Hi,
We are using the NRF51822-CEAA flip chip BGA package in a custom board we recently received back from the manufacturing and assembly. Unfortunately, we are running into problems where the chip performs inconsistently and can even "degrade", becoming no longer operational over a period of time. We have observed this behavior in all of the 7 tested boards thus far. Our assembly house has verified all assembled chips with x-ray inspection. We suspect either the chip's optical susceptibility, thermal susceptibility, or some other issue of which we are unaware is the cause, but we need help from you at Nordic to drill down to the root cause.
Our design uses the 1.8V ceaa reference supply . We supply 1.8V to VDD, AVDD_nRF, and DEC2 pins, with an accurate 1.8V LDO (TPS71518) coming after a 3V CR coin battery. We disable the internal DCDC through the firmware.
-
Let's start from Board A. We took out board A from the box, and powered it up. It cannot be programmed. We thought the problem comes from the light sensitivity of ceaa BGA package as we had not yet coated the sides, and the chip may be killed because of exposure to light. Please notice that all the boards may already be exposed to light during assembly. In this case, what's the constraints on the environment brightness we should obey? From this point forward, we protect all the boards using liquid electrical tape (is there a recommended optical shielding material?).
-
For all other PCB boards we have, we move them to a dark room (with very dim light), and put liquid electrical insolation glue on the Nordic chips' top and side surfaces, before we do anything else. Then we tried flashing board B. And it didn't work. We guess the reason is the 1.8V power supply, so we moved to board C.
-
For board C, we applied 3V directly to the VDD and AVDD_nRF, and broke the connection between DEC2 and the external power supply, DEC2 is only connected to a bypass cap. Then I was able to flash the chip and run the function we wrote in firmware. I continued debugging other chips (not Nordic) on our custom PCB, where I desoldered and replaced one external memory chip on the other side of the PCB, with a 580F heat gun and very small wind air blow-in, for 5~10 minutes. Then I can't program the Nordic chip. What are the thermal cycle constraints related to the Nordic chip and could this have exceeded them?
-
For board D, I only changed the power supply to 3V as what I've done in 3), and didn't do any soldering, desoldering with heat gun. As expected, it worked on the first day. But when I continued work on the second day, the chip without explanation can no longer be programmed any more.
For all the above failures in programming, the error information displayed in JLink commander is "Could not measure total IR len. TDO is constant high."
We feel like too many possible reasons causing the failure, and they are entangled together. till now, we are still not sure of the exact source of failure. We really need to know every possible issue that's going to kill the board, during the debugging and assembly of Nordic ceaa chip. Lastly, are we able to recover them if they have reached this state where they cannot be programmed and give the above j-link error?
Thanks, Jamie
