Database Info

Tracks from Captivate Compared To Other Devices

Using JH3 with SuperRooter and lagfix.
Did play with LBSTestMode and change settings, but those settings did not change "jupiter.xml" which I believe is where the GPS stores its settings, since that is where we turn off the GPS Log File.
Even with decent locking, tracking doesn't follow very well. Attached images show the differences between the Captivate (Yellow Pins) and Treo Pro (White/Black Squares).
Will try a few things that I've seen and post up some more results after I drive to work in the morning.
Does anyone know name of a good NMEA logging app, to log the NMEA statements directly.
Attached images are snippings from google earth, showing tracks...

Only changes made is used Epic 4Gs jupiter.xml file from this thread:
http://forum.xda-developers.com/showthread.php?t=759641
I did see some improvements in tracking, previously all recorded turns or even lane changes resulted in an overshot track by over 100 ft, that took 10+ seconds to return to active track. These are from today's test with above changes, I drive 45 minutes to work, interstate, US Hwy, and city streets so its a decent test.
(EDIT) Red is Treo Pro recording NMEA with VisualGPSce / Blue is Captivate running GPS Logger from market (/EDIT)
Cap 1 is the only truly overshot at speed turn, at about 40 MPH, but the previous turn was perfect (in the same image)
Cap 2, only appeared on interstate @ 70 mph (on there 11 miles), had about 3-4 lil bumps like this that went off track by about 100-150 ft and returned, seems to last about 6 seconds but could be enough to throw navigation program off track
Cap 3, examples of improvements
Cap 4, examples of improvements

Joe, what's the red and what's the blue tracks? That way we can compare properly

dougp.me said:
Joe, what's the red and what's the blue tracks? That way we can compare properly
Click to expand...
Click to collapse
Edited post, that kind of slipped my mind since I knew the colors, but others need to know too DUH...
If anyone wants to know I used gpsvisualizer.com to convert the NMEA and GPX file to a track/waypoint based KML. I know I could get KML from GPS Logger, but I wanted to files with same format

Good research...
For comparison purposes do you also have a tracks of the PRE-Fix??? Like to see how much the GPS actually fixes it, that's not a placebo effect?

TexUs said:
Good research...
For comparison purposes do you also have a tracks of the PRE-Fix??? Like to see how much the GPS actually fixes it, that's not a placebo effect?
Click to expand...
Click to collapse
I didn't save any tracks before laying around in the 1472365 (LBSTestMode), but I believe any changes made in test mode are only a placebo affect. The first post is before changing jupiter.xml. I'm playing on the angle that that menu/program is exactly what it says, a "Location Based Services Test Mode" for testing different settings. Changes made there are saved in system.db but not in jupiter.xml
From the fact that you have to edit jupiter.xml to disable GPS logging in JH2/JH3 means that settings are definatly being pulled from that file.
I'm actually going to change the SUPL settings in jupiter.xml to the google ones and see what happens.

The first set of images is pre-fix and the second is post-fix?

TexUs said:
The first set of images is pre-fix and the second is post-fix?
Click to expand...
Click to collapse
Yes but last night record some more data and it was back to the tracks over shooting turns and accuracy being off by up to 100 feet so not sure what's up with this thing.
Can see all the satellites with decent signal, so I doubt its a hardware issue but more likely firmware or drivers depending on where the calculations are done.
I know the devs over in the kaiser and vogue forums did work to improver gps accuracy in the kernel.. It may just end up being a wait and see since we don't have source for the samsung kernel
Sent from my SAMSUNG-SGH-I897 using XDA App

A interesting GPS Test with surprising results...

TL;DR version:
1)Start Google Navigation, pick destination, start driving. Result: Sporatic or no lock and definitely no turn by turn working.
2)Start Google Navigation, pick destination, start GPSTest, start driving. Result: Almost constant lock and turn by turn works astonishingly well.
What gives?
Alright, please read this whole post, try it out, then comment on whether there is something to this or maybe I'm just crazy (a very definitely possibility). I was playing around with the GPS today and tried the Vibrant GPS hardware fix (pulling up the contact so that the GPS antenna is making better contact with the metal back). There wasn't much of a difference, so I decided to take the back off and run the GPS without the metal backing. I found that interestingly enough I was getting a better SNR with the back off than with the back on it was 8/8 at about 25-30 with the back on (using GPSTest) while with the back off it was 8/8 with 30-35.
Now comes the really interesting bit. On my drive home today I decided to see how well I could hold a moving lock. I turned on Google Navigation, pointed to my house, and started driving. Needless to say I was not able to hold a lock for very long. However, if I left the navigation app running and started GPSTest, all of a sudden I was getting the turn by turn voice pretty accurately. I then exited GPSTest and probably 10 seconds later lost my lock again. Open GPSTest, BOOM lock back on and turn by turn working again. Am I just hitting a really lucky coincidence or is GPSTest doing something we're not aware of? Please post here if you can either a) replicate what I'm seeing or b)think I'm blowing smoke cuz you aren't seeing a lick of difference.

smoking blow/10
anyways I'll try it out

comdei said:
smoking blow/10
anyways I'll try it out
Click to expand...
Click to collapse
BTW... this is GPS Test by Chartcross Limited, NOT the one by Mike Lockwood. You're welcome to try both however.

Previous GPS threads have repeatedly documented the positive effect of moving your nav app to the background. Theories on it center around backlight, temperature, or CPU/GPU. My own myTracks tests seem to point towards CPU/GPU. Or, at least, they seemed to rule out temperature or backlight as a factor.

So did you have GPS Test in the foreground during this test? This will pretty consistently give you improved performance over having Navigation up. If you turn the screen OFF altogether, it almost works as well as a real GPS! Check out the difference by recording a few drives in MyTracks -- it's night and day, on my phone at least.
My personal, uninformed theory (guess) is that the CPU/GPU and screen during Navigation are starving the GPS subsystem of power, causing either reception issues or causing the clock to desync and lose the position of the satellites it's tracking.
Ah well, such is life.
Perception 10.2 | SpeedMod K13D | I9000XXJQ1

I noticed this as well with my old setup (phone bought in August 2010 running Cognition & JK4 radio firmware). I got one of the refurbs from the random power off fiasco and GPS seemed to be just as unreliable. I've recently flashed the JL2 radio firmware, and GPS locks on within seconds and stays locked around town - as it should. I think a lot of people's issue would be completely resolved if they played around with radio firmwares...

Ikonomi said:
So did you have GPS Test in the foreground during this test? This will pretty consistently give you improved performance over having Navigation up. If you turn the screen OFF altogether, it almost works as well as a real GPS! Check out the difference by recording a few drives in MyTracks -- it's night and day, on my phone at least.
Click to expand...
Click to collapse
Yeah, GPS Test was in the foreground, navigation in the background. I get the feeling however its more to do with the application than the hardware. I say this because having other apps in the foreground doesn't quite have the same effect for me. Is that the case for everyone else as well?

knyghtryda said:
TL;DR version:
1)Start Google Navigation, pick destination, start driving. Result: Sporatic or no lock and definitely no turn by turn working.
2)Start Google Navigation, pick destination, start GPSTest, start driving. Result: Almost constant lock and turn by turn works astonishingly well.
What gives?
Alright, please read this whole post, try it out, then comment on whether there is something to this or maybe I'm just crazy (a very definitely possibility). I was playing around with the GPS today and tried the Vibrant GPS hardware fix (pulling up the contact so that the GPS antenna is making better contact with the metal back). There wasn't much of a difference, so I decided to take the back off and run the GPS without the metal backing. I found that interestingly enough I was getting a better SNR with the back off than with the back on it was 8/8 at about 25-30 with the back on (using GPSTest) while with the back off it was 8/8 with 30-35.
Now comes the really interesting bit. On my drive home today I decided to see how well I could hold a moving lock. I turned on Google Navigation, pointed to my house, and started driving. Needless to say I was not able to hold a lock for very long. However, if I left the navigation app running and started GPSTest, all of a sudden I was getting the turn by turn voice pretty accurately. I then exited GPSTest and probably 10 seconds later lost my lock again. Open GPSTest, BOOM lock back on and turn by turn working again. Am I just hitting a really lucky coincidence or is GPSTest doing something we're not aware of? Please post here if you can either a) replicate what I'm seeing or b)think I'm blowing smoke cuz you aren't seeing a lick of difference.
Click to expand...
Click to collapse
this worked really well for me. am on andromeda 3. but was getting on locks without staying on continuously. with the approach. gps was spot on.!! a BIG Thanks

I've noticed this for some time when running the GPS Application Test in lbstestmode. Run the GPS test first and then move it to the background and launch your normal GPS program and you'll usually be better off. Needless to say, I'm doing another warranty exchange in hopes of getting one with better GPS hardware soon.

Try this:
Get a mobile hotspot (or someone who has wifi tethering enabled on their phone). Put your phone into flight mode, then turn WIFI on and connect to that mobile hotspot. Fire up navigation, and go for a drive. You'll be amazed at how accurate your GPS is.
That leads me to believe that there is an interference issue in the GSM/GPS antennas. Part of the reason why putting navigation in the background helps improve the GPS performance is because it stops refreshing the map, which means it's not trying to use the GSM radio. But if you fire up Pandora, even with Navigation in the background, the GPS will start going nuts again.
I'm pretty sure AT&T is aware of this, and I even think they have a... well, not a fix, but a way to mitigate the problem.
A friend of mine did a warranty exchange on his Captivate last Monday for a problem with his USB port. The Captivate he got in exchange has been getting much better performance with it's GPS. And since his worked just fine all week, I exchanged mine yesterday. Now, quite often after flashing a new ROM, the GPS will work fine for a day or three. So I'm not going to be convinced mine is working as well just yet. But there are some notable differences from what I've seen before... When I'm tracking my position on Google Maps, my actual position now stays within what the GPS reports as it's error, even when I'm driving around. Previously, my phone might report a 10m error, but my actual position might be 200m or more from where it thought I was. In addition to that, I'm seeing that my phone is locking into more satellites, and holding that lock even as the SNRs drop due to obstructions. And finally, my GPS works just as good while it's in the car dock, whereas before it would only intermittently get enough signal to fix a position. Overall, while my GPS still isn't as good as other phones I've had in the past, it is now actually good enough to use and trust.
I think that AT&T isn't advertising this because it's not something that they can push out over the air or through a Kies update. I'm guessing that either they modified the antennas slightly, or that they installed a new firmware to the GPS chip (at a lower level than the driver)... or maybe both.
Either way, my friend's GPS has been working great all week long, and mine appears to be doing so as well. I'll not say that it's permanently fixed yet, but I'm cautiously optimistic. We'll see how it's doing after another few days go by.
UPDATE 3/28:
Two more days of testing have passed. So far, the GPS is holding up nicely. The only issues are with the car dock. When the phone is in the car dock, the GPS will occasionally drift 5~10 meters off of my position. This doesn't really surprise me, as the electronics in the car dock are right up against the GPS antenna, and simply putting the phone in the dock will drop the reported SNRs by more than 25% - usually a lot more.
Still, even with the car dock, the GPS is perfectly usable. And without the car dock, the GPS is now exactly what it should have been from the get-go.

I've also found that if I run either gps status or gps test to get the fix, then start whatever app I need to use, the performance is much better. Recently I just use lbs test to get the fix, then proceed with whatever app i need to use. Seems to work better, as lbs test continues to work in the background

i have seen this in the past but trying to prove that it is not random has been a problem as i cant repeat the results.
also i dont thing i have ever observed the phenomenon while using control plane mode.
maybe another coincidence, i cant tell but i have been using the control plane version of da_g's fix for a couple months now in combination with the jl3 modem and have gps that many smart phones would envy. i find jl3 to have the most stable positioning and tracking, less blue circles ect. i also find agps settings to hinder tracking. i only set agps mode to standalone. not sure how supl works or if controlplane supl setting is actually doing anything because i was under the impression that supl was part of agps but it is a combination that started working for me and i am sticking to it.

GPS Again

I had a Vibrant and the GPS never worked adequately. It took 10 minutes to lock if then. I knew that some of the Vibrants had a hardware problem (a poor connection that needed to be soldered), and rather than try and fix it myself I got T-mobile to swap for a SGS4G.
Well that didn't help, the SGS4G has approximately the same problem. I presume there is no known similar hardware issue with the SGS4G, and the problem is come combination of software and hardware design.
I have tried the LBSTestMode tweaks and don't see a difference.
So far I've stuck to stock roms, currently KD1.
Is the bottom line that the USA Galaxies just have poorly implemented GPS's? Do any of the custom rom's for the SGS4G really materially improve it?

What do your settings look like at the moment in LBSTestMode?

FBis251 said:
What do your settings look like at the moment in LBSTestMode?
Click to expand...
Click to collapse
I'm using the defaults with the exceptions
application->op mode = MS Based
supl/cp setting ->server = supl.google.com
-> port = 7276
-> secure socket = off
I have also varied agps between suppl and conrol pane.

Same settings I have. You can try downloading GPS Status
https://market.android.com/details?id=com.eclipsim.gpsstatus2
Click the menu key > Tools > Manage A-GPS state and click on reset, then go into the same menu again and click download. Restart your phone, and open the app again, making sure that you're outside where you can get GPS signal. After doing this my phone was able to get a GPS lock quickly.

kmandel said:
Is the bottom line that the USA Galaxies just have poorly implemented GPS's?
Click to expand...
Click to collapse
I'm going to say that I think that this is indeed the case.
Some people will have you believe that your settings are responsible, and that you should always get a lock within a minute, or even seconds with the so-called correct settings.
The fact of the matter is that if you VERY RECENTLY locked and your phone has the correct ephemeris data on it, then your re-lock will be VERY quick. This is a hot start. With a hot start, it should relock within seconds, if not instantly. I don't have an exact amount of time that constitutes a hot start or when the ephemeris data expires... obviously it also has to do with whether you're in motion or not.
If you locked recently, but not VERY recently, it will be somewhat slower by a few minutes at LEAST. This is a warm start.
Anything longer and you'll have to revert to a cold start.
AGPS helps somewhat, but the fact is that the GPS implementation on our GS class phones just plain suck. My girlfriend has a Vibrant and experiences the same issues. I did the hardware fix for her on it, and it helped reduce the signal bouncing around significantly, but the lock times... still just suck.
The people like the person above me (no offense) who claim to get quick locks likely haven't forced their phone to do a cold start or use their GPS frequently enough that it's a warm start at worse. If you want to actually induce a cold start, I find that GPS status doesn't really reset the GPS state fully. In order to force a cold start, you go into LBSTestMode and do a reset GPS. THEN you try and get a new lock with GPS Status and see how long it takes.

kmandel said:
I'm using the defaults with the exceptions
application->op mode = MS Based
supl/cp setting ->server = supl.google.com
-> port = 7276
-> secure socket = off
I have also varied agps between suppl and conrol pane.
Click to expand...
Click to collapse
try MS assisted. it worked for me when I was on froyo.

I have both, Vibrant and Galaxy S 4G, and GPS on Galaxy S is much better than on Vibrant. It is not perfect by any means, but it is much improved.
Here are my lock times:
1. Cold start - around 21 sec
2. Warm start - around 17 sec
3. Hot start - 1 sec
These are used using the supl.google.com settings on stock KD1.

Google maps and Latitude soooo slow!

anyone else experiencing slow tracking and movment when using google maps or latitude?
my GF's Wildfire S is perfect while my Galaxy S plus is slow to get GPS if it gets it at all and incredibly slow while using maps or latitude, its like when you used to play a game on a pc and the HDD would start thrashing lagging your game like crazy.
its as if something is trying to access cpu in the background, im getting sick of this phone it was supposed to be a bit of an upgrade to my Wildfire S but all i see is the Wildfire S kick its arse even with that 600Mhz cpu compared to my 1.4Ghz cpu.
can you check please how it runs by clicking the compass icon top right while looking at the map and setting it to follow you so the map moves around you when you turn.
Thanks

Hi BackDaws,
This is a common problem it seems across the SGS+ phones as I have also been facing the same problem. This is what I have found and you may check it at your end aswell.
1. Enter the start and the end direction and the route comes up.
2. Now tap the Blue icon which states your current position and it will now come to the centre of your screen.
3. Then press the cmpass icon on the top right. Now your screen will refresh as you move but now the problem that you will face is that the whole route has been compressed in one screen and you will not be able to make out if there are frequent turn in your rout.
4. Now go the Zoom out tab that is there at the right bottom corner and Zoom it to a level where you can see the route clearly with the turns.
See if this helps... It has helped me but there are times when the refresh rate is slow because of the data transfer rate.
Also do let me know if you find any other way out
Thanks
Sunil

GPS accuracy improvements?

Does anyone else have issues with the GPS accuracy? When looking at the map on activities, it seems pretty close. The total distance it records is way off. Over 25%. At the third lap around the track it's saying I hit a mile... If I look at the map, it shows me on the track for each lap. Don't get it. Any recommendations?

Same issue here. I'm constantly seeing 33%-40% less total distance when running short laps (~400m). Have returned my watch.

I used my watch >10x without any issue.
To all who believe having problems or concerns, please keep in mind the gps accuracy is limited by its technology. Every device calculates, averages and interpolates a bit different. And even after a gps fix it may need some time ...seconds,minutes... to find more satellites until max accuracy has been reached.
In general i would not expect a very precise value at short distances like 400m ...

the problem nice spoken / beautifully spoken
without any solution

It seems to me that, because of a low GPS refresh rate, it goes wrong when you make a turn. Instead of recording a distance of A+B, it records sqrt(A^2+B^2). Makes me feel less good about myself after running
At least, this happens when I let Google Fit record my running session. It might be that GPS refresh rates are determined by hardware, in which case using another application will not improve this. If it is determined by software, other running applications may work better. Anyone any experience?