Hey guys,
I have a customer who is observing a 20-25 % failure rate with his lamp product that has come off the production line. The CC2541 firmware is getting stuck in a loop waiting for the 32 MHZ oscillator to stabilize. The code is lifted directly out of TI example projects so I don't think that the software is really the issue.
A couple of details:
- The crystal they are using now is not the same crystal used in their 20 prototype boards (which worked). The crystal they previously used that functioned correctly was the CTS 403C11A32M0000. The oscillator is stuck at a DC voltage in lamps that don't work.
- He is also concerned about solder/joint quality on the thermal/analog ground pad to the PCB. Everything seems to be connected correctly but he is also looking into the specifics (stencil pattern, thickness) to see how much solder paste needs to be applied.
- A few other pieces of interesting information are that: 1) some units seem to be "on the edge" and can start or stop running over time; 2) some stuck units can be made to start by heating the crystal or CC2541 a bit with a soldering iron; 3) some stuck units seem to "get better" either by themselves or by #2 and continue to function correctly for a couple days; and 4) some units that do run have difficulty with Bluetooth connections.
I asked the customer to elaborate on what he meant by "difficulty with Bluetooth connections' since this could be a myriad of things. I also asked him to double-check on the BOM to see if there were any other discrepancies, besides the crystal, in his final product compared to the functioning prototype.
Lastly, this E2E post said that you should ensure the load capacitor on the 32 MHZ clock is 10 pF or greater. This is indeed the case with my customer so I don't think this is the cause of his issues either.
Any support that can be provided is much appreciated!
Best,
Sean