Wesley's Tool-Box

There has been some talk about the sensor (accelerometer, gyroscope, and compass) readings of iPhone 5S being inaccurate, and since those are handled by M7, some are blaming the chip for the problem. I can't verify this claim, but I have multiple iPhones, so I can tell you if iPhone 5S strays far from the norm.

First of all, the compass seems to be doing fine. As you can see here, 5S is getting the same reading as 5, and not far from 4S's. As I moved these devices around, the readings from 4S, 5, and 5S were basically within 5 degrees of each other. 3GS was the actual standout, giving noticeably different readings from the rest of the devices. This could be due to the device's old age.

I should note that putting two phones close to each other, as some people have done to show the differences, is not a good idea. The phones affect each other's magnetic readings from the electromagnetic noise that they generate, making both units' compass deviate. It's akin to putting a magnet next to a real compass. This is why I gave the phones a bit of room. I would disregard any reports of compass error that show the photos of phones being right next to each other.

But what about the measurement of tilt as detected by gyroscopes and accelerometers? Let's take a look at those, too.

If the phones are put in the vertical orientation, the inclinometer included as part of the iOS 7 compass app shows the same reading for all three iPhones. It seems correct visually, as well.


But when they are laid flat on the table, something's different. iPhone 5S shows a slightly different amount of tilt, as well as its general direction, compared to others. This is reinforced by holding the phone horizontally. Typically, the phone reports about two to three degrees of tilt when there's supposed to be none. It's a small amount, but it's visually noticeable (humans are that sensitive). The other phones generally are accurate to within a degree.

This seems to be the central point of the issue with iPhone 5S's sensors - it's not reading the horizontal tilt as accurately as it should. This would affect a lot of games that sense the tilt to operate, as the player would drift to a side unintentionally, unless the game provides a separate calibration. Hopefully, Apple would be aware of this issue and issue a patch soon.

Update: Patch has indeed been released in the form of iOS 7.0.3. I've written a follow-up on this.