Compass calibration

[Zarhan]

How am I supposed to calibrate the compass? Google maps instructs me to do this figure-eight track, but the accuracy never improves. I've also tried rotating the phone along all of it's axises individually (720°), but the accuracy as reported by google maps is always "low". Is there some trick to this?

[spook]

I might be wrong, but I think you are never going to get better than low accuracy. I think the magnets that hold the device together are strong enough that they interfere with the sensitivity of the compass. Having said that, I have tested the compass and mine is pretty spot on despite it complaining of low accuracy.

[Daniel W]

How am I supposed to calibrate the compass?

Compass issues came up as an off-topic discussion on page two of this thread on Firmware Update:www.oesf.org/index.php?showtopic=3604. I reported my findings in [a href=\'index.php?act=findpost&pid=294415\']this[/a] and [a href=\'index.php?act=findpost&pid=294431\']this[/a] post.

Since I haven't found a way to hold the Cosmo such that I can safely and conveniently do the "figure 8", I calibrate mine by rotating it two full turns along each axis.

A safe way is to put the Cosmo on a (non-magnetic) surface, grip it where the screen meets the keyboard on both sides and turn it upside down four times such that two full turns are completed. Then grasp it by the frame in front of space bar and the top middle above the screen, and turn it upside down two full turns that way. Finally, hold it by the keyboard and turn the Cosmo such that the keyboard is facing away from you, and repeat that motion for two full turns. All of this can be done in the air too, but be very careful, not to drop the Cosmo.

After calibrating, when I hold my Cosmo with the screen level (parallel to the ground) and turn the top of the device (the long edge of the screen facing away from the keyboard) towards north, the compass apps I've tried - for example "Compass Steel 3D" by "SimplyWerks" - finds north quite accurately, that is, the compass points to the north-facing top of the device.

But after that things starts getting weird. As I turn away from north, the compass doesn't keep pointing north, but, as detailed in the posts linked above, the further away from north I get, the bigger the error gets, and, at least on the apps I've tried, the compass seems to get progressively slower, until, when pointing south, it's really slow and points 180 degrees wrong. It seems to be repeatable, so it's not just a random error, but, it seems, something wrong with the compass hardware driver.

So, to summarize, I can calibrate the compass and find north, but that seems to be all that works with the current system software. Perhaps I should report this to Planet, unless it's been done already, but, lacking a real bug tracker, there's no way to really know, unless someone explicity says they have reported it, and, as far as I know, that hasn't happened yet, not at least here on OESF.

[spook]

I had to rack my brain - but here is where I read about the compass being disabled on the Gemini: https://support.planetcom.co.uk/index.php/Troubleshooting:

"Despite the compass sensor being present on the unit, the compass functionality has been removed from the firmware because of interference with the magnets used to keep the unit closed and make the device altogether thinner by pressing the keys down."

It seems like the compass has been enabled on the Cosmo, but could the strange results be due to this same issue? Strangely enough last time I tested my Cosmo compass was in the countryside a couple of weeks ago. The app said "Compass accuracy low" but it was actually working OK. I compared it to my other Android phone and both were showing all directions accurately. Now, I'm back in the city and the compass is all but useless - even after calibrating like Daniel suggested. It won't even point north correctly.

I have very little knowledge and experiences with compasses and magnetic fields, but is it possible that the magnets on the Gemini and Cosmo are quite strong, thus make it difficult for the device to find magnetic north? In the Gemini, they decided to simply disable the compass. But could it be possible that in the Cosmo, they adjusted the compass calibration to compensate for the magnets? Thus it is able to detect magnetic north, but the field is very weak due to the magnets.

I have no idea how if this is right, but if I compare it to - say - audible noise and sound waves. Let's pretend the speakers of the device are constantly emitting a loud sound. This would cause the microphone to have trouble picking up the sounds of voices for voice recognition. So maybe they tweak the device to "filter out" the specific sound from the speakers. This would enable the device to more accurately interpret the voices it hears. But the sound is still there - saturating the airwaves. So it would be ok if the voice was the only other sound or if it was very loud. But if there were also lots of other sounds as well - like a noisy room - then the device would have a lot more trouble "hearing" the voice than a device that wasn't already making its own sound... I may not have explained that very well...

So comparing this to the compass: in my case in the country side - where there might not be a lot of other things giving off a magnetic field. Even though the "accuracy" was low (because the device has to compensate for it's own magnets) there was enough magnetic "silence" to be able to "hear" the earth's magnetic field and point accurately in every direction. However, back in the city, there is too much "noise".

Sorry for my obviously completely non-technical jargon and also if this guess is completely off and that's not haw magnetic fields work! But to me, if this is an accurate guess it would explain the compasses behaviour...