Got a new S22ultra and ran the App. NOW I have questions...

729 views
Skip to first unread message

Robert Enger

unread,
Mar 22, 2022, 6:30:21 PM3/22/22
to GPSTest
I finally broke down and bought a new phone.  (No more LG V20).
I now own an S22 Ultra with Snap 8 gen 1 chip.

I loaded up the app and saw some stuff I don't understand.  May I bounce some peanut-gallery-level questions off the gurus on this list?

1.  Eleven high chip rate signals are available at this time (see screen cap).  WHY is position computation using any of the comparatively imprecise L1 signals.  Doesn't use of L1 signals in the computation lower position accuracy?
2.  Why are satellites that don't have ephemeris data being used in position computation.  Doesn't that introduce some sort of imprecision in the position computation?
3.  Why don't any of the Galileo satellites show ephemeris data promptly.  (It took 7 or 8 minutes for Ephemeris to show for Galileo, while Navstar info showed up promptly.)
4.  No SBAS are listed.  The old V20 used to list one (or even two sometimes).  I don't know if the positioning system in the V20 was smart enough to use the correction info transmitted by the SBAS signals, but presumably the fancy chip in the S22 ultra IS.  Why is SBAS being ignored?
5)  Although I see one QZSS satellite (at very low angle), NO BeiDou satellites are shown.  Is Qualcomm still blacklisting BeiDou, even in the new Snap 8gen1 chip?

Screenshot_20220322-134356_GPSTest.jpg

Sean Barbeau

unread,
Mar 22, 2022, 6:46:04 PM3/22/22
to Robert Enger, GPSTest
Good questions Bob!

>Eleven high chip rate signals are available at this time (see screen cap).  WHY is position computation using any of the comparatively imprecise L1 signals.  Doesn't use of L1 signals in the computation lower position accuracy?

The number of available L1/E1 signals from satellites is still much higher than L5/E5a, so depending on the overhead satellite geometry, signal strength, multipath, etc. the positioning engine might still value them in addition to or over L5 in some cases. This would vary per chipset depending on the exact positioning algorithms used by the hardware.

>Why are satellites that don't have ephemeris data being used in position computation.  Doesn't that introduce some sort of imprecision in the position computation?

Great question, I've always wondered this too. In short, yes - it would introduce a large degree of error. So I've always assumed that the use of a signal that doesn't have ephemeris data is an error in the "E" flag from the hardware - or the definition of "has ephemeris" that the hardware is using is different than what we think it is. The Android spec docs just say "Reports whether the satellite at the specified index has ephemeris data."

For example, maybe the "E" flag definition used by Qualcomm is that the ephemeris was decoded over the air (OTA), although it could still be fetched over other channels (e.g., via SUPL over IP network) and still be used in a fix even if it hasn't been decoded from an OTA signal yet. The Dashboard feature I've been working on will flag these issues explicitly as errors to make them more obvious.

>Why don't any of the Galileo satellites show ephemeris data promptly.  (It took 7 or 8 minutes for Ephemeris to show for Galileo, while Navstar info showed up promptly.)

Likely same issues as above - either error in "E" flag or hardware definition isn't what we think it is.

>No SBAS are listed.  The old V20 used to list one (or even two sometimes).  I don't know if the positioning system in the V20 was smart enough to use the correction info transmitted by the SBAS signals, but presumably the fancy chip in the S22 ultra IS.  Why is SBAS being ignored?

SBAS visibility varies by time of day - make sure you're testing at the same time. If you're testing at the same time, it could just be the ongoing issue of SBAS flakiness on devices. Even on the same device, sometimes I'll start up the app and see SBAS, kill the app, and then start it up again and SBAS will be missing. There are also known issues on Android with definitions here - we don't know what the "U" or "used" flag means for SBAS. Does it mean corrections are used, or just the L1 signal broadcast?

>Although I see one QZSS satellite (at very low angle), NO BeiDou satellites are shown.  Is Qualcomm still blacklisting BeiDou, even in the new Snap 8gen1 chip?

If you're in the US, the FCC (and therefore Qualcomm) is still restricting use of BeiDou here - see https://barbeau.medium.com/where-is-the-world-is-galileo-6bb7bfa29e. QZSS seems to have been exempted along with GLONASS due to its early use in chipsets - for some reason the FCC only cracked down on Galileo and Beidou.

Sean

--
You received this message because you are subscribed to the Google Groups "GPSTest" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gpstest_andro...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gpstest_android/9a03ac6d-b424-41a6-bb38-b6fd3194969an%40googlegroups.com.

Robert Enger

unread,
Mar 22, 2022, 6:50:43 PM3/22/22
to Sean Barbeau, GPSTest
 Thanks!

lodrog...@gmail.com

unread,
Mar 23, 2022, 4:28:42 AM3/23/22
to GPSTest
from my experience i will accept "what you pay, is what you get". Amateurs devices ARE not professional devices. All of those problems / issues / restrictions, even into US area with a professional GNSS receiver does not exists.
this is my opinion

An Stable

unread,
Aug 24, 2022, 9:58:50 AM8/24/22
to GPSTest
I've just aquired a Galaxy Note 10 running Android 12 and I only see US Gps sattelites in Norway.
My old S7 shows both Russian and Chinese. Not European though since it is too old.
I've tried to find any missing setting, but so far I no luck. 

Thanks for a very nice utility !
anst

ndo...@gmail.com

unread,
Aug 26, 2022, 10:46:17 AM8/26/22
to GPSTest
Great answers from Sean , thanks!  One clarification and one addition.

>SBAS visibility varies by time of day

To be clear, SBAS utilizes geosynchronous satellites that are always in the same spot relative to the user - meaning the SBAS system message stream availability does not change.  I think what Sean intended to say is: visibility might change because of obstructions as you move around, or because of the general android flakiness in regards to tracking SBAS.  If you had a high quality, certified aviation receiver in the cockpit of a plane in the sky with no obstructions, the SBAS system is always there providing corrections from typically 3 geo satellites.  For WAAS (USA SBAS), there has been no complete loss of signal-in-space since it was commissioned in 2003.  There have been (typically short duration) single-geo outages due to maintenance or failures - but never all the Geo's at once.  There's a ton of redundancy.  Current GEO footprints for WAAS: https://www.nstb.tc.faa.gov/DisplayGEOStatus.htm



>  WHY is position computation using any of the comparatively imprecise L1 signals.  Doesn't use of L1 signals in the computation lower position accuracy?

The receiver also must track and utilize both L1 & L5 frequencies from a single satellite to eliminate the largest source of error in the GPS "error budget" - ionospheric error.  Induced ionospheric error is proportional to the signal frequency.  If you have two frequencies, you can nearly eliminate it completely.  As opposed to utilizing the Klobuchar model for single frequency GPS users, or the more detailed Iono model from an SBAS for SF users.  The SBAS Iono models are based on many more measurements, and provides a denser (tighter grid of lat/longs) correction stream when compared to the GPS model.  Klobuchar was intended to remove ~50% of the iono error world-wide, while SBAS removes nearly all of it for its smaller region of precision-approach availability.  Both models are inferior to the user using dual-frequency to directly measure & eliminate it - which is why the U.S. military has been doing it since the get-go.  Only now that additional civil frequencies are available do we start to enjoy the advantages.



Валерий Заподовников

unread,
Sep 29, 2022, 5:46:13 AM9/29/22
to GPSTest
I have S22 Ultra Snapdragon SM-S9080. 0 model has triple-frequency chinese BeiDou (B1, B2A, B1C), as described in https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8838512/pdf/sensors-22-01289.pdf

Absolutely crazy thingy! 3.8/2.5 accuracy.

Actually, all signals are the same. There is of course that GLONASS 
has bigger time to fix even with SUPL data cause of how its time transmission works...

Glonass satellites transmit time information once per 30 seconds (L1OF, L2OF signals). GPS satellites transmit time information once per 6 seconds. Of course you can take time from a different constellation...

Also see this, it shows that Galileo may be problematic.
My Galileo also works good, but see this, there is also some russian CDMA on the second screen https://github.com/barbeau/gpstest/issues/264 

Screenshot_20220928-200954_GPSTest.pngScreenshot_20220927-151623_GPSTest.png

lodrog...@gmail.com

unread,
Sep 30, 2022, 12:22:39 PM9/30/22
to GPSTest
On Thursday, September 29, 2022 at 12:46:13 PM UTC+3 val.za...@gmail.com wrote:
I have S22 Ultra Snapdragon SM-S9080. 0 model has triple-frequency chinese BeiDou (B1, B2A, B1C), as described in https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8838512/pdf/sensors-22-01289.pdf

Absolutely crazy thingy! 3.8/2.5 accuracy.

i will agree with you .... "crazy". Of course this is not logical and the money you spend is lost. From year 2011, my old and still working android device Sony MT11i has 3 meters static accuracy using the same app - the GPSTest. When we go back to year 2017 who can forget the announce of "smartphones will have 30 cm accuracy"

that's all folks
 

Валерий Заподовников

unread,
Sep 30, 2022, 12:57:46 PM9/30/22
to GPSTest
2011 did not even have glonass or dual frequency, not even talking about triple. What is the confidence for that 3 meters, is that vertical or horizontal, does it work in the room without open sky, does it have AI for multipath and without GNSS work on wifi strength signals or time sync to a nanosecond using Stratum 0, not possible for NTP? It may have 30 cm accuracy, I did not test ground truth, one need to use normal maps, QGIS for this and PPP. That thing only reports info from the chip for 68.27% confidence (what is called 1 sigma). But that does not mean that the actual accuracy is not higher ever since GPS migrated to G2139 (again, QGIS) in 2021, in fact even that 68.27% is just estimated and is often wrong, as author of gpstest noted.

As for money I did not even know it will have it and that BeiDou will work, in fact I bought 0 model last moment (see how powerful Trump was, wow)! My previous S21 FE snap 888 had those features blocked, even GPS L5. I bought it to test armv9-a, new CPU architecture. So far a failure, only armv8.2-a works. Still faster than armv8, but kinda pathetic. Clang 15 will work better, I hope.

lodrog...@gmail.com

unread,
Sep 30, 2022, 3:00:05 PM9/30/22
to GPSTest
2011, sony mt11i had snapdragon s2 with both support to gnss systems of GPS and glonass. My device works from this period just fine for both systems. Still does.
see the specs

over 30 years now, location satellites had multi band and multi frequency, of course not at low price for amateurs devices. Now we have all of them at low cost 100-200 dollars .... but the 3 meters static accuracy has stuck at the same value. No matter if you pay 150 or 1500 dollars you will have the same navigation accuracy.
My personal opinion is probably it is unlikely that we will see something better for many more years. Expensive smartphones for navigation accuracy reasons is money for trash.


lodrog...@gmail.com

unread,
Sep 30, 2022, 3:13:52 PM9/30/22
to GPSTest
- Sony MT11i works all this years also inside house, one room away from window with no issue. At year 2012, a Garmin nuvi 3760 never did it to work. Not even 3 meters away from the window.
- 3 meters accuracy is location data
- there is no AI nowhere. It is a good way from the companies to foolish the customers. Every new year we have new functions with updated software programming instructions. This is the real fact.

Валерий Заподовников

unread,
Sep 30, 2022, 3:46:22 PM9/30/22
to GPSTest
"one room away from window with no issue."

Impossible. Even my G22 only gets a fix only near the window.  AI is needed to fix phased array, "smart antenna". With normal antenna one do not need AI, but it is what it is. Anyway, can you phone do RTK? Only Glonass? Okay. New Exynos S22 can do RTK, but not in 4nm SOC Snapdragon, it also uses Broadcom.

lodrog...@gmail.com

unread,
Sep 30, 2022, 4:52:01 PM9/30/22
to GPSTest
On Friday, September 30, 2022 at 10:46:22 PM UTC+3 val.za...@gmail.com wrote:
"one room away from window with no issue."

Impossible. Even my G22 only gets a fix only near the window.  AI is needed to fix phased array, "smart antenna". With normal antenna one do not need AI, but it is what it is. Anyway, can you phone do RTK? Only Glonass? Okay. New Exynos S22 can do RTK, but not in 4nm SOC Snapdragon, it also uses Broadcom.

from 2012 all my devices can do the same, at 3 houses i have test them. the list is
- sony mt11i
- sony LT30
- samsung s3
- 1+ 5t

not all of them at the same year, not in the same house. the max number is 3 of them in the same place

Валерий Заподовников

unread,
Oct 1, 2022, 11:59:20 AM10/1/22
to GPSTest
My Ultra can also get GPS 1 or 2 satellities and one chinese far from window. That is not enough to get a fix, unless you have an atomic clock synced to Geneva. And even then, this quality of the signal is horrible as can be seen in GNSSlogger by Google. And I suppose I would like to dissapoint you, but your phones are lying about accuracy. Paper from author of GPStest (c. 2011 phones): https://www.researchgate.net/publication/231849997_Positional_Accuracy_of_Assisted_GPS_Data_from_High-Sensitivity_GPS-enabled_Mobile_Phones

lodrog...@gmail.com

unread,
Oct 1, 2022, 12:33:37 PM10/1/22
to GPSTest
just for the record

when the devices are tested with GPSTest internal the house, according the sky conditions, i read variable measurements of accuracy. Those values can be from 10-120 meters. They are very stable for a long time. The time to TTFF can be from 3-20 minutes. I have also run navigation apps like the osmand,locus,orux to verify the real location. Don't forget that my device is into airplane mode - with no sim card

Валерий Заподовников

unread,
Oct 1, 2022, 2:23:48 PM10/1/22
to GPSTest
Apparently my S22 can do it too, just checked in the restaurant. In seconds, after 2 minutes rather high precision. Previous tablet S4 cannot do it, wow, even after 5 minutes. Thanks for making me do these tests.Screenshot_20221001-211543_GPSTest.png


суббота, 1 октября 2022 г. в 19:33:37 UTC+3, lodrog...@gmail.com:

lodrog...@gmail.com

unread,
Oct 1, 2022, 7:06:59 PM10/1/22
to GPSTest
I am using this test to determine the quality of sensitivity from a gnss receiver. The meaning of the word "sensitivity" for the receiver has to do with how good is the antenna and the amplifier to the front part of the electronic design of the device. No matter if we are talking about one or more bands at the device. Every device must to be tested.

How to do that before we buy the device ?
We can visit a local shop, that have demo devices.At most cases, we can found access to internet via wifi of the shop. Then we use google chrome to download the GPSTest. If this is not possible, we can transfer the GPSTest apk via bluetooth from our smartphone. It takes a few seconds.

results
now we are able to see if a device into problematic conditions can have 1. location, 2. how fast, 3. what accuracy

see this video and replace those test devices with android smartphones. Of course, this is very hard for all of us to make it. That's why i suggest a shop

binghy17

unread,
Oct 3, 2022, 5:13:05 AM10/3/22
to GPSTest
Expensive smartphones for navigation accuracy reasons could be money for research.

The fact to have opened access to the "black-box-chip" to refine measurement processing (by the means of algorithms) is a huge step towards this aim.

lodrog...@gmail.com

unread,
Oct 3, 2022, 12:26:47 PM10/3/22
to GPSTest
On Monday, October 3, 2022 at 12:13:05 PM UTC+3 m.gian...@gmail.com wrote:
Expensive smartphones for navigation accuracy reasons could be money for research.

the navigation accuracy and research is not the same conditions. The navigation is to travel from the point A to point B. No one cares about accuracy below 5-10 meters. The user only need to go to the desire place fast and economy. Even a 100 $ plastic android smartphone can be used for safe navigation needs.
research is for the scientists. Don't try to confuse the customers.

 
The fact to have opened access to the "black-box-chip" to refine measurement processing (by the means of algorithms) is a huge step towards this aim.

 if you like to have "play time" you can buy one exterrnal GNSS receiver. From 160$ with bluetooth + usb simultaneous output, dual band, 50cm (50% CEP) 150cm (95% CEP), Update Rate 1Hz / 5Hz , Messages: GNRMC, GNGGA, GNGSV, GNGSA (Default); GNGLL, GNVTG (Optional)

gpsfan

unread,
Oct 3, 2022, 12:54:29 PM10/3/22
to GPSTest
Got a link for that receiver ? Cheapest Dual-Band I'd seen so far were the Polaris SkyTraq offerings but that's way more https://www.polaris-gnss.com/Shop

lodrog...@gmail.com

unread,
Oct 3, 2022, 1:57:13 PM10/3/22
to GPSTest

Got a link for that receiver ? Cheapest Dual-Band I'd seen so far were the Polaris SkyTraq offerings but that's way more https://www.polaris-gnss.com/Shop

Yes my friend, i have this. The company has also official resellers. The accuracy is only from satellites. No need for RTK / RTX

product

OS compatibility

drivers

shops

gpsfan

unread,
Oct 3, 2022, 3:32:40 PM10/3/22
to GPSTest
Thanks, I had a quick look and it does look very interesting. It must be the Broadcom 47755 chip like the logger that was discussed last year ? I can't see any reference to the chip on their site, a bit odd.
They don't mention any RAW format either that would allow for post-processing or RTK.

lodrog...@gmail.com

unread,
Oct 3, 2022, 5:31:04 PM10/3/22
to GPSTest
On Monday, October 3, 2022 at 10:32:40 PM UTC+3 gpsfan wrote:
Thanks, I had a quick look and it does look very interesting. It must be the Broadcom 47755 chip like the logger that was discussed last year ? I can't see any reference to the chip on their site, a bit odd.

there is no official info about the internal gnss chip, but the model p10 pro that use the BCM47755 has 100% the same receiver specs like the p7 pro. logical they will use the same chip


i am very glad to learn that the 47755 is a dead product. This info helps me explain this ....
the columbus products P10 pro and P7 pro, both of them have supports for Navic L5  ... with 47755. But this chip is not NAVIC capable
according Broadcom official web pages
https://www.broadcom.com/products/wireless/gnss-gps-socs/bcm47755
https://www.broadcom.com/products/wireless/gnss-gps-socs/bcm47765
the 47755 is not NAVIC capable, only 47765 is. So, maybe .... the columbus products is using 47765 with Beidou B2a disabled. This is a personal thinking



They don't mention any RAW format either that would allow for post-processing or RTK.
 
the sure is only that those products have NMEA 0183 output. Is this help you for post processing ?

Валерий Заподовников

unread,
Oct 4, 2022, 1:28:48 AM10/4/22
to GPSTest
https://www.youtube.com/watch?v=5byHT3yr50U

That is very interesting, but Garmin (watches, that is) is a real joke, see https://www.dcrainmaker.com/2022/07/gps-accuracy-issue-impacted-garmin-suunto-and-polar-watches-over-past-week.html cause they had a bug in this file: https://github.com/StevenMaude/epo-bin/tree/epo-bin  
and also see this https://youtu.be/Kn96cza2BZE 

Now iPhone 13 does not support dual freq. Besides that is iPhone, who knows how it works on their closed OS and their ARM fork of a chip.

As for GPS external, use FPGA with GNSS-SDR, see we even added support for GLONASS Hamming code correction. https://github.com/gnss-sdr/gnss-sdr/pull/655/commits/533e107781ae423f3ad874d6fd10dce1cab512e8

gpsfan

unread,
Oct 4, 2022, 4:44:16 AM10/4/22
to GPSTest
Thanks, unfortunately NMEA data can't be used for post-processing. I sent them an email asking about the availability of RAW data.

Lukoviczki András

unread,
Oct 4, 2022, 5:32:25 AM10/4/22
to lodrog...@gmail.com, GPSTest
the BCM47755 is able to receive NAVIC signals with an update , the Broadcom did it already in 2018 (the Indian SAC asked them) and they demonstrated it with a Xiaomi MI8  , and the Pixel5 is able to receive it by default (if You check the GPSTest database, there are many Pixel5 phones with newer dll and with Navic signals )

here is the full article:

probably (it's my opinion) the Columbus could get old 47755 chips on very low price (as EoL models) ~1.5-2 years ago and they assembled (because they still have on stock , they couldt sell enough P10Pro)  the P7 pro with this chip also. if they run out from these chips, they should swap to 47765 or the newest 4778 (but with a new model name like P7 Pro Mark II, due to they cant "upgrade" the sold models to B2a signal with a simple fw update.)

--
You received this message because you are subscribed to the Google Groups "GPSTest" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gpstest_andro...@googlegroups.com.

lodrog...@gmail.com

unread,
Oct 4, 2022, 8:39:17 AM10/4/22
to GPSTest
Thanks Andras for the tech info

is seems that broadcom does not like to inform end users at the official web peges it has, about updates specs of an older design (47755) or even worst to say more details into their new products (4778). Especially for the new v3 gnss chip 4778 the company seems to afraid to say how good or worst is compared with the older models

Валерий Заподовников

unread,
Oct 5, 2022, 7:13:43 PM10/5/22
to GPSTest
Look into this madlad who tested L5 in his Apple Watch Ultra by mowing a lawn: https://9to5mac.com/2022/10/05/apple-watch-ultra-gps/

Meanwhile new firmware of Garmin watch is still somehow winning. Hilarious. https://youtu.be/ufI-uOE49xk
Reply all
Reply to author
Forward
0 new messages