CC2650MODA: CC2650MODA Stop work in a cold room : -25°C

The datasheet lists this device to be able to run at -40C.

Can you be more specific as what you mean by "stops after 20 days"?

After 20 (approximatively) days the microcontroller stop work. Power is ok but no signal.... A reset has no effect.
If we take off the module from the cold-room, we wait half an hour with ambiant temperature and the module go on ...
We don't know why.

Can this be reproduced with a TI example on a launchpad? Are you maintaining a BLE connection, advertising, or anything specific? Are you bonded?

I didn't test with a launchpad but it's a possibility.

I don't use BLE ==> Only 6lowpan (802.15.4).We have a mesh network.

We deployed a system for reality test before production : (in our laboratory, all is ok but not in real)

On the same network we have 32 modules (sensors) connect on a rpl border router.

Our sensor is based on the CC2650MODA with a lithium battery and an i2c sensor TMP112. We send temperatur measurement every 45 seconds to the border router and store them.

All sensors works correctly except sensors who are on cold room with very low temperatur : -20°C to -25°C.

The other sensor are on cold room with positive temperatur like 5°C and works fine. Our bowe is tight (IP65).

I can see for test with sensortag or launchpad but there is a long time to have a result .....

??????

Can you narrow down cause of the failure any more?
Are you able to determine if this is failing due to the sensor or I2C bus?
Can you follow up with our hardware checklist (processors.wiki.ti.com/.../CC26xx_HW_Troubleshooting) to see if we can isolate if this is hardware vs. software issue?

Hi,

i think it's not a software issue because :

1) all 32 boards have the same software and only boards who are on the cold rooms -25°C have issue = We test change them and the issue folow the temperatur.

2) when the CC2650 fail : If we try to make a reset the board didn't go on : We have a led status blinking when the soft start : It's necessary to wait a long time before board could be restart.

Sensor issue ==> I think not because if the sensor failed CC2650 work and send a bad value but it works fine.

....

Try the same test but running from mains power (or similar) to rule out battery chemistry issues. Are there components that might not be speced to -25? Have the devices gone through many cycles of heating/freezing? Thermal expansion stress could have caused solder joint cracks which only show when it's cold. Hope you sort it out.

Hi,

Since next week i made a lot of tests.

1) I program a Launchpad with my software. The Launchpad is powered by the same kind of battery i use for the other tests.

At first the The Launchpad failed but i put TROPICOAT (silicon varnish) on the board for humidity protection. Now all works fine. ==> Soft and battery are ok.

2) I try put TROPICOAT on CC2650MODA but it's very diificult due to metal cover. For test, i took the cover off and TROPICOAT the entire board ==> All works fine.

I think that my problem is caused by the humidity which forms in the case and especially under the metal cover of the CC2650MODA. That's the reason why it's necessary to wait for a long time before system go on.

To have a system on a cold room, there is a problem with the CC2650MODA. In a cold room the temperature changes, there are peaks of defrosting, the temperature goes up for a short time regulary.

There is a little humidity on the case and below the metal cover. After a long time, there is to much humidity below the metal cover and CC2650MODA fonctionnality brakes.

To have a very good product i think it will be necessary to protect components before adding the metal cover with a silicon varnish for example.

What do you think about those tests ?