NaNs in statistical data

Hello Matt,

Using UI 8.6.0, some of my statistics are changing to NaNs when sliding my dual marker cursors around with downsampled data. I have not verified if this occurs only in downsampled data, but my settings in Preferences were for a 1 minute buffer, 1Hz reduction, and 20Hz sample rate.
I see from release notes that this may be a known issue, but wanted to present my details, and hopefully get this fixed sooner rather than later, as it makes summarizing battery charging results more difficult than just grabbing a screenshot…

Dual markers correct here.
Move right marker to the right and NaNs appear.

Hi @wired,

This does sound similar to Issue #51. The NaNs are likely due to USB sample drops on your machine, but the stats should normally ignore sample drops. If possible, could you post the Joulescope log file and the JLS with a short snippet that moving the markers closer causes the signal to go from good to NaNs?

Matt, I do not have the live data available anymore, so I pulled up the .jls file I saved. I cannot reproduce this issue when traversing the file with dual markers on the .jls file.

I believe this is the correct log file that I have attached.
joulescope_20200318_120413_51372.log (3.0 MB)

I also had an issue with Joulescope crashing when trying to export data from dual markers. I can’t remember if this log file has that crash or if it was a later one. Can data only be exported from live data, or should I be able to export a subset of a .jls file as well?

1 Like

Hi @wired - Yes, you should be able to export dual markers from JLS files. The log does contain that error, and it tells me exactly where to look. I also see the same error as in Data Recording not always working. It also has a division by zero error, which obviously should not happen. I will investigate.

Thank you for the log file!

Hi @wired

I have fixed this issue in pyjoulescope 23b91fd. The fix will be in the next UI release.

Thanks for the log file and your time helping to trace down this issue!