Friday, May 29, 2020

Can I trust my phone?

Recently, I was checking some of my past measurements and I found the latitude and longitude I recorded at The Bluffs to be off by about 130 miles. (I checked the GPS recording from the Physics Toolbox Pro with the latitude and longitude from Google Maps.) That's a tiny error in respect to the circumference of the earth. I still calculated that to within 1.01% on the same hike. But it would be pretty bad if I were telling someone where to meet me!

So I wondered how my other phone sensors perform. After all, the inside of a smartphone is a noisy place with so much electronics packed into such a tiny space building up heat, and so much energy swirling around in my urban setting. I decided to check my magnetometer so I set my phone in a quiet place (in my bedroom) using the Science Journal to record magnetic fields and left it for about an hour.

Here's the recording.

Not a lot of detail here but there are a few things that jump out at me. The first, oh, 23 minutes are rather "fuzzy" and then the tracing settles down into a nice flat line. After that, there are a few spikes somewhat larger than the earlier fuzz; then, at the end (maybe give minutes), the tracing drops and suddenly jumps up).

I don't see any drift. In other words, the tracing doesn't seem to drift up or down over time.

I am going to guess that the early fuzziness is internal noise as my magnetometer adapts to internal noise in the phone. Is the noise enough to worry about? Well, all instruments have errors, even manual tools like rulers and scales react to environmental noises like changes in heat and drafts. To check how much the tracing moved around, I saved the tracing as a comma separated values (cvs) file and loaded it into my DANSYS statistical spreadsheet. All I needed to check was the range of values in a section of the tracing.

The range here is 0.0165 μT. The average field strength is 52 μT (these figures are x10). The standard deviation is 0.028 μT. That last figure indicates that 66% of the samples will deviate from the average of 52 μT by only 0.028 μT, and 98% will deviate by 0.056 μT. I can live with that!.

The 52 μT background field strength is very close to what I found and reported in the Earth Specs blog. That's pretty close to the normal field strength of the Earth's magnetic field (25 to 65 μT).

By the way, background magnetism isn't constant over the Earth's surface. Dense rock and metal content in the Earth's crust can concentrate the magnetic field in certain areas much like iron cores in transformers and electromagnets can concentrate the magnetic field produced by an electric coil. This fact can be used by prospectors to find ore deposits and the magnetometer has long been in the prospector's tool chest. This close to the Front Range of the Rocky Mountains, it doesn't surprise me that the magnetic field strength is toward the upper range of Earth's background magnetism.

As for the spikes in the last half of the tracing, those do not surprise me either. The air system clicked on a couple of times while I was recording and electric relays and motors can kick out some heavy fields.

You always have to be concerned about the accuracy and reliability of your instruments. Different phones will have different sensors and driver circuits. You might want to check out your sensors. NASA has a cool book called "A Guide to Smartphone Sensor." It's a free download and you can get it here:


Lillian said...

Wolf, thanks for a good post on knowing the limits of your tools and data.

Concerning my phone’s ability to do accurate GPS locations, in the past I have noticed my location point wander around the map on my phone. I assume this relates to the phone having trouble picking up sattelite signals. I have always assumed that if I see the point land where I know I am, then the lat/long would be correct. And I have assumed that if I pull a lat/long off of Google Maps, it will be correct.

GPS accuracy is something I wish I had studied more. It would be cool to be able to map trees and perennials for a landscape plan using GPS. I know that would require survey-grade equipment, and I wonder if technology has pushed the price down.

Wolf VanZandt said...

I keep hearing that the new generation of GPS will push the accuracy of positioning to 2 meters. I just wonder when things are going to start crashing into each other up there, what with all the automobiles and beer cans. Well, I guess there's only one cat out there orbiting the sun but space junk is a real concern to astronomers, astronauts, and the space industry.