[Discussion] GPS Problems, Lock speed, Accuracy, Fixes - The Poll

Beowulf_pt · Aug 3, 2010

larryccf said:
the rubber plug is on the far left just under the beveled corner on the sim car - those arrows are from my asking if that pc of black tape on my vibrant was a covering over the GPS antenna, given the socket near it's left end was identified as a GPS socket

I see you added the photo now. I do have such an antenna (used with my Holux BT GPS) but unfortunately I do not have the Samsung to see how well it improves things. The location of the socket doesn't seem very practical if you need to use the antenna often but at least it has one (my Nokia E72 doesn't).

appelflap · Aug 3, 2010

sjdean said:
Ok, I've compared the NMEA reports and the MyTracks results. Not good. Looks like the underlying NMEA information is flying off the handle too.

Not good.

Looks like the NMEA information is wrong too:

http://maps.google.com/maps/ms?ie=U...77579,-1.794891&spn=0.005594,0.01929&t=h&z=16

(Blue Squares indicate NMEA positioning, Blue Track is MyTrack)

Somewhat good news in other regards however, there are times NMEA reports show 10 satellite fixes, but the SGS still only reports on 8.

I only see blue squares on google maps.

Concerning the accuracy of the blue squares: is it really that bad? Most of the time it is tracking you on the road. Sometimes you seem to drive through some houses and the forest

but isn't that an acceptable error (I measured an error average of 22 m. there) . Ok, I see 2 very extreme jumps (did you loose you're fix there), but overall it is surely good enough for a vector based navigation program with some good prediction algorithms.

larryccf · Aug 3, 2010

Beowulf_pt said:
I see you added the photo now. I do have such an antenna (used with my Holux BT GPS) but unfortunately I do not have the Samsung to see how well it improves things. The location of the socket doesn't seem very practical if you need to use the antenna often but at least it has one (my Nokia E72 doesn't).

i originally thought it was a radio antenna for use similiar to a sony ericsson phone i had - it had a plug in the rear case cover that you could remove, and the sony ericsson car holder had the male prong built in so that when you dropped the phone in, it would automtically connect the phone to the option antenna

wouldn't be difficult to do the same here but for GPS, but i didn't realize the antenna needed it's own driver

but still, maybe samsung had this in mind, ie a modified backplate, and a car holder with similiar setup

boomerbubba · Aug 3, 2010

appelflap said:
sjdean said:

Ok, I've compared the NMEA reports and the MyTracks results. Not good. Looks like the underlying NMEA information is flying off the handle too.

Not good.

Looks like the NMEA information is wrong too:

http://maps.google.com/maps/ms?ie=U...77579,-1.794891&spn=0.005594,0.01929&t=h&z=16

(Blue Squares indicate NMEA positioning, Blue Track is MyTrack)

Somewhat good news in other regards however, there are times NMEA reports show 10 satellite fixes, but the SGS still only reports on 8.

Click to expand...

Click to collapse

I only see blue squares on google maps.

Concerning the accuracy of the blue squares: is it really that bad? Most of the time it is tracking you on the road. Sometimes you seem to drive through some houses and the forest but isn't that an acceptable error (I measured an error average of 22 m. there) . Ok, I see 2 very extreme jumps (did you loose you're fix there), but overall it is surely good enough for a vector based navigation program with some good prediction algorithms.

I see only blue squares, too. I do agree with sjdean that the tracking there is not good. Not only are there areas where it wanders away grossly, the track along the road is pretty far off-road in too many places. My trusty old G1 does much better, showing me not only on the road, but in the right lane.

BTW, I have corresponded with the developor of GPS NMEA Monitor (and of GPS Test) who was very helpful. He is investigating my problem running the NMEA utility on my Vibrant. Interesting that it runs for you. I assume on a Galaxy I9000?

Interestingly, he tutored me a bit on what is happening under the covers. In the Android programming interface there are indeed different programming hooks for the NMEA output and for the other GPS sensor data outputs:

To answer your question GPS Test and other GPS apps use two Android "listeners" to get GPS data, the Location listner for lat/lon, speed, Altitude etc, and the GPSStatus listener for satellite signal and position data. They do not parse the raw NMEA data. GPS NMEA Monitor uses the NMEA Listener to get the raw NMEA strings.

I wonder if these different Android listeners are being affected differently (or the same) by the Broadcom driver, which I assume is running at a lower level.

appelflap · Aug 3, 2010

boomerbubba said:
I see only blue squares, too. I do agree with sjdean that the tracking there is not good. Not only are there areas where it wanders away grossly, the track along the road is pretty far off-road in too many places. My trusty old G1 does much better, showing me not only on the road, but in the right lane.

I agree it is far from perfect, but at least the data are useable for a decent navigation program. I think that with this kind of data it passes with ease a navigation test. But it isn't useable for "offroad" tracking, far from that.

BTW, I have corresponded with the developor of GPS NMEA Monitor (and of GPS Test) who was very helpful. He is investigating my problem running the NMEA utility on my Vibrant. Interesting that it runs for you. I assume on a Galaxy I9000?

Yes, a Galaxy I900 S... I'm very curious about the cause of your problem.

Interestingly, he tutored me a bit on what is happening under the covers. In the Android programming interface there are indeed different programming hooks for the NMEA output and for the other GPS sensor data outputs:

Aha, I thought so, but now I know for sure, I'm very interested to see what the difference is between what the NMEA listener generates and the jupiter log file.

I wonder if these different Android listeners are being affected differently (or the same) by the Broadcom driver, which I assume is running at a lower level.

I'm going to write a program to do some testing with the different listeners. One thing I want to check is if the moments the GPS Status listener and the Location listener recieve their data are in sync with the nmea output (raw and from the NMEA listener). With other words if the events that triggers the listeners are correct. Also, I want to check what the best time and distance is between different GPS measurements.

larryccf · Aug 3, 2010

suspect this is going to affect samplings taken over the next couple days

http://www.engadget.com/2010/08/03/violent-sun-throws-fit-massive-blast-of-plasma-to-hit-earth-tom/

appelflap · Aug 3, 2010

larryccf said:
suspect this is going to affect samplings taken over the next couple days

http://www.engadget.com/2010/08/03/violent-sun-throws-fit-massive-blast-of-plasma-to-hit-earth-tom/

lol

Maybe the Galaxy has a GPS designed to travel through the Galaxy and will it be very accurate in these kind of circumstances. We'll see, if we still exist after tomorrow that is.

Beowulf_pt · Aug 3, 2010

appelflap said:
lol
Maybe the Galaxy has a GPS designed to travel through the Galaxy and will it be very accurate in these kind of circumstances.

Well the Samsung's GPS has already been proven to be quite effective during high-speed navigation. especially in a fairly straight line.

Problems start only when you eventually land your ship or even beam yourself down to surface and try to do some more interplanetary geocaching. That's when your cheap ass 3.99€ GPS antenna bought at Earth's Ebay might come in handy - either that or your copilot's HTC Desire.

ghosty_uk · Aug 3, 2010

Kilack said:
So the terrible track you just got, was that on the new sgs?

yes but it was only at the start, i checked the whole route (bare in mind i have never used my tracks before), and althought the start was a mess (just outside my house, then for about 750 yrds round the side roads), when i hit a main road it got considerbly more accurate, in fact it did not glich once, it even tracked round a mini roundabout, if someone can tell me how to post the data ill be more than happy to.

sjdean · Aug 3, 2010

appelflap said:
I only see blue squares on google maps.

Concerning the accuracy of the blue squares: is it really that bad? Most of the time it is tracking you on the road. Sometimes you seem to drive through some houses and the forest but isn't that an acceptable error (I measured an error average of 22 m. there) . Ok, I see 2 very extreme jumps (did you loose you're fix there), but overall it is surely good enough for a vector based navigation program with some good prediction algorithms.

Let me clarify.

Firstly we see in the NMEA logs than sometimes it does achieve ten satellite fixes, but when reported in Android in programs like GPS Test, the accuracy only ever reaches at minimum 16.4ft and never fixes onto more then eight satellites.

This gave us some hope that maybe something else was going on so that whatever was reported to Android in terms of positioning was different to raw data and we only had to rely on Samsung to get everything sorted out.

But unfortunately, the NMEA info is out too.

For those curious, I have links to both the NMEA data track, and the MyTracks log:

NMEA
http://maps.google.com/maps/ms?ie=U...=52.474338,-1.799355&spn=0.0149,0.045319&z=15

MyTracks
http://maps.google.com/maps/ms?ie=U...2.476011,-1.793132&spn=0.014899,0.045319&z=15

ghosty_uk · Aug 3, 2010

ok i have no idea how to post my my-tracks file, but from the basic under standing of maps and roads, the nmea data seems pretty spot on, aside the odd jump here or there.

what im now thinking is althought i dont know andriod tooo well (came from windows mobile), looking at it from mearly a person interested in programming (doing a degree in it atm) that it is an interface issue not a hardware issue.

i sumise it is more how the software interacts with the hardware that is af fault, as looking over this topic, and all the raw log files ect, the hardware is seeing the satalites ok, it just the software dont seem to be able to read the data the hardware is outputting correctly.

i have just put a silicone case on my SGS (dont want it to get scratched) and even with that on i get a good clear gps fix in lnb test and gps test (i paid for the plus version as i like the HUD functionality).

sjdean · Aug 3, 2010

ghosty_uk said:
ok i have no idea how to post my my-tracks file, but from the basic under standing of maps and roads, the nmea data seems pretty spot on, aside the odd jump here or there.

what im now thinking is althought i dont know andriod tooo well (came from windows mobile), looking at it from mearly a person interested in programming (doing a degree in it atm) that it is an interface issue not a hardware issue.

i sumise it is more how the software interacts with the hardware that is af fault, as looking over this topic, and all the raw log files ect, the hardware is seeing the satalites ok, it just the software dont seem to be able to read the data the hardware is outputting correctly.

i have just put a silicone case on my SGS (dont want it to get scratched) and even with that on i get a good clear gps fix in lnb test and gps test (i paid for the plus version as i like the HUD functionality).

The huge problem I see is that it is intermittent. Fine one day, bad the next three weeks. Plus the map of the NMEA data was as off as the track in MyTracks, so I don't think either are particularly great. I've seen tracks that have been absolutely spot on, and HTC Desires are perfect.

In terms of uploading your map though from MyTracks...

1) Press the menu button and go to "Tracks" to show your saved tracks.
2) Select the track you want to upload
3) When the track is loaded, press the down arrow button in the bottom right corner, and you'll see a "Send to Google" option
4) Select to upload to Google My Maps, and choose Create a New map (I recommend).

5) Once uploaded, log into maps.google.com with your google name and password.
6) Click on "My Maps" top left, and you should see the list of tracks you've uploaded. Select your track...
7) Then there is an option top right to "Link", where it gives you a link you can paste here.

Kilack · Aug 3, 2010

sjdean said:
Let me clarify.

Firstly we see in the NMEA logs than sometimes it does achieve ten satellite fixes, but when reported in Android in programs like GPS Test, the accuracy only ever reaches at minimum 16.4ft and never fixes onto more then eight satellites.

This gave us some hope that maybe something else was going on so that whatever was reported to Android in terms of positioning was different to raw data and we only had to rely on Samsung to get everything sorted out.

But unfortunately, the NMEA info is out too.

For those curious, I have links to both the NMEA data track, and the MyTracks log:

NMEA
http://maps.google.com/maps/ms?ie=U...=52.474338,-1.799355&spn=0.0149,0.045319&z=15

MyTracks
http://maps.google.com/maps/ms?ie=U...2.476011,-1.793132&spn=0.014899,0.045319&z=15

Hi,

I was all hopeful yesterday that we were onto something with the NMEA data and that it was accurate and so we were just looking at a software issue. Now you burst my bubble.
Interesting seeing the tracks from the iphone that you posted, very similar tracking there. I was reading an article the other day that was saying tomtom are selling a lot of GPS car kits for the iphone, almost one to one with the software they are selling , I can see why now hehe.

They then go on to say that the chipset is a good one and the bad data is probably caused by lack of antenna and rf interference from other parts of the phone etc as if that is all one can expect from a gps in a phone.

So why do the HTC phones do so well? do they have an antenna? (and also curious what chipset htc used?)

We are back at square one then really, could be hardware again and now likely looking at the results of the iphone using a very similar GPS chipset.
Road navigation seems mostly useable to me, sometimes it might jump a street etc, i mean its pretty shocking but still useable...

boomerbubba · Aug 3, 2010

sjdean said:
Let me clarify.

Firstly we see in the NMEA logs than sometimes it does achieve ten satellite fixes, but when reported in Android in programs like GPS Test, the accuracy only ever reaches at minimum 16.4ft and never fixes onto more then eight satellites.

This gave us some hope that maybe something else was going on so that whatever was reported to Android in terms of positioning was different to raw data and we only had to rely on Samsung to get everything sorted out.

But unfortunately, the NMEA info is out too.

For those curious, I have links to both the NMEA data track, and the MyTracks log:

NMEA
http://maps.google.com/maps/ms?ie=U...=52.474338,-1.799355&spn=0.0149,0.045319&z=15

MyTracks
http://maps.google.com/maps/ms?ie=U...2.476011,-1.793132&spn=0.014899,0.045319&z=15

Yes, I see that. :-(

The same sort of interpolation algorithms seem to be in play in your examples. Do I presume correctly that the places with the gross errors -- the intersections of Queens Rd & Meadway, and of E Meadway & Tile Cross -- were places where traffic was stopped? That would fit the pattern of my own experience on the U.S. Vibrant using My Track. Rapid driving seems to enable some buffering, interpolating, etc., and that post-processing creates a smoothed track along straight or gently curving roads decently while in motion. But standing still or walking, the coordinates go haywire easily.

Is this sort of post-processing of raw data going on within the Broadcom chip or its driver? Or is it something Samsung is doing in its own firmware? Both would seem to be operating at a level below the Android programming interface.

The fact the the NMEA output reflects these interpolation artifacts is fishy. Might that not be a property of the Broadcom chip itself?

sjdean · Aug 3, 2010

I think HTC Desire has Qualcomm MSM7201A gpsOne

jmreed97 · Aug 3, 2010

Kilack said:
Hi,

I was all hopeful yesterday that we were onto something with the NMEA data and that it was accurate and so we were just looking at a software issue. Now you burst my bubble.
Interesting seeing the tracks from the iphone that you posted, very similar tracking there. I was reading an article the other day that was saying tomtom are selling a lot of GPS car kits for the iphone, almost one to one with the software they are selling , I can see why now hehe.

They then go on to say that the chipset is a good one and the bad data is probably caused by lack of antenna and rf interference from other parts of the phone etc as if that is all one can expect from a gps in a phone.

So why do the HTC phones do so well? do they have an antenna? (and also curious what chipset htc used?)

We are back at square one then really, could be hardware again and now likely looking at the results of the iphone using the same chipset.
Road navigation seems mostly useable to me, sometimes it might jump a street etc, i mean its pretty shocking but still useable...

Do you have a link for that article?

Kilack · Aug 3, 2010

jmreed97 said:
Do you have a link for that article?

Why yes I do!, I was actually googling "android tomtom" when I found it, was checking if there was any news on tomtom coming over to android but they are still refusing to comment on that issue...

http://www.pocket-lint.com/news/31768/tomtom-android-ipad-versions-question

""We can't answer that question," was the reply from senior VP of marketing for TomTom America when Pocket-lint asked him whether an Android app was on the cards in the future."

"Furthermore the company also confirmed that it has seen a massive uptake in the car kit that not only cradles the iPhone in the car but also boosts the GPS signal.

"Damn near one to one sales" was the quote from Tom Murray."

Ahh maybe you were meaning the other part where I was talking about the antenna issues with the galaxy, I just read that from the link that simon posted. I think I managed to combine the two sites in my head

. I have read so many GPS articles these days, it is hard to keep track of them all.
http://www.gpspassion.com/forumsen/topic.asp?topic_id=135768

sjdean · Aug 3, 2010

boomerbubba said:
Yes, I see that. :-(

The same sort of interpolation algorithms seem to be in play in your examples. Do I presume correctly that the places with the gross errors -- the intersections of Queens Rd & Meadway, and of E Meadway & Tile Cross -- were places where traffic was stopped? That would fit the pattern of my own experience on the U.S. Vibrant using My Track. Rapid driving seems to enable some buffering, interpolating, etc., and that post-processing creates a smoothed track along straight or gently curving roads decently while in motion. But standing still or walking, the coordinates go haywire easily.

I was stationary about 30 seconds at Queens Road and Meadway. Im just reviewing the map again and I can see it was fine until just after the roundabout at Shirley Road and Olton Blvd when it appears to lose signal. Then up Yardley Road, it just starts making things up and has me driving in a cemetary.

Those points were both normal driving at around 30mph. The bit at East Meadway was also 30mph... But still good thinking there.

I can live with a few glitches. I can live with the slight divot at Tile Cross Road/East Meadway. What I don't enjoy seeing is where it thinks Im driving 10 metres to the side of the road along the entire road or cutting through a cemetery. I don't like those funny loops it puts in. Quite bizarre.

For me I think I get the most issues usually when I turn a corner.

Ironically, this is probably one of the most accurate maps I've had.

boomerbubba said:
Is this sort of post-processing of raw data going on within the Broadcom chip or its driver? Or is it something Samsung is doing in its own firmware? Both would seem to be operating at a level below the Android programming interface.

The fact the the NMEA output reflects these interpolation artifacts is fishy. Might that not be a property of the Broadcom chip itself?

Clearly I think Samsung is doing something in the interpretion of accuracy and satellite fixes. But I think Broadcom are responsible for the location information as indicated in the raw NMEA data.

boomerbubba · Aug 3, 2010

Kilack said:
Ahh maybe you were meaning the other part where I was talking about the antenna issues with the galaxy, I just read that from the link that simon posted. I think I managed to combine the two sites in my head . I have read so many GPS articles these days, it is hard to keep track of them all.
http://xdaforums.com/showpost.php?p=7436848&postcount=22

I had not read that link myself. Disheartening, because of the implication that the innacuracies are inherent in the hardward design and form factor of the phone. The standard of "GPS performance about as good as the famously weak iPhone 3 but not as good as iPhone 4" was not what I was aiming for as a customer.

It would be interesting to see such benchmark track comparisons of the Samsung Galaxy S models head-to-head with the Apple products.

My own benchmark standard has been the performance of my old G1, the very first Android phone ever sold. It has excellent GPS performance, and outperformed my Samsung Vibrant over both driving and walking courses -- especially walking.

Driving Test.

Walking Test.

(Edited to fix links)

Is this as good as it gets for this Samsung design?

sjdean · Aug 3, 2010

Is this as good as it gets? We still don't know. It's kooky. It works half the time, doesn't work the rest.

Whether this means that there's a faulty chip that's burning out (probably due to the heat from the screen or when recharging), or a bad aerial, or just a programming bug in driver whether Broadcom or Samsung, we don't fully know.

Every day I investigate I come up with a bit of new information suggesting something else.

Certainly the GPS receiver can see and fix 10 satellites, so that probably isn't the problem.

Im going to try another update tomorrow. I've been noticing people move /data/data or something elsewhere.

Thing is, there's a load of files under /data/gps including control pipes and sockets which will be used for something. Im gonna move those to either the /dbdata or /system and create a symlink for /data/gps and see if it works better.

Maybe it is just a speed issue with the file system under /data.

Cya
Simon

[Discussion] GPS Problems, Lock speed, Accuracy, Fixes - The Poll

Your GPS oppinion

I use GPS a lot and it works great.

I use GPS a lot and sometimes it's Ok, other times unusable.

I use GPS a lot and it's really bad.

I don't use GPS a lot, when I try - it's OK.

I don't use GPS a lot, when I try - it's really bad.

Other/See Results.

Senior Member

Inactive Recognized Developer

Senior Member

Senior Member

Inactive Recognized Developer

Senior Member

Inactive Recognized Developer

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Similar threads

Top Liked Posts