Hi
We have a strange problem with a product with a nRF52840 chip. When we do the final test of the product, Bluetooth does not work on some boards and after some investigation we have found that the HFXO (32 MHz) does not start.
We have double-checked that the schematic and layout follow all of Nordic's recommendations. We have tried changing the crystal and capacitors but nothing helps. We also tried to resolder the nRF52840 several times on several non-workings boards but it is impossible to get the 32 MHz to start.
After further investigation, we have found that to get the non-working PCBA to work, is to unsolder a nRF52840 chip from a working PCBA and solder it to a non-working PCBA, then HFXO starts and Bluetooth works.
We have had soldering problems in the past and it is of course easy to believe that we have the same problem now, but the strange thing is the measurement results we get when we measure on the crystal pins.
It is unclear to me what the design looks like in the nRF52840 chip, but on a working chip I can measure a voltage drop of about 0.64V when measuring with the multimeter (positive on GND and negative on one of the 32 MHz crystal pins). Approximately the result you get if you measure on an ESD diode.
When I measure on a non-functioning nRF52840, I get an infinite voltage drop, just like the pin has no contact with anything internal to the chip.
On some of the non-working nRF52840s, there is infinite voltage drop on one of the crystal pins, and on some it is on both pins.
On working chips, I can always get approx. 0.64V voltage drop by measuring in this way.
Can someone explain what is going on, it seems to me that the crystal pins are not making contact with anything internal to the chip. But how can this have happened?
