Poor GPS performance.

[bfguns]

I went to a different sprint store and got it swapped out. Several things i've already noticed - GPS picked up right away. the power button is more responsive, the shortcut buttons work better, and best of all, 4g is no longer connecting / disconnecting.

Interesting thing though is the build date on the previous phone is the same as this one.

Sprint Store on Fenton Ave in Silver Spring, MD. Don't go there unless you want to be treated like absolute crap.

What a nice way to end the summer.

 

[drizek]

I went to a different sprint store and got it swapped out. Several things i've already noticed - GPS picked up right away. the power button is more responsive, the shortcut buttons work better, and best of all, 4g is no longer connecting / disconnecting.

Interesting thing though is the build date on the previous phone is the same as this one.

Sprint Store on Fenton Ave in Silver Spring, MD. Don't go there unless you want to be treated like absolute crap.

What a nice way to end the summer.

Is te 4g issue common. I have been hba;ng the same problem, can't keep a signal for more than a couple of minutes.

Sent from my SPH-D700 using Tapatalk

 

[boomerbubba]

Thanks so much for the info:

Do you have any information about the different settings and what they do. I changed my GPS from MSBased to MSAssisted, will see if this allows it to work on stand-alone mode when radio is off.

You really are better operationally using the default, which is MS Based. You might want to experiment with Standalone just out of curiosity, to see what performance is like with the aGPS assistance turned off. (In the boonies, with no network available, this is the best you've got. But normally, better to take advantage of the network, which is what aGPS is for.)

The last option, MS Assisted, was really designed several years ago for dumb handheld devices with simple GPS receivers but poor processing power. Instead of calculating the final position on the phone, the satellite data is received by the handheld and forwarded to the land-based server, and the final answer is returned. This is certainly not needed by a modern device like the Galaxy S phone platform.

Generally, what MS Based mode does is accelerate the Time To First Fix by obtaining high-volume, quasi-time-sensitive data from the server instead of downloading it over the very slow satellite channel. Once this almanac, ephemeris and maybe clock data is received, calculating the final fix is much faster.

In any case, the final fix depends on satellite data captured by the handheld's GPS receiver.

Note that the early Galaxy S variants shipped with aGPS turned off, so they all ran effectively in standalone mode. When GPS problems were first reported, some of the first workarounds posted on the forums (erroneously called "fixes") just enabled MS Based mode. Unfortunately, that was not the end of the problems. It takes programming changes in firmware, including low-level drivers programmed by Broadcom, the supplier of the GPS chip, to fix everything.

So while this new GpsSetup menu is interesting, it probably is not the answer to the locking problems sometimes reported here. It is pretty clear that a new bug, specific to the Epic, is responsible for those issues. This problem seems to occur after a period of GPS inactivity, when the phone fails to get a lock. Apparently, that ephemeris data is being cached incorrectly, allowed to become out-of-date, or corrupted.

You can cure it by rebooting, but that is just a crude workaround. Another workaround is to install the latest version of the GPS Status utility. (The new version was just posted in Market a day or two ago.) When running GPS Status, you now can to to Settings and, with a couple of clicks empty the cached GPS data, then reload it from the Internet.

[Edit: Another user tried this workaround and said it didn't work. He still had to reboot.]

In the long run, Samsung needs to fix this and at least one other bug that has cropped up in the new Epic firmware. But the actual accuracy of the tracking seems to work pretty well, which it did not on earlier Galaxy S platforms

For a fuller discussion of GPS testing on the Epic, see this thread on another forum.

 

[Aero]

You can cure it by rebooting, but that is just a crude workaround. Another workaround is to install the latest version of the GPS Status utility. (The new version was just posted in Market a day or two ago.) When running GPS Status, you now can to to Settings and, with a couple of clicks empty the cached GPS data, then reload it from the Internet.

[Edit: Another user tried this workaround and said it didn't work. He still had to reboot.]

I can confirm what the other user noted, gps status is not clearing expired (eg 8hour old) ephemeridal data out of the cache.

 

[boomerbubba]

my GPS performance is very poor, cant get better than within 30 Meters acurracy on Google Maps where my wifes evo gets within 4 meters almost instantly. tested Full on Clear skys out doors.

What you are seeing is not the actual accuracy, but the Epic's report of its own estimated accuracy. There is definitely a bug there, newly introduced on the Epic platform. The imputed accuracy being reported is always 30.0 meters, as if that value is hard-coded into the firmware.

To test the actual accuracy, record some tracks using Google's My Tracks app, and compare the points plotted to your actual position on the road, trail, etc. I have done this, comparing it to a simultaneous benchmark on my venerable G1 with proven GPS perfomance, and the actual results were just as good. Meanwhile, during this walking test (the most demanding test of all) the G1 was reporting estimated accuracy of 2.0-4.0 meters, while the Epic was erroneously reporting estimated accuracy of 30.0 meters.

I have done similar simultaneous driving tests, where the Epic slightly outperformed the G1 in actual accuracy -- all the while with a bogus estimated accuracy of 30.0 meters.

So the 30.0-meter estimated accuracy is definitely a bug. Too bad for Samsung their miscreant programmer did not just hard-code 3.0-meter accuracy instead. Then most people would be happy.

 

[boomerbubba]

I can confirm what the other user noted, gps status is not clearing expired (eg 8hour old) ephemeridal data out of the cache.

Another good theory ruined by over-verification.

 

[Aero]

Thanks so much for the info:

Do you have any information about the different settings and what they do. I changed my GPS from MSBased to MSAssisted, will see if this allows it to work on stand-alone mode when radio is off.

no no no. The "ms" is your phone. you do not want ms assisted. this has sprint calculate your position (if it is working right) and is almost always much less accurate than GPS, aGPS or even hybrid, and often is not updated more than once an hour.

Also proper total confirmation of standalone does not involve phone radios turned off since they are not really ever off and some assistance data can be passed to phone. If that assistance data is required it would not really be standalone capable but seem that it was. Although I think epic is standalone capable, one of us has to take the phone completely and 100% certainly out of all cdma range to confirm it is true standalone. (eg, the palm treo 800w claimed to be standalone GPS but no one could get it working overseas, it turns out the GPS would not enable without a opaque network time call making the device forced aGPS and not Standalone.)

 

[Aero]

Another good theory ruined by over-verification.

Brother you maybe onto a fixable problem. Either by a user able to get to the cache in ways gps status isn't or at least if brought to Sammy's attention

 

[Aero]

anandtech's take on Epic's GPS woes:
Samsung Epic 4G Review: The Fastest Android Phone - AnandTech :: Your Source for Hardware Analysis and News

 

[AndroidOne]

no no no. The "ms" is your phone. you do not want ms assisted. this has sprint calculate your position (if it is working right) and is almost always much less accurate than GPS, aGPS or even hybrid, and often is not updated more than once an hour.

Also proper total confirmation of standalone does not involve phone radios turned off since they are not really ever off and some assistance data can be passed to phone. If that assistance data is required it would not really be standalone capable but seem that it was. Although I think epic is standalone capable, one of us has to take the phone completely and 100% certainly out of all cdma range to confirm it is true standalone. (eg, the palm treo 800w claimed to be standalone GPS but no one could get it working overseas, it turns out the GPS would not enable without a opaque network time call making the device forced aGPS and not Standalone.)

With the Treo 800w if you turned on airplane mode, reboot the phone (thus clearing the network GPS data stored on the cache) and turn on the GPS it never got any fix even inside a strong Sprint signal area, this phone (contrary to the Treo Pro) did not have stand-alone GPS functionality - this is exactly what I experience with my Epic ~ I have asked people that say they get GPS service with the radio off to try this and see if the GPS is actually capable to get the satelite data by itself with no network assistance.

If I do this with the Evo it gets a fix in a less than a minute (cold start) and is able to navigate perfectly off-line using NDrive. With the Epic, I can get a fix when the phone radios are on and once the fix is obtained, turning the radios off does not affect the GPS performance or accuracy, I can navigate with no problem. Getting the initial fix from a data off fresh reboot is not possible on my unit. I want to know if this is the way others work or is a problem with my Epic.

BTW, I returned the GPS settings to the original as the ones I selected rendered the GPS non-operational - did not try the standalone setting.

 

[Argyle]

With the Treo 800w if you turned on airplane mode, reboot the phone (thus clearing the network GPS data stored on the cache) and turn on the GPS it never got any fix even inside a strong Sprint signal area, this phone (contrary to the Treo Pro) did not have stand-alone GPS functionality - this is exactly what I experience with my Epic ~ I have asked people that say they get GPS service with the radio off to try this and see if the GPS is actually capable to get the satelite data by itself with no network assistance.

If I do this with the Evo it gets a fix in a less than a minute (cold start) and is able to navigate perfectly off-line using NDrive. With the Epic, I can get a fix when the phone radios are on and once the fix is obtained, turning the radios off does not affect the GPS performance or accuracy, I can navigate with no problem. Getting the initial fix from a data off fresh reboot is not possible on my unit. I want to know if this is the way others work or is a problem with my Epic.

BTW, I returned the GPS settings to the original as the ones I selected rendered the GPS non-operational - did not try the standalone setting.

This is exactly the behavior I am seeing, too. I tried various combinations of "Use wireless networks" unchecked, GPS in standalone mode, GPS in cold start mode...as soon as I turn on the phone radio, I seem to get a GPS lock in a matter of seconds. It's as if the airplane mode shuts off the GPS radio, too... Hopefully it's as simple as that and not a requirement that the phone radio actually connects to something...I can't think of any good way to test that here.

Bummer because I was planning to take the phone overseas on vacation in a few weeks and use the GPS in standalone mode (with airplane mode turned on)...maybe it'll work if I never turn it off?

 

[AndroidOne]

This is exactly the behavior I am seeing, too. I tried various combinations of "Use wireless networks" unchecked, GPS in standalone mode, GPS in cold start mode...as soon as I turn on the phone radio, I seem to get a GPS lock in a matter of seconds. It's as if the airplane mode shuts off the GPS radio, too... Hopefully it's as simple as that and not a requirement that the phone radio actually connects to something...I can't think of any good way to test that here.

Bummer because I was planning to take the phone overseas on vacation in a few weeks and use the GPS in standalone mode (with airplane mode turned on)...maybe it'll work if I never turn it off?

Is not that the Airplane Mode turns off the GPS radio, but rather turns off the data/cell radio not allowing the phone to download almanac/ephimeris info from the network. If the phone has this information already on the cache and has aready fixed your position, then turning off the radio makes no difference as the GPS is only using satelite info to navigate at this time.

However, if you turn on Airplane Mode and reboot the phone, the cache is cleared, with the data/cell radio off the network almanac/ephimeris data is not received and when you turn the GPS on it must then get this info directly from the satelite... that is what a standalone GPS such as those on most HTC phones and all dedicated GPS navigation units do. It takes longer to acquire that data and then fix your position, that is why with the network on it can lock very quickly as the data is downloaded very rapidly from the network. The problem with my Epic is that it does not seem able to download the almanac info directly from the satelite, therefore it can never get a fix where my Evo and even my Treo Pro (HTC built Palm branded handset) can do so and get a fix in about 30-45 seconds under similar conditions.

 

[boomerbubba]

The problem with my Epic is that it does not seem able to download the almanac info directly from the satelite, therefore it can never get a fix where my Evo and even my Treo Pro (HTC built Palm branded handset) can do so and get a fix in about 30-45 seconds under similar conditions.

Have you tried setting your Epic to Cold Start mode and trying this? I would expect that setting to force a download direct from satellite. By default the Epic is in Hot Start mode.

 

[Aero]

this phone (contrary to the Treo Pro) did not have stand-alone GPS functionality - this is exactly what I experience with my Epic ~ I have asked people that say they get GPS service with the radio off to try this and see if the GPS is actually capable to get the satellite data by itself with no network assistance.

If I do this with the Evo it gets a fix in a less than a minute (cold start) and is able to navigate perfectly off-line using NDrive. With the Epic, I can get a fix when the phone radios are on and once the fix is obtained, turning the radios off does not affect the GPS performance or accuracy, I can navigate with no problem. Getting the initial fix from a data off fresh reboot is not possible on my unit. I want to know if this is the way others work or is a problem with my Epic.

BTW, I returned the GPS settings to the original as the ones I selected rendered the GPS non-operational - did not try the standalone setting.

I am aware of the 800w standalone issue, in fact I was the first one to test the 800w overseas to determine that Palm's claims were objectively false.
800w problems were not a requirement of ephemeris and almanac population by the network but the mere exact network time, which actually is not made by a data call but a simple interrogation for network time though the same means as voice calls are made. Absent a cdma signal the GPS would never acquire a fix under perfect conditions after an infinitive length of time -- for this reason alone.
The following-on device to the 800w the Treo Pro, and excellent and excellently implemented GPS, but that "Palm" device was not made by Palm it was made by HTC and branded Palm.

That being said the 800w GPS sat RF capability was excellent. What happened was there was an intentional disabling of GPS as standalone, quite frankly most of us surmised as the entire 800w production as slated for Verizon (clearly evident form the Verizon splash screens you could see in root), and Verizon bailed and sprint bought the enitre stock at a fire sale. Verizon routinely cripples access to features for revenue schemes (politely in business called "the walled garden" model), and this was likely intentionally done to control usage of GPS.
device

On the Epi though other things are going on. I think this has inherently marginal to nominal RF performance for GPS band reception. I think they are using a scatter shot method of compensation for that. I also think their worldwide market, cdma and gsm, users paying for data connections and users who do not, has made for a buggy implementation. for one There does seem to be a problem either clearing stale ephemeris/almaanc from cache, or Epic knowing it is stale. Hence sometimes states of 5 to 10 birds in view, seen by Epic, and no location after infinite time. In such a condition of stale data a proper implementation would be, depending on network RF condtions (data connection absent or preent): a) populate this data with new data from network, b) absent network connection, pull data from a bird in open sky.

As a duplicatable test, under such failure to fix, a reset, presumably clearing cache or at least some marker of data reliability in cache, and gives you quick fix in the same place.

there is one other weird thing going on for me at least. whcih is absent a fix from birds my epics are sometimes not falling back to tower fix attempt (inside and outside).


ere are some additional resources:
sprint white paper on os's and lba's
Device OS LBS Specifics

nice sprint elementary explanation of location methods:
GPS and LBS 101

 

[Aero]

This is exactly the behavior I am seeing, too. I tried various combinations of "Use wireless networks" unchecked, GPS in standalone mode, GPS in cold start mode...as soon as I turn on the phone radio, I seem to get a GPS lock in a matter of seconds. It's as if the airplane mode shuts off the GPS radio, too... Hopefully it's as simple as that and not a requirement that the phone radio actually connects to something...I can't think of any good way to test that here.

Bummer because I was planning to take the phone overseas on vacation in a few weeks and use the GPS in standalone mode (with airplane mode turned on)...maybe it'll work if I never turn it off?

You need to boot and attempt a long test in clear view without moving the device to get valid results in airplane mode. if the devices initial settings are in another hemisphere this woudl be a factory stat and you need to wait a long time. if it an actual cold start (the MS (eg teh Epic) has been programmed to assume north america and could take about 20 minutes for an actual non agps cold start. (amanace take about four minutes and the birds repeat almanac brodcast every 15 minutes or so)

 

[AndroidOne]

Have you tried setting your Epic to Cold Start mode and trying this? I would expect that setting to force a download direct from satellite. By default the Epic is in Hot Start mode.

Not yet, that is the next setting I am going to test when I get home.

 

[AndroidOne]

On the Epi though other things are going on. I think this has inherently marginal to nominal RF performance for GPS band reception. I think they are using a scatter shot method of compensation for that. I also think their worldwide market, cdma and gsm, users paying for data connections and users who do not, has made for a buggy implementation. for one There does seem to be a problem either clearing stale ephemeris/almaanc from cache, or Epic knowing it is stale. Hence sometimes states of 5 to 10 birds in view, seen by Epic, and no location after infinite time. In such a condition of stale data a proper implementation would be, depending on network RF condtions (data connection absent or preent): a) populate this data with new data from network, b) absent network connection, pull data from a bird in open sky.

Agreed with your post completely. So far I have not been able to get my Epic to lock (or even see) any satelite while on Airplane Mode even after 30 minutes trial. My issue, asside the issue of GPS inherent lack of accuracy, is that this phone does not allow for the GPS to work as standalone regardless of the ultimate culprit for this (be it the lack of network provided time reference or almanac/ephimeris data). I see several people claiming they are able to get a GPS fix wihtout network (BTW, if I turn off the Use Wireless Network setting even with the cell/data radio on, my GPS has not been able to see or lock to ANY satelites, but I have not tested this thoroughly yet).

What I need you all to clarify is how your GPS is REALLY working to determine if this behavior is a problem with MY unit or is the way this phone is setup to work. In the current configuration the only way my GPS seems to work is with Use Wireless Network checked and data/cell radios on, otherwise is dead on the water. As I said above, I will change the setting to cold start on the hidden GPS menu and see what happens.

What I suggest is for you guys to do the following in an effort to gather some data:

1. Turn off Use Wireless Network off and try to get a fix.

2. Reboot the phone with the above setting off, and try again.

3. Turn Airplane Mode on and try to get a fix.

My Epic so far can't get any satelite info in any of these setups. I will report back once I make the change to cold start to see if there is a difference, but I doubt it... beginning to suspect my unit has additional hardware problems with the GPS itself.

EDIT: Some may ask why is all this important if the GPS "seems" to work well with network on and all... why should I care.

Imagine the following, you are out there in the proverbial boonies and suddenly need guidance for whatever reason. You whip out your trusty phone and there is no cell signal on your area. In the best of cases your GPS may be able to get a fix based on the last data downloaded to the cache and all is fine, you will be able to navigate to your destination (assuming of course you have a self contained nav program on your phone as Google/Sprint Nav won't be useful here). When you are done and need guidance back, your GPS can't fix because the data on the cache is now stale and can't get any updated data from the network... if the GPS is able to get data directly from the satelite (cold start), it will take longer to get a fix but it will work and you are good (this is the case with the Evo's GPS). If the GPS is not able to get this data by itself, navigation with the phone becomes impossible, you'll have to get directions some other way.

As far as the positioning errors with this GPS goes, as long as you are using a road navigation program that usually forces the reported position to "stick" to the closest road, you may feel the GPS positioning is quite accurate - and granted, this what most of us will use the phone nav capabilities for on a day-to-day basis. Once you change to a different program for off-road navigation, the position error (if substantial enough) may place you in the wrong place when you need accurate positioning the most.

Are these critical issues for the average user, problably not, but is always good to know the capabilities and limitations of your equipment so you can plan accordingly.

 

[kemfer]

GPS

I'm having issues with the GPS on my Epic. I usually have GPS turned off to save battery power, but when I turn it on it takes forever to find a GPS signal. Sometimes it can't find one at all. Is there anything I can do or change that can remedy this? I know Samsung was supposed to fix some of its GPS issues with the Epic but it doesn't seem to work for me.

 

[AndroidOne]

I will refer you to this thread:

http://forum.androidcentral.com/sprint-epic-4g/33087-poor-gps-performance.html

 

[Aero]

I'm having issues with the GPS on my Epic. I usually have GPS turned off to save battery power, but when I turn it on it takes forever to find a GPS signal. Sometimes it can't find one at all. Is there anything I can do or change that can remedy this? I know Samsung was supposed to fix some of its GPS issues with the Epic but it doesn't seem to work for me.

the cache is sticking leaving old ephemeris and almanac (assistance data), and Epic is not calling for new ephmeris data. I spoke with Sprint advanced device support Epic team this morning. they recommending a reset before using GPS if you get stuck with no GPS lock. This is the same thing we figured out on android forums and sda a couple days ago.

With any luck samsung will in future update add somethign to gps stack that refreshes or removes old inaccurate ephemeris and almanac. for now if stuck reboot.

If you download and install from market either "GPS status" or "GPS test" you can see symptoms of this bug: 3 to 11 birds in view, none "in use" and no position even with full view of sky and ample time to pull ephmeris and almanac off of one bird or sprints assistance servers.