How to properly test active GPS power consumption

[Stanovoy]

I noticed that a lot of you posted your battery test results, especially for the 2600mah Seido battery.

Many of you stated that you had "GPS on" during testing.

The problem with the method used

The problem is having the GPS "on" doesn't actually mean the GPS is consuming any power. GPS will not start looking for signals unless a GPS enabled application is running. This is drastically different from other things such as 3G, WiFi and Bluetooth. Those three constantly try to seek out signals, and thus consume battery power even when they are not used.

The recommended method and sample test results

The recommended method to test the effect of GPS on battery life would be to load a program such as GPSTracker and keep it running as a background process.

To demonstrate the difference, I have conducted a simple experiment. Under both test trials the following were held constant:

1. 3G off
2. WiFi off
3. Bluetooth off
4. Automatic sync off
5. Night Clock was loaded for the entirety of the test (thus the screen was just the clock)
6. Watt was loaded for the entirety of the test
7. No other application was loaded for the entirety of the test
8. The battery used was the stock one
9. The battery started at 100%
10. The test was conducted indoors without a clear view of the sky

Test #1: GPS on without loading a GPS enabled application
5 hours later: 92% battery remaining

Test #2: GPS on with GPSTracker running in the background
5 hours later: 39% battery remaining

Difference: 53 percentage point difference. The active GPS drained the battery almost 8 times faster than the GPS that was on but inactive.

Why is this important

It is important because the G1 takes a long time to acquire GPS signal from a cold start, i.e., from a state where no GPS enabled application was running. This can take up to 10 minutes.

In order for many applications, including ParkMark and Maps, to work instantaneously, GPS must be active at all times. This way, almost immediately after the application is loaded, an accurate position can be determined. Standing around the car for 10 minutes in the middle of a freezing night in a large city is not fun.

[Pinesal]

Agreed. The GPS turning itself off is the reason it works so poorly. It it were on all the time it would work flawlessly. But it drains more battery. Quite the pickle.

[Stanovoy]

I just tested this. Power-cycled my phone, leant over to the window and timed how long Google Maps took to locate me. Took 18 seconds. That's better than my TomTom, NavMan or N95 can warm up from cold... Are you sure that the G1 is so slow to acquire lock?

The speed of GPS acquisition depends on a number of factors. Two of the most important factors are (1) whether there is a clear view of the sky and (2) the density of interfering structures around.

Obviously, if you are in a location where you have a wide angle of view of the sky and the sky is free of clouds, the acquisition should be almost instantaneous. The wider the angle, the more satellites the G1 will likely "see." The clear the sky, the stronger each beacon signal will be when it reaches the G1.

Tall buildings in close proximities will not only decrease the angle of view and interfere with the signals, but will also cause significant noise. The noise comes from the signals bounding off the outside of the buildings before reaching the G1. This would cause multiple and often conflicting signals. GPS receivers can correct many of the conflicts. But when the buildings are close by and in close proximity to each other, the rebounding may be so significant that by the time it reaches the G1, the rebounded signal might be half a second after the direct signal. This cannot be corrected.

So, to answer your question, it is entirely possible for the G1 to acquire signal within a very short period of time. But for people who are in major metropolitan areas, it would take a lot longer on average. In DC, at least I can eventually get a location lock. There are places in NYC (e.g., downtown Manhattan) that a lock is nearly impossible to obtain regardless of how long the user waits.

A cold start would significantly increase this time because the G1 has to acquire all 3 satellites before even attempting to triangulate. A warm start, on the other hand, should already have at least 1 or 2 satellites locked. The G1 would only have to acquire the remaining balance before triangulating.

[Jorsher]

Interesting I honestly haven't used the GPS too much. I don't like leaving it on since so many apps use it when I don't need it to be used. When I wouldn't mind it being on, I don't feel like bothering with switching it on.

App idea - let me selectively choose what apps will start the GPS automatically, and turn it back off when they lose focus for more than 15-30 seconds