Apple’s iPhone 5s is already getting bad press for bugs and errors, with recent complaints focusing on an unreliable gyroscope/accelerometer. Some outlets are speculating the problem is tied to Apple changing the manufacturing source of its accelerometers for the 5s and 5c, but TechHive believes the compass accuracy problems extend beyond Apple’s two latest smartphones.


On Thursday we ran a series of tests, and found the Compass app in both iOS 6 and iOS 7 reported screwy results across a wide range of iPhones—from the iPhone 4 to the 5s and 5c. We we re-calibrated each phone multiple times, we and often found the results changed wildely from one test to the next.


The inconsistencies occurred in multiple locations around our office, and continued when we took our testing outside. Apple’s support docs warn that environmental factors can affect the phones’ digital compasses, which is why we tested the phones in several different locations, and made sure to re-calibrate each phone after moving it around or picking it up.


Left to right: The iPhone 5s, the iPhone 5, and the iPhone 5c. Click the photo for a close-up look at the inconsistent readings.

We first noticed that something was amiss while trying to replicate Gizmodo’s findings that the iPhone 5s has orientation sensors to the ones in the iPhone 5. Unlike Gizmodo, we didn’t find anything wrong with the leveling capabilities of the multiple iPhones we had lying around. We did, however, find that their sense of direction was completely unreliable, with readings across different phones disagreeing wildly. Worse yet, the problem persisted even after re-calibrating the phones eight different times.


Our methodology and results


To test the accuracy the various phone compasses, we purchased a $15 Suunto A-10 recreation compass and took an armful of phones out in front of our office building, away from the electromagentic interference within. Before testing each phone, we quit and restarted the compass application, calibrating whenever possible.


We ran the test at least three times each, restarting the apps between trials, testing the phones one at a time, as we noticed that having other phones nearby affected the reading of the compasses.


We set the Suunto compass to north and placed it on top of a map of San Francisco. We used the map's grid lines to line up the phones with the Suunto. We tested five types of phones: the iPhone 5s, iPhone 5c, iPhone 5, iPhone 4s, and an LG G2. All of the iPhones except the 4S were running iOS 7 (the 4s was running iOS 6).


iPhone compassimage: Michael HomnickSometimes iPhones were close, sometimes they were 20 degrees apart.

The iPhone 5s showed that it was pointing to 12 degrees when lined up with our Suunto compass set to zero degrees—true north. Restarting the app and recalibrating the compass found the same iPhone 5s was now 22 degrees off from the Suunto. Restarting the app and calibrating a third time brought the iPhone 5s the closest yet, with a reading of 10 degrees off of our conventional compass. The iPhone 5c's results were much more consistent with readings of 12, 11, and 9 degrees, respectively. We took five readings with the iPhone 4S and found the results ranging between 9 and 19 degrees off of the Suunto, averaging about 14.5 degrees wrong.


We ran the test on three different iPhone 5 models and found the worst deviation there. One model was not very accurate at all—15 degrees away from zero at best and 28 degrees at worst. That's pointing starkly northeast! Another iPhone 5 did very well, just 5 or 6 degrees from zero, while a third iPhone 5 fluctuated between a respectable 6 degrees away from north and a truly misleading 21 degrees away.


Just to give an Android phone a whirl, we replicated our tests with the LG G2 using Compass Pro from Gabenative, and found the Android phone's readings were the closest to our Suunto, just 3 to 4 degrees off.


An inaccurate compass is a problem, but one that is sometimes accurate and sometimes not is just as bad. If your iPhone tells you that you're heading east, you don't want to worry whether you happen to have a model that delivers consistent and accurate results. If you recalibrate and get results that are 20 degrees different, you don't want to wonder whether your readings where right thefirst time.


Serenity Caldwell contributed to this report




James Galbraith

James Galbraith , TechHive


James is the director of Macworld Lab.
More by James Galbraith










Subscribe to the Smartphone News Newsletter










Thank you for sharing this page.




Sorry! There was an error emailing this page