> I knew of the electronic noise issues with other devices, and had heard 
> discussions that the device produced its own problem.  However, it seems odd
> that one can’t compensate for the devices own noise?

Hi Terry,

I suspect that the iPhones and iPads actually do a pretty good job of compensating for their own noise, and the noise around them. That is what is happening every time you see the black calibration screen with the little red ball. That pops up when the device decides that there has been a significant change in the device's electromagnetic environment and it needs to take a bunch of readings to help it re-calibrate. It is quite important to do that calibration process properly if you want good results from your magnetometer. For those with iDevices, pay attention to the little grey ticks as the ball rolls around and you will see they have two lengths. I would recommend that you try and roll the ball around the circle twice: once at a shallow angle so that you only get the short ticks, and again at at least 45 degrees to get the longer ticks. This will help your device to get a good idea of it's environment because, after all, the magnetic field is a three dimensional field and often inclined.

As for Android devices, developers really have their work cut out. As has been discussed before, the problem with calibrating magnetometers is that they suffer a lot from noise and magnetic biases in the hardware around them. Device manufacturers should think a lot about calibration and where they position magnetometers in their device relative to things like speaker magnets or antennas, but I imagine they are more concerned about ticking the "has magnetometer" box than "produces great scientific readings" when a consumer is comparing devices. This is where Apple really have an advantage because they only produce one set of devices, so they can spend some time getting the positioning and calibration process right. Of course this is not always the case as we have seen with the iPhone 5S.

So trying to calibrate magnetometers becomes the app developers job, which makes it a much more difficult task. The shear amount of different devices that an Android app can end up on means that there are always going to be devices that don't work well, if at all. There are some general things you can do like mixing in readings from the gyroscope and accelerometers to help remove noise, and we do those things in FieldMove Clino. But the real problem is calibrating the readings from the magnetometer. There are methods out there, but they are often quite computationally expensive and often need tuning to a specific device. A lot of papers have been done on calibrating just one type of magnetometer for a specific task, so you can imagine that to get something that works well across a lot of different devices is going to take some time. Recently, some faster methods have been published and we have been looking at those at Midland Valley to see if we can use them in FieldMove Clino and our up coming tablet app.

I think that as we see technologies like virtual reality and wearable devices become more popular we will see a big improvement in the quality of magnetometer hardware and calibration algorithms available, because they have to work well for those technologies to take off.

Nathan


Nathan Collins
Software Engineer
________________________________

Midland Valley Exploration Ltd.
2 West Regent Street
Glasgow G2 1RW
United Kingdom

T: +44 (0) 141 332 2681
F: +44 (0) 141 332 6792
E: [log in to unmask]
W: www.mve.com