Help GPS issues?

[Andyman10]

The longer it takes the less I believe it's really just a software bug. The only thing that realistically takes months to 'fix', is trying to come up with better error-correction algorithms and other trickery to compensate for a faulty hardware design.

Sadly, I expect you are correct. All these tales of 'mine works fine' and 'mine doesn't' or 'it worked on Monday, didn't Tuesday' suggest a hardware design and/or manufacturing tolerance issue. The low and wildly changing SNR values suggest that it's due to noise coming from the other internal gubbins on the device itself.

It's actually not very surprising that different users, or even the same one in different locations, see widely varying results. GPS is VERY sensitive to local electronic noise and power management on mobile phones is very sophisticated indeed. The farther you are from a mobile phone base station the higher the power the device needs to transmit signals at to maintain contact, so power consumption and likely system electronic 'noise' varies all the time. Add that to variable processor loads and speeds on a modern mobile phone and it's pretty difficult to do an 'apples for apples' comparison between users with different firmware/apps/settings running in different locations. Heck it's even difficult with exactly the same ones.

The basic problem I suspect is one of hardware design and manufacturing tolerances. Design it too close to the edge to begin with and those little tolerances on each component, distance from base-station, processor load affecting power rail noise etc etc all add up and either it works...or it doesn't.. even though it's ostensibly the 'same phone'.

 

[Beards]

Sadly, I expect you are correct. All these tales of 'mine works fine' and 'mine doesn't' or 'it worked on Monday, didn't Tuesday' suggest a hardware design and/or manufacturing tolerance issue. The low and wildly changing SNR values suggest that it's due to noise coming from the other internal gubbins on the device itself.

It's actually not very surprising that different users, or even the same one in different locations, see widely varying results. GPS is VERY sensitive to local electronic noise and power management on mobile phones is very sophisticated indeed. The farther you are from a mobile phone base station the higher the power the device needs to transmit signals at to maintain contact, so power consumption and likely system electronic 'noise' varies all the time. Add that to variable processor loads and speeds on a modern mobile phone and it's pretty difficult to do an 'apples for apples' comparison between users with different firmware/apps/settings running in different locations. Heck it's even difficult with exactly the same ones.

The basic problem I suspect is one of hardware design and manufacturing tolerances. Design it too close to the edge to begin with and those little tolerances on each component, distance from base-station, processor load affecting power rail noise etc etc all add up and either it works...or it doesn't.. even though it's ostensibly the 'same phone'.

The problem is to do with the new Broadcom chip.
It's not defective, simply a software issue.
As said it's a new chip which controls not just the GPS but BT & Wifi as well.
The Broadcom engineers have been busy this past couple of months sorting or the problem then handed it back to Samsung.
Samsung then had to do their work but org was delayed again & again.
When it finally is released it will be with Froyo and should also have fixes for GPS, Filesystem, Wifi & Lag.

 

[compi]

The problem is to do with the new Broadcom chip.
It's not defective, simply a software issue.
As said it's a new chip which controls not just the GPS but BT & Wifi as well.
The Broadcom engineers have been busy this past couple of months sorting or the problem then handed it back to Samsung.
Samsung then had to do their work but org was delayed again & again.
When it finally is released it will be with Froyo and should also have fixes for GPS, Filesystem, Wifi & Lag.

I really hope that you are right. But being a software engineer developing large-scale commercial products day after day, I can see what an average consumer cannot about Samsung's development process. This whole GPS fiasco tells a lot about it, and if this type of fault can go thorough their QA, it means non-existent or deeply flawed QA process in the development phase. Even with the prototype phase someone should have been testing it in an urban jungle with a reference GPS receiver (making some track recordings on both and compare them later in the lab). This is not rocket science is it? If this type of test cases are missing, it means that Samsung simply does not test its product in real-life environment. I can totally believe that the GPS receiver passed absolutely fine when fed with some signals from some kind of signal generator. But I'm quite sure they did not test is in a real urban environment, which is absolutely necessary for this type of devices.

To make it short(ish): What I learned about the internal development process of Samsung from this case makes me believe that the last product I bought from them was my i9000. It's clear that they does not give adequate testing for their products and they release them way too early. They do the same with their devices as Microsoft does with it's software - but it is a lot harder to fix a bogus device sold to millions of customers than an office suite or an operating system on an internet enabled PC. As I started to avoid Microsoft products everywhere I can years ago, from now on I will try to avoid buying Samsung products and being unpaid beta tester of them.

 

[daveleather]

The problem is to do with the new Broadcom chip.
It's not defective, simply a software issue.
As said it's a new chip which controls not just the GPS but BT & Wifi as well.
The Broadcom engineers have been busy this past couple of months sorting or the problem then handed it back to Samsung.
Samsung then had to do their work but org was delayed again & again.
When it finally is released it will be with Froyo and should also have fixes for GPS, Filesystem, Wifi & Lag.

That's a reassuring update, thanks.

 

[AlexanderA]

The problem is to do with the new Broadcom chip.
It's not defective, simply a software issue.
As said it's a new chip which controls not just the GPS but BT & Wifi as well.
The Broadcom engineers have been busy this past couple of months sorting or the problem then handed it back to Samsung.
Samsung then had to do their work but org was delayed again & again.
When it finally is released it will be with Froyo and should also have fixes for GPS, Filesystem, Wifi & Lag.

I too hope this is true, but I wish there would be an official statement from Samsung that said this rather than an anonymous source. Not that I don't trust you but because I can't verify the accuracy of such a statement that originates from an unnamed source.

 

[spark001uk]

Good news and bad news from T-Mobile UK support:

POSTED 22/9/10:
"Hi there,

We are pleased to announce that there will be a software update for the samsung galaxy s at the end of the month. This will update your phone software to Android 2.2 and will be available through the Samsung Kies PC application.

Regards
Stef
Forum Team"

POSTED 25/9/10:
"Hi there,
We recently announced that there will be a software update for the Samsung Galaxy S at the end of the month, we are still looking to roll this out as soon as possible but have been told the release date may slip by up to a couple of weeks. Of course as soon as we have any more info we`ll let you know.
The update will still be made available through the Samsung Kies PC application.

Regards
Forum Team"

 

[karnka]

Of course 'upto' a couple of weeks would put us before the 20th of October date that has been suggested so I think T-Mobile are still fibbing.

 

[Andyman10]

The problem is to do with the new Broadcom chip.
It's not defective, simply a software issue.
As said it's a new chip which controls not just the GPS but BT & Wifi as well.
The Broadcom engineers have been busy this past couple of months sorting or the problem then handed it back to Samsung.
Samsung then had to do their work but org was delayed again & again.
When it finally is released it will be with Froyo and should also have fixes for GPS, Filesystem, Wifi & Lag.

That's great news if so..do you have an internal contact at Samsung or Broadcom that has verified this?

I think the Galaxy S/Andoid combo is generally a brilliant device that betters the Iphone in many areas (I have both) but reliable GPS is a must for a lot of today's really useful apps ..and it doesn't have that now.

I found something interesting the other day that I have posted in another forum but perhaps it may be useful to post it here and catch a wider audience to try it.

Leaving settings at default (standalone mode, network assistance off). Disable the GPS, put the device into flight mode and then enable GPS and use GPS Test (downloadable from the market for free) to check the speed and stability of the GPS lock, noting particularly the variation in SNR on the same satellites whilst stationary. Once it's got a good lock turn flight mode off to enable wireless again and again check via GPS test what kind of SNR and stability of lock you are getting. For me and a few others this seems to have a significant effect on performance which lasts until you actually power the phone down. SNR levels are much higher with the wireless turned off, which you might expect if it is producing interference noise, and interestingly the wild variations in SNR you normally see on 'in view/in use' satellites seem to be reduced considerably even when you turn wireless back on.

I have some theories about why this might produce the improvement I am seeing (to do with initial AGC lock and tracking) but it would be very interesting to get more feedback from others with GPS problems trying this simple test.

 

[Beards]

That's great news if so..do you have an internal contact at Samsung or Broadcom that has verified this?

I think the Galaxy S/Andoid combo is generally a brilliant device that betters the Iphone in many areas (I have both) but reliable GPS is a must for a lot of today's really useful apps ..and it doesn't have that now.

I found something interesting the other day that I have posted in another forum but perhaps it may be useful to post it here and catch a wider audience to try it.

Leaving settings at default (standalone mode, network assistance off). Disable the GPS, put the device into flight mode and then enable GPS and use GPS Test (downloadable from the market for free) to check the speed and stability of the GPS lock, noting particularly the variation in SNR on the same satellites whilst stationary. Once it's got a good lock turn flight mode off to enable wireless again and again check via GPS test what kind of SNR and stability of lock you are getting. For me and a few others this seems to have a significant effect on performance which lasts until you actually power the phone down. SNR levels are much higher with the wireless turned off, which you might expect if it is producing interference noise, and interestingly the wild variations in SNR you normally see on 'in view/in use' satellites seem to be reduced considerably even when you turn wireless back on.

I have some theories about why this might produce the improvement I am seeing (to do with initial AGC lock and tracking) but it would be very interesting to get more feedback from others with GPS problems trying this simple test.

Hi Andy,
As much as I would like to divulge my source's name they have requested to remain incognito.
I know this does not help to ascertain the factualness of the messages and I appreciate any scepticism towards the truth.

I would not in the slightest be offended if there were any members who would doubt the credibility of such messages.

Re your tip on getting the GPS to lock ~ an interesting idea as it does follow in with one of the problems found.

EDITED:
Just tried your GPS tip but found it to be completely unsuccessful.
With the phone in Airplane mode and GPS On. although it found several satellites it simply refused to lock on to any.
It wasn't until I had unchecked Airplane Mode and switched on Wifi did it then manage to lock on to a few satellites.
Looking at the track in My Tracks showed my position again like a dog chasing a rabbit over fields.

 

[Andyman10]

"As much as I would like to divulge my source's name they have requested to remain incognito.
I know this does not help to ascertain the factualness of the messages and I appreciate any scepticism towards the truth.
"

That's understandable..I don't suppose he would remain a source very long if you disclosed his name I am much cheered at the prospect of a real fix for this. Surprised though that I and others see a huge difference and you see none...the mystery deepens eh?

I should have perhaps mentioned that I'm running JM2 firmware, I have no idea if that is significant but if it's a driver issue I guess it may be.

 

[mnemon]

That's great news if so..do you have an internal contact at Samsung or Broadcom that has verified this?

Leaving settings at default (standalone mode, network assistance off). Disable the GPS, put the device into flight mode and then enable GPS and use GPS Test (downloadable from the market for free) to check the speed and stability of the GPS lock, noting particularly the variation in SNR on the same satellites whilst stationary. Once it's got a good lock turn flight mode off to enable wireless again and again check via GPS test what kind of SNR and stability of lock you are getting. For me and a few others this seems to have a significant effect on performance which lasts until you actually power the phone down. SNR levels are much higher with the wireless turned off, which you might expect if it is producing interference noise, and interestingly the wild variations in SNR you normally see on 'in view/in use' satellites seem to be reduced considerably even when you turn wireless back on.

I just have tried your solution... Went outside switched Filight mode on... tryied GPStest and i got quick lock with 7 satelites with snr around 35-40
When with Fligth mode off i am getting snr around 20-30 and far longer lock time...it is not stable either
It seems that there is some sort of interference between signals which would make me think that the problem is not with a software but hardware... Hope i am wrong and they will get it sorted

 

[AlexanderA]

Bugger. If GPS performance becomes measurably worse depending on what other pieces of hardware are enabled in the phone, that does sound like a hardware interferance issue.
That pretty much dissolves my hope of it being a software problem. It does explain why it's taking so long for a fix and the lack of any promise that the fix will really be a 100% fix. All that samsung promises is 'improved gps functionality', perhaps by having the new software disregard data from the GPS when it cant be realistically correct (traveling at the speed of sound, etc). Such measures would certainly improve the GPS but it would still be way behind a properly working GPS.

I have 4 days left to pick a new phone, but so far the wait has only given me more doubts about this issue being really fixed. A real shame. On Samsung.

 

[Beards]

If it helps.... the information I was told was that the Broadcom drivers first needed correction. Then the filesystem needed tweaking which then brought on another problem as the filesystem was (is) problematic anyhow and needed looking into first.
In other words they couldn't get a GPS update until the filesystem had first been addressed.
As time went by it became closer and closer to Froyo being released, which again would be bringing in changes.
As such all further updating went hand-in-hand with Froyo.
It's taking time and it's only right that owners and would be owners are troubled and frustrated.

I really do feel for those like Alexander who are in a time-line, wondering whether or not to risk it.
I have no idea the exact date and obviously no idea if it will all work in the end.
My best guess is it will solve quite a few problems but there are bound to be either new one's brought in or further work to be done on problems which have been worked on.

Depending on the hardware you are looking for ~ if it's already a proven platform (i.e. the Desire) then I would say go for it.
If however the hardware is not yet out (i.e. Desire HD) then I'd be tempted to get the GSG and try it out...... If it does not fit in with your requirements, return it and get the other phone of your choice.

 

[Andyman10]

Bugger. If GPS performance becomes measurably worse depending on what other pieces of hardware are enabled in the phone, that does sound like a hardware interferance issue.
That pretty much dissolves my hope of it being a software problem. It does explain why it's taking so long for a fix and the lack of any promise that the fix will really be a 100% fix. All that samsung promises is 'improved gps functionality', perhaps by having the new software disregard data from the GPS when it cant be realistically correct (traveling at the speed of sound, etc). Such measures would certainly improve the GPS but it would still be way behind a properly working GPS.

I have 4 days left to pick a new phone, but so far the wait has only given me more doubts about this issue being really fixed. A real shame. On Samsung.

Well, don't abandon all hope just yet.. Once I've done a set-up from cold with wireless off it continues to perform well (even locks indoors..which it didn't do before!) when I turn wireless on it just takes a little while to 'settle' but again works much better...as apparently it does for some other users .i.e. this one:

"Interesting observation, Andyman10. I just turned on airplane mode, went outside and ran GPS Test. Within 30 seconds I had 8 in view, 5 sats in use. at 2:02 I had 11 in view, 7 in use. My SNR on the used sats was very steady at 33 up to an observed high (on this phone) of 41. I went back and turned off airplane mode and reaccessed GPS Test and it relocated all the sats within seconds but the SNR was back down in the 24-33 range on all in use sats. Might bear further testing.
Oh, and sitting at my desk which is right next to a window, about a 14" below the sill, I normally can't lock any sats though it will usually see about 2. With the airline mode turned off I saw and locked 8/5 almost immediately. Once airline mode was turned back on the phone still sees/uses 7/3 up to 8/6. Perhaps the answer is to go airline mode, lock your sats, then turn your Captivate into a phone again?

As it seems that other people are observing exactly what I am I'll repeat my original post on another forum here.

"That's pretty much what I'm seeing now..certainly the SNR values being much better and more stable when the phone is off and it's acquiring locks..and that noticeable increase in performance continues when you turn the wireless back on. It does give me some possible ideas as to what the issue might be.
I can't seem to find the exact chip they are using in this phone to look at a datasheet or programmers guide.. seems it might be one of three..but they're all a similar architecture. One common thing you have to do with the host processor assisted type designs that are being widely used now is control, via a Java or direct API, the tracking of the satellite signal being received and the initiation of the AGC or Auto Gain Control circuits. Basically a teeny weeny signal from the satellite comes in to the GPS device at say 50 millivolts, and to bring it in optimum peak range of the decoding circuits, let's say 0 to 1.8V you apply gain or amplification to the signal until it nicely fits in this range, so you get a good look at it. If you start this gain adjustment process in the presence of noise spikes greater than the signal you want to receive then the gain is set incorrectly for the intermittent noise rather than the required signal and there isn't enough amplification to reliably digitise and quantitise the required signal.
Conversley if you set the AGC tracking whilst the noise is completely absent then it ramps up the gain and when the noise does appear intermittently at a higher level and is amplified it hits the limits (the 1.8V say) and gets clipped, causing all sorts of nasty harmonics and distortion.. and a sudden whacking change in SNR.
It's all actually a bit more complicated than that but that's the basic principle. These AGC circuits are, as the name implies, automatic and they try and track changes in signal level to dynamically adjust the gain. So they don't do this at the slightest tiny pulse they have a programmable hysteresis or delay so it keeps a decent average value. Problem is they need something generally representative in the first place to get a coarse lock before they can get a fine one and a dirty great spike while it's setting up, say from a mobile phone transmitter, could mess that up.
All mobile phone modules, even if they aren't being used for voice or data at the time, perform periodic 'local area updates' where they send a short burst of information to the cell so that the network knows whether it's still attached and still on the same cell. This sudden transmit burst also suddenly ramps up the power consumption. In a good design a high capacity capacitor (or super cap) is used to provide this short-term peak power demand and the effect on the rest of the power rail is minimal..in a bad one it causes a sudden huge increase in load across the whole power rail and often noise on the rest of the system.
So..do other users see significant average SNR changes and better locks after first tracking with wireless off (i.e. in flight mode) ? Would be very interesting if that's a common observation and would indeed point to intermittent locally generated noise and something like incorrect AGC tracking as a root cause. Might even be fixable in the driver if they change the start point/duration/sampling interval on the AGC ..or even make sure the phone isn't transmitting noise at the time it performs the initial AGC lock. Might also explain why some users get good GPS performance and some don't..they're just lucky when it sets up the AGC.
Can we get a straw poll of the efect from other users with GPS problems?"

So..it may well be possible to re-configure the AGC circuit in the driver and possibly synchronise the initial AGC lock with a period when the wireless components aren't generating large pulses... if that is the problem.

To summarise I think Samsung got the basic design wrong.. and that makes it very sensitive to fluctuating local noise.. but that doesn't mean that it can't be worked around to a suitable level to give reasonably good GPS functionality. A device packed with the amount of radiating electronic gubbins that the SGS is will never be as good as a properly designed stand-alone GPS module but it should be adequate for purpose.

 

[AlexanderA]

At the chance of everyone already being aware, AT&T has been rolling out an official update for the Captivate version of the Galaxy. It includes GPS fix and several other fixes but no Froyo yet.

Article can be found here: AT&T Samsung Captivate GPS fix | CareAce

MyTracks result maps are added. The GPS seems to function a bunch better, with tracks looking reasonably smooth although still leaving the actual road every now and then and it seems to be still not really as it should be when it comes to suddenly changing speed and/or direction.

To me it still looks like a hardware issue thats partially compensated by better error correction and filtering. However it seems like a decent effort at that, making the GPS performance probably acceptable for most purposes, which is good step in the right direction.

 

[mnemon]

Guys (and ladies)... Few minutes ago i went outside to test GPS again with a solution above... Got quick lock and SNRs around 25-35... after while i got bored and i put galaxy on the roof of the car.. SNRs jumped up and i got most of them above 40, very stable aswell...
So all we have to do is carry a slab of steal to improve our GPS

sorry for a poor quality of my english its not my first language

 

[karnka]

At the chance of everyone already being aware, AT&T has been rolling out an official update for the Captivate version of the Galaxy. It includes GPS fix and several other fixes but no Froyo yet.

Article can be found here: AT&T Samsung Captivate GPS fix | CareAce

Samsung lied / had no idea what they were actually rolling out.

Samsung Captivate gets GPS fix, other Galaxy S versions wait patiently -- Engadget

Engadget confirmed with AT&T there was no GPS fix included.

 

[Beards]

Take it from me guys...... As of yet, there is no official GPS fix release for any of the Galaxy S models.

 

[spark001uk]

Take it from me guys...... As of yet, there is no official GPS fix release for any of the Galaxy S models.

Is there any official word that a gps fix will be included with the froyo release?

 

[stoatwblr]

The problem is to do with the new Broadcom chip.

*Shudder*

Broadcom have repeatedly shown themselves to be open-source hostile. Getting drivers to run under linux (Which is what Android is) can be quite difficult as a result and getting ongoing support to OEMs is nearly impossible.

Samsung should be doing their homework better. Whatever they saved by using Broadcom silicon will have been more than wiped out by the added costs of making the parts actually work.

 

[AlexanderA]

Well the month is just about over and I have a choice to make tomorrow..
Unfortunately the promised update never came, making it hard to choose for the Galaxy which was my favorite option. I have trouble opting for the Desire though due to not-so-great camera and some other things.

Right now im tempted to preorder the Desire HD, but that too has its downsides as nobody has reviewed it properly yet so it may or may not have its own problems, and the 1230mAh battery seems pretty shy for a phone with a 4.3" screen. And ofcourse itll be a bit pricier but thats overseeable.
What also plays is that I have some doubt if I could still fall in love with the SGS after this whole debacle, even if the GPS becomes acceptable.

I still can't make up my mind so if anyone has a good argument either way it's welcome

 

[karnka]

I don't have a really good argument but if I was within any kind of return period I would have sent back my SGS by now.

I am still considering Ebaying and paying the large chunk of cash it would cost to go up to a Desire HD. Given it would be very expensive for me I am going to wait to see one in the flesh which gives Samsung another couple of weeks.

I agree the battery is slightly worrying and the screen won't be as nice as the SGS but I think the HD is a more attractive phone and I'm a little shallow

 

[Beards]

Well the month is just about over and I have a choice to make tomorrow..
Unfortunately the promised update never came, making it hard to choose for the Galaxy which was my favorite option. I have trouble opting for the Desire though due to not-so-great camera and some other things.

Right now im tempted to preorder the Desire HD, but that too has its downsides as nobody has reviewed it properly yet so it may or may not have its own problems, and the 1230mAh battery seems pretty shy for a phone with a 4.3" screen. And ofcourse itll be a bit pricier but thats overseeable.
What also plays is that I have some doubt if I could still fall in love with the SGS after this whole debacle, even if the GPS becomes acceptable.

I still can't make up my mind so if anyone has a good argument either way it's welcome

Depending on where you would be getting the Galaxy S phone from you could order the phone, see how you get on with it and if the Froyo and other fixes have not surfaced before having to return you could then return the phone as it is not fit for the purpose you purchased it for.
At least this method gives you a stop-gap and allows you to try it out first hand.

Regardless, it's a tough call which phone to go for.
The camera on the DHD does not look that good and HTC are known for not providing the best camera hardware.
The screen may be larger but it's not even Amoled. It may be bright but the colours are going to be washed out when compared to the likes of the SGS.
I watched a movie being displayed simultaneously on:-
a) My HD2 with it's 4.3" LCD screen
b) My Brother's Desire with it's 3.7" AMoled screen
c) My SGS with it's Super Amoled 4" screen

In order of worst to best:-
a) HTC HD2 ~ A lot of detail missing. Colours looked washed. Blacks were like dark greys. Pixel blotches quite strong especially in dark areas.
b) HTC Desire ~ A lot of detail still missing. Colours not as strong. Again Blacks not true black. Blotches still evident in corners.
c) Samsung Galaxy S ~ Perfect detail, like watching a Blue-ray movie. Colours vibrant. Blacks are true blacks. No pixel blotches in dark areas.

Of the three the SGS was the only one where it stood out by a mile.

I also think the Speaker is the strongest of the three.

However, if things like Notification LED's and a camera flash are important the the DHD is your choice.
P.S. Bear in mind the although the SGS does not have a camera flash, in my opinion it does not need one and I've never missed not having one in all the low light photo's I have taken and I take many, many evening shots.
Besides, not having to use a flash makes shots look much more natural.
The SGS has a very clever sensor and one of the lowest f-stops found on a camera phone which enable low light photography at higher ISO levels which in turn mean more detail and not much light required.

 

[karnka]

Mr Beards - I wish you weren't so knowledgeable about this. You're making it harder for me to give up my SGS

 

[Beards]

Mr Beards - I wish you weren't so knowledgeable about this. You're making it harder for me to give up my SGS

Thank you for the compliment..... I hope to look after all just as I would if you were on my operating table.

Deciding what to do is a tough call.
For me I just hang on to the facts of what I have seen and used with the SGS.
It's more than enough to keep my interests high... but I fully understand we are all different and one mans meat is another mans poison.