Does anyone have any views/data on acceptable magnetometer noise for cave surveying? The reason I ask is I have been investigating the magnetometer in my phone and it is capable of repeatable results and exhibits next to no observable drift (tested by taking an average of a set of readings every 10 mins over 8 hours). However it seems to me to be pretty noisy, at about +/- 1.5o.

Here's a plot of a sample taken over about 10 secs

I'm not quite sure what you're plotting - is that compass bearing derived from a fluxgate magnetometer over ten seconds - 295 samples? Anywhere near a ring main and you'll be seeing the 50Hz mains hum.

For cave surveying (anything, really), the accuracy of your measurement is more important than the precision. So as long as the noise is evenly distributed you should be ok. And of course, you typically take a series of (non correlated) measurements in succession and take the mean (average), the precision will increase with the square root of the number of samples.

Yes it's compass bearing, I was outside and about 10' from the nearest mains supply (I thought that would be far enough but maybe I'll try to get further away), the phone was on top of a stone wall.

Tried again, this time about 20' from the nearest mains, similar result (but over a longer period). Interestingly the dip in the middle of the chart is when the power saver kicked in, it recovers because I woke it up again. This would seem to suggest that it is listening to it's own internal noise (hardly surprising). Like I said the results seem to be nicely repeatable.

Mmm, yes, it certainly does look like it's listening to its own heart tick!

Give it a go would be my best suggestion. No need for any other bits and bobs (or coding), just close a flat loop with metre stick (or any other distance measurement) and see what happens when you plot the data!

Your magnetic chip is likely to have an accuracy of a 3-5 degrees so long as it is in a benign magnetic environment. Do not expect better than that. These are cheap mass production sensors that are very close to lots of other emitting stuff.

The DistoX uses a PNI sensor chip with three orthogonal coils for sensing the magnetic field in 3D. Capable of 0.5degree subject to a good calibration and using it sensibly.

The DistoX2 under development uses a single Honeywell 3D magnetic sensor chip which should give better than 1 degree.

You could try to map your phone mag sensor by taking readings through a full circle and comparing against truth to determine a correction map. Also try this on a full battery charge and when almost flat to see any direction variation. Now throw in temperature variations, tilt, power modes, cellular signal variations, touch screen actions and any Bluetooth use to really get the full picture ! Talking of which, does it vary when you take a picture and with flash ?

Well thanks, but the question was about are typical noise figures - do you have any for DistoX or X2???

My hunch is that repeatability is the important thing, i.e. as long as when I point it in a particular direction I always get the same reading then the rest is down to correcting that reading into a bearing.

BTW, I expect there will be little Cellular signal variation underground and the phone doesn't have a flash.

Repeatability is dependent on all the things I mentioned.
Can you guarantee exactly the same observing conditions ?
There may not be a cellular signal underground but your mobile will still be trying to find one.
Also is your mobile correcting the magnetic signal by using an inbuilt gyro ?

As to DistoX repeatability, you can take a look at the PNI 11096 chip specifications if you can find them, this is now an obsolete component.

The DistoX2 prototype uses the Honeywell HMC5883.

Up to you how far you want to go to characterise your mobile magnetic compass.
Far more interesting to look at the latest gyro technology which can find true north without magnetics or can be used to supplement magnetics.