Bending the antenna breaks the connection

[Eric_Viso]

Hi Devzone,

We are using the nRF52 DK in a prototype which is under stress, and we noticed that any bending of the board breaks the BLE connection.

To easily replicate the problem, one can put something within the 2 holes in the centre of the board (like screws) and just put some stress between them, the connection breaks for a few seconds. Vibrations can also replicate the symptoms. 

I wanted to know if this is a known problem since I couldn't find any other complaint about it. Could it be batch specific?

Is there a recommended product to use an external antenna on the nRF52DK? 

For now, we are trying to minimise the mechanical stress to make sure it holds, but any other idea is welcome. 

[run_ar]

Hi,

What clock sources are you using?

[Eric_Viso]

Hi run_ar, 
We are using the default HF (XTAL) frequency and the RC in the case of the LF Clock source.
Do you think it could be what's impacted by the bending?

Eric

[run_ar]

Most likely there is something wrong with the card. Remember that you are not supposed to bend these. What kind of force are you applying? Have you tried multiple boards, or only one?

What ppm value are you using for the LFRC? any chance you have a ble sniffer trace?

[Eric_Viso]

Hey,

We are not bending the board on purpose, and the force is applied on hard piece of plastic on which a PCB is screwed, and the DK is plugged to the PCB. So it's not direct, but still enough to break the connection. We've tried 3 different sets of PCB/DK/plastic, and it's the same behavior.

I can give more detail in a private chat if necessary ^^

The LFRC is using the default value, +/- 250 ppm. 

No BLE sniffer. If it helps, we tried data transfer through USB cable and it works well and never breaks. 

Over BLE, there were a few data length issues, but they've been solved. And it's still breaking over stress.

Any external antenna option?

[run_ar]

The LFRC should be set to 500ppm. Could you try that and see if it improves. If not, it would be interesting to know if it works if you use the external 32kHz crystal on the DK instead of the LFRC. I do not think it is related to the antenna, but bending the card might affect the internal rc, or you could have a bad connection somewhere.

Also which version of our softdevice are you using?

[Eric_Viso]

Hey Run_ar, sorry for the delay,

The LFRC seems to always be at the best ppm it can be, which in the documentation seems to be +/- 250: https://infocenter.nordicsemi.com/index.jsp?topic=%2Fcom.nordic.infocenter.nrf52832.ps.v1.1%2Fclock.html

I agree the bending might be affecting the RC clock since passive components value can change with stretching ... I would discard bad connection since it happens with every DK board we have (tried 5 so far).

We can't use the external 32kHz for our application because the pin 0.01 is used to read a sensor =/

Softdevice S132 v5.

We noticed another thing, the BLE connection breaks a lot more with the 2 dongles we use (one PCB antenna and one ceramic antenna) than with another DK plugged in the computer. Could it be that a DK talking to another DK is more robust because they have the exact same antenna? 

[run_ar]

No, I don't think it matters if they have the same antenna or not as long as it is properly tuned. You should notice that the recommended PPM value for the LFRC is 500 ppm due to errata 146. I would also recommend that you update to S132 v6 as this has some improvements for LFRC in case there is unexpected noise, which could happen due to mechanical stress.