phyphox Forums
NaN values in GPS location - Printable Version

+- phyphox Forums (
+-- Forum: App-Feedback (
+--- Forum: Bugs and Problems (
+--- Thread: NaN values in GPS location (/showthread.php?tid=1825)

NaN values in GPS location - kkun - 10-31-2023


I'm trying to use Phyphox to measure inclinations and a GPS position at the same time. I've created my own experiment with the online editor, building from existing experiments.

However, when I export the results, I get a lot of NaN values for both latitude and longitude.
The strangest thing is that it doesn't seem to be related to moments when the GPS wouldn't work (at the beginning of the measure when we're not yet getting good GPS fix, or indoors), but it's always at the end.
For example, on my last measurement, the experiment lasted 1126 seconds (according to the export file), but from 489 seconds onwards, I have NaNs instead of my GPS positions.

With the initial Location (GPS) experiment, I don't have these kind of results.

Could this be due to a wrong configuration of my experiment?

attached: the xml file of my experiment

RE: NaN values in GPS location - Jens Noriʇzsɔɥ - 11-02-2023

Hm, I can reproduce a problem with the GPS data: my .xls export has empty cells after a certain time (600 and 700 something values in two tries). I need to dig into it more deeply…

The data containers for GPS in your file are restricted to 1000 values. Putting a “0“ for size (i.e. unconstrained) does not solve the problem though.

Which phyphox version are you using and which smartphone?

RE: NaN values in GPS location - Jens Noriʇzsɔɥ - 11-08-2023

In fact, a rather easy explanation: the analysis block is run roughly each 0.5 seconds, however, there is GPS data just about every second. Accordingly, the time stamps of inclination and gps data do not coincide.

You have filled the GPS data into a column with inclination time stamps. These do not match and there are simply less GPS values. I have found that the empty cells in the .csv export are filled with "" (empty strings). These are likely interpreted as NaN.

Please find attach a modified experiment with a separate file / tab for the GPS data.