Timestamps incorrect

We have been using the MetamotionC sensors for a few months. In the last 2 days, we have been having trouble with the data downloading very slowly, disconnecting frequently, and also timestamps aren't accurate. We can't use the data we've been downloading from the sensor because we can't trust that it's accurate. How can we ensure the timestamp will be correct moving forward?

Comments

  • Would you mind sharing data from the sensor (showing the bad timestamps), your setup steps (how you program it/code or metabase app setup) and also the device's diagnostic results?

  • Thanks for the quick response. I have attached a screen shot of the sensor diagnostic results from the sensor I just downloaded. I just checked the file of that report and it looks like it's just not logging many hours also. The file was started last night at 7:30ish and I downloaded it this morning around 8am and it only has data for 5:08am-6:33am.

    The file I attached is screen shots from one of the previous files with strange timestamps. The data started logging at about 7:30 on 8/3/21 and was downloaded around 7am on 8/4/21. It's saying there is data from 7/31/21 to start the file, then it jumps to 7/28/21 and then ends up jumping to 8/4/21 and records time points in the future.

    We have also used multiple devices to download data. We have used an iPad mini (5th generation) running software version is 14.7.1. I have used my phone, also, which is an iPhone 12 running software version 14.6. The screen shots I have attached were from a file downloaded to my phone.

    Please let me know what other info you need from me. Thanks!

  • I have done more troubleshooting and there really doesn't seem to be a pattern to this problem. I actually just downloaded another file from a sensor that was started at 8am and downloaded at 1:55pm. The data put the timestamps as 10:42-13:56. When I check the number of recordings, there is enough for 5 hours and 55 minutes of recording, which is exactly 8am-1:55pm.

  • Hey @jckulp , we think it's a really simple issue in the CPP library when the timestamp gets created and somehow there's a missing bit or some data getting dropped. I will look into it.

  • That would be great. It's really messing up our data right now. There seems to be no consistency to the problem. For example, all weekend, data downloaded correctly. On Monday morning, when I downloaded the data, it skipped recording data for the first 3 hours. Then, it worked correctly for the rest of yesterday and today, but the data file I downloaded at 2:30pm has data from an incorrect date and is missing data from a random hour in the middle of the collection window.

  • We are having the exact same issue with (Timestamps incorrect) with metamotionrl. We were struggling with this issue for the last two months now. We would say, this works 70% of the time without any issues.

    We understand there is no RTC in these devices but it's a high accuracy crystal. Timestamps come from the future or from the past. Something going on with the timelines when the scrip started to calculate timestamps at firmware lavel. Sometimes it goes back to the UNIX 1970 (some times goes to the future like 2-3 months or 30mins). This is random. This happens when we download (iOS / Android) the data from the device.

    Looks like the JS SDK on Raspberry PI is stable. We are still testing this on Linux.

    Please fix these issues for us and we are on firmware 1.5.1. We also bought more than 60 of these devices and the data we get is useless at this stage. This is a big pain for us at this stage. Hope you guys can fix this and update the firmware ASAP.

    Many thanks in advance. :)

  • @asanga, @jckulp, can you send us data (not screenshots) as well as replicable steps of how you get to the messed up timestamp (which sensors are on, what metasensors model and so on).

Sign In or Register to comment.