

Markus

RE: Thingy91 GPS without AGPS never acquires lock?

Arnout — Mon, 12 Jul 2021 10:12:25 GMT

Hi Markus,

Until now I was testing with the device hanging just outside my window. I now placed it +- on the roof, (see picture) and ran some tests again.

(Obviously its pretty close to the metal. When lifting it 50 cm in the air for 5 mins, it got 2 sats more, but the results stayed the same.. (no lock))

So:

- I am monitoring the logs through a BLE connection with my smartphone
- Running asset_tracker_v2 in sdk1.6.0 now, with nrf cloud enabled (forcing even AGPS to work!), with the patch from the first post to check progress.

Result:

Up on the roof, it finds like 7-8 stats within 30seconds. That's good and all, BUT, again, it says all of them are "unused". I waited 10 mins on the demo code, and it just repeats the process over and over again, (Starting from 0 stats, finding 7,8, but never "using them".. I never seem to get a lock..

Tracking: 8 Using: 0 Unhealthy: 0

Note that the gps demo code DOES work, but uses different code for this (directly parses the nmea messages instead of using the nrf9160_gps lib) + does not do any cellular stuff..

(Maybe some more info: On the gps demo v1.6.0/nrf/samples/nrf9160/gps/: It also shows tracking 5 satellites, 0 used, but within 60 seconds, that suddenly becomes "using 5", and leading to a lock..)

RE: Thingy91 GPS without AGPS never acquires lock?

Albrecht Markus Schellenberger — Wed, 07 Jul 2021 07:47:23 GMT

How “open” is the sky for your Thingy:91 while trying to get a fix, Arnout?

Regards,

Markus

RE: Thingy91 GPS without AGPS never acquires lock?

Arnout — Tue, 06 Jul 2021 12:16:02 GMT

Hi,

Well, for SUPL I am currently running into https://devzone.nordicsemi.com/f/nordic-q-a/77050/supl-library-fails-at_cmd-response-buffer-not-large-enough that issue.

I quickly tried unplugging the USB (using battery) and monitoring the AWS logs. Did not see anything change in the device shadow regarding positioning info for 10 mins, so I guess not. (Haven't tried BLE uart, though that sounds like something that I should definitely try out:)

RE: Thingy91 GPS without AGPS never acquires lock?

Markus Tacker (he/him) — Tue, 06 Jul 2021 11:11:29 GMT

Definitely enable the SUPL client, which will speed up time to fix significantly.

The newer Thingy:91 antenna is affected if plugged in via USB, which will degrade GPS performance. So make sure to unplug it. You can observe logs using BLE UART.

RE: Thingy91 GPS without AGPS never acquires lock?

Albrecht Markus Schellenberger — Mon, 05 Jul 2021 11:19:34 GMT

Hello Arnout,

[quote user="Arnout"]Is it possible the LTE (nbiot) is interfering with the GPS for some reason?[/quote]

LTE/NB-IoT is not interfering with the GPS, but the LTE socket is prioritised unless configured otherwise and the GPS socket runs when the LTE socket is idle.

The antenna of the Thingy:91 has limited performance, especially when the USB interface is used. You could disconnect the USB from the Thingy:91 and log with the RTT interface instead and see if that makes any difference. Also, please make sure that you have an open sky without any buildings etc. that could interfere.

Regards,

Markus

RE: Thingy91 GPS without AGPS never acquires lock?

Arnout — Fri, 02 Jul 2021 14:10:40 GMT

Is it possible the LTE (nbiot) is interfering with the GPS for some reason?

What strikes me as the most obvious difference, is that in the non asset tracker case, the amount of stas used goes up eventually (after 100s), while in the asset tracker case, it -always- stays at zero

[00:03:41.701,782] <dbg> nrf9160_gps.print_satellite_stats: Seconds since last fix 221
[00:03:42.615,509] <dbg> nrf9160_gps.print_satellite_stats: Tracking SV 32: not used,     healthy
[00:03:42.624,786] <dbg> nrf9160_gps.print_satellite_stats: Tracking SV 14: not used,     healthy
[00:03:42.634,155] <dbg> nrf9160_gps.print_satellite_stats: Tracking SV  8: not used,     healthy
[00:03:42.643,493] <dbg> nrf9160_gps.print_satellite_stats: Tracking SV 28: not used,     healthy
[00:03:42.652,832] <dbg> nrf9160_gps.print_satellite_stats: Tracking: 4 Using: 0 Unhealthy: 0
[00:03:42.661,834] <dbg> nrf9160_gps.print_satellite_stats: Seconds since last fix 222

=> How do I debug this further? The gps example seems to use a direct socket connection to the gps, while the asset tracker uses the nrf9160_gps library. Both seem to be working, but .. for some reason the asset_tracker works .. worse.

Please advise

RE: Thingy91 GPS without AGPS never acquires lock?

Arnout — Fri, 02 Jul 2021 13:40:55 GMT

Hmm, build the gps application, and there I do get a lock after +- 3 minutes.

So now the issue becomes why the asset_tracker_v2 does not work, and the gps sample application does.

For reference, the log of this

devzone.nordicsemi.com/.../gps_2D00_application.txt