Yes, this is part two of the review (part one HERE), and yes, it only covers GPS.
Why cover one narrow feature so early in the review, and why give it such a large focus? If you’re asking those questions, you’ve never been a Galaxy S owner. When I canvassed the tech forums finding out what people wanted to know about the SGSII, the question in the forefront of nearly everyone’s mind was “Is the GPS fixed?”
You see, the first Galaxy S had problems with its GPS, rather large problems as it happens. While it could get a lock in an adequate timeframe, albeit slower than nearly every GPS-enabled handset I’ve owned since the N95, it just couldn’t hold it well enough to be relied upon. As a consequence, navigation was a rather painful experience at times. This was particularly evident in areas with densely packed and winding streets – your position would jump from street to neighbouring street continuously, triggering route recalculations each time. The result? Garbage voice guidance, confusion, frustration. Trying to do GPS-based tasks like sports-tracking, or geocaching that require even more accuracy? Forget about it.
So onto the big question, does GPS work properly in the SGSII? For my part I’m going to say yes, although I see reports in various forums that not everyone is having a happy experience. It is extremely difficult to advise prospective buyers about the GPS performance given the presence of conflicting opinions, even though those reporting problems are in a relative minority. So, what I’m going to do here is talk about my experience, and also provide as much raw data as possible, and let you decide whether the SGSII GPS will suffice.
NOTES ON GPS
Firstly, let’s have a quick education session on GPS tracking. You need to know that civilian GPS units are restricted compared to their military counterparts. While a military GPS can place itself within a foot, a civilian unit’s maximum accuracy is around 3 metres. There are a few other things that have a bearing on GPS performance that you need to be aware of too – needing a clear view to the sky, atmospheric conditions, and the presence of tall buildings or trees near the unit. Indoors, on overcast days, or when surrounded by tall buildings or trees, GPS performance is degraded.
As you might have gathered from the background on the Galaxy S GPS woes, there is more to a good GPS lock than just initial lock speed – maintaining a good lock over time is also important. If anything, it’s much more important. That’s why most of the coverage around GPS so far has been woefully lacking (Engadget, I’m looking at you!). You can tell the reviewers don’t really get the concerns with the Galaxy S GPS, and you have to wonder whether they really used the GPS functions much at all in their test period. I’ve been using the GPS continuously for a week now, and while I’ll grant you this doesn’t represent a massive sample size, there hasn’t really been much variability within this period, so I feel comfortable that I have enough of a handle to make some reasonably informed commentary.
In terms of lock speed (bearing in mind that I always have A-GPS enabled, which uses cell tower position data to speed lock times, and uses a small amount of cellular data to do so), I can happily report that my first lock took only around 25 seconds. Subsequent locks have consistently clustered around the 7 second mark, sometimes a few seconds quicker if the GPS has been recently used, and only rarely a few seconds slower in adverse conditions.
Ok, so this one is for all the marbles – how well does the SGSII maintain a lock? In my experience: very well. Using GPS Test the lock accuracy tends to fluctuate a bit initially, but then settles quickly. In the video attached to the review you’ll see that accuracy tends to lie in the 4-10m range, with the line of best fit showing good accuracy, however in less GPS-adverse conditions it sits very solidly around the 4-6m mark (see here http://db.tt/Yimn0Yu, here , and here , for some more GPS test videos – their accuracy is better than the main video, and fluctuations relate to overhead bridges and trees etc). For navigation purposes this is perfectly fine, however those with specific requirements requiring accuracy consistently around the 4m mark may need to look elsewhere. To back all this up I’ve got some illustrative screenshots taken from Google’s free ‘My Tracks’ software. Furthermore, I have many full tracks available for you to peruse at your leisure. In order to view them, you need to copy the following links into Google Maps in your browser and search for them there.
(this last one is the My Tracks route from the video, taken on the same trip as the GPS Test data)
In terms of context, it’s worth noting a few things about the tracks. The phone is sitting under the windshield in its car mount, it has a good clear view to the sky in that position (you can see exactly where in the video below). Also, in their favour, is the relative lack of tall structures surrounding the routes shown. The tracks were mostly taken with an overcast sky – that just happens to be what May has served me up over here so far. There are other things you can’t know unless you were in the car with me – that corner I swung wide on because a car was parked close, that corner where I cut the apex because I’m a bad man and didn’t slow down for the corner, and so forth.
If you look at them in Google Maps in the map view, and then in satellite view, you will often notice slight discrepancy between the positions of roads. I have no idea if one of them is more accurate than the other. Furthermore, the satellite imagery is quite old. For example, those bits where it looks like I’m driving in water beside the bridge, I’m actually on the bridge extension that was completed around nine months ago, and the tracking is just fine. Newsflash: Google Maps is not perfect.
Up next is a short video taken in the car, recorded with a Galaxy S cunningly positioned besides the drivers’ headrest with a gorillapod to capture the action. I drove the route twice, and what you will see is the first trip in the main window of the video, two smaller windows, one with video of Google Navigation running during that same trip, and another showing GPS Test data over the same route on a second trip. The onscreen videos, or screencasts, were captured using the Root application ShootMe. If you’re worried that I used Root privileges to use apps like FasterFix to cheat the GPS test, let me put your mind at ease – besides having an insecure kernel to allow Rooting, my SGSII is completely stock. All I’ve done with Root privileges is restore apps with Titanium Backup (seriously, no way was I grinding through Gun Bros again to get a half decent weapon), and install ShootMe to facilitate screencasting. There is no trickery or sleight of hand going on here: I’ve not done anything to improve GPS performance.
So far my experience of the GPS has been peachy. For navigation, I feel confident to say that the SGSII is just fine. Those of you with special GPS-related interests, such as sports tracking, or geocaching, will need to take stock of everything I’ve shown and draw your own conclusions.
I’ll be interested to hear your feedback on your own SGSII GPS experience, your criticisms of my methodology, suggestions for other tests to conduct, and, well, whatever else you have to say. I’m under no illusions that my methods are perfect, or that there isn’t some other really good way to illustrate the GPS function that I’ve neglected, so don’t fume at me in silence – have your say! If people have great suggestions for additional tests I will try and do them, providing they are not unreasonably onerous, and I will take criticisms onboard for future reviews. At the very least I promise to put in more effort than this.
See you next time for a run down on media and browsing!
PS: There are two text-to-speech engines in the SGSII – Samsung TTS and Pico TTS. The Samsung TTS has a much more pleasant, less robotic, voice, but if you enable it you will lose the reading of street names in Google Navigation!