thingy 52 crash

[Marty de Vries]

We recently purchased a few Thingy 52s, and we're very impressed with the simplicity!

After playing with the Thingy 52,and upgrading the FW to 2.2.0, I noticed that the device locks up after a while. This manifests itself as a blinking yellow/green LED, and it's not possible to connect to the Thingy anymore.

The only way to get back control is to turn the power off and then back on again.

In 'normal operation', the blue LED is flashing away happily, and when the device enters sleep, pressing the button on the front of device or shaking it usually works. However, this 'locked up' state appears quite often.

Any ideas?

-Marty.

[Karl Ylvisaker]

Hello,

Marty de Vries said:

Placing two thingys next to each other, and also an analog and digital indoor weather station device next to the thingys, I compared the temperature readings. The analog and digial weather station were very comparable, but the Thingys were a little high.

The temperature seems a little better today, when comparing this with a digital thermometer. I assume that the device needs to be on for a while for the readings to settle. Is this true for temperature? Another explanation could be that the when the Thingys were on for a while (1 hour), the battery got hotter.

I have not heard anyone else mention an increasing offset in temperature readings from the Thingy:52, but I suppose it is reasonable following 1 hour of operation, especially considering that you could feel a temperature increase in the battery, along with the temperature sensors close proximity to the battery.
Could you remove the PCB from the plastic casing with the battery(leave the battery connected to the PCB, but leave the battery in the casing), and repeat the measuring procedure? If it is the battery causing the offset in temperature then this should be evident by this test. The PCB is secured in the casing by the two clips.

The sensors can continue writing to the nrf52 while it is sleeping if they are using easyDMA. This will only require that the MCU wakes up when the data is written to RAM, to process it.
However, I do not think that the 2.2.0 firmware uses easyDMA for its sensors by default. Looking into the source code on github it seems that they do not use the easyDMA feature.
Are you able to see data for the time the device has been sleeping when you wake it up from sleep?

Best regards,
Karl

[Marty de Vries]

I ran a test on three thingys sitting next to each other. There was about a -/+2 degree difference between them. Removing the thingy from the casing didn't improve matters either. Perhaps the temperature sensor is a little crappy, or that there's a issue with the compensation algorithm?

[Karl Ylvisaker]

I understand, thank you for performing the test.

Did all the three thingy's had the same temperature measurement, regardless of if they were still in the casing or not, and regardless of how long the battery had been in use?
This sounds very strange - I have not heard any other complaints about offset in the temperature sensors, to my knowledge they are fairly accurate. From the datasheet of the supplier the minimum accuracy is stated to + - 0.5 C within the 15 - 40 C operating range.
Could you provide me with the production week numbers for the Thingys you have on hand? There should be a small sticker on the side of the thingy:52 box.
With this, I can go through the logs to see if there has been any other mention about inaccurate measurements earlier.

Best regards,
Karl

[Marty de Vries]

Thanks. There was about a -/+2 degree difference between them. 

batch number: 2019.42

[Karl Ylvisaker]

Thank you for the information, Marty!

I will discuss this with my colleagues to see if anyone else have heard about similar measurements.

Best regards,
Karl