Problem with UART TX or RX lines

[Bartosz22]

I have a question regarding an issue we observed in some of our products. In a few units from our large production batch, we are experiencing problems with the UART TX or RX lines on the NRF52833.

Specifically:

• In some units, the NRF52833 does not respond to commands, preventing data transmission via UART, although we still receive data logs, so TX line doesn't work but RX works
• In other units, we can send data via UART, but we cannot receive any data. For example, we can send a reboot command, and the MCU reboots successfully, so TX works, but RX doesn't

This issue does not seem to be related to our PCB design, as the UART is connected to a pin header and we use an external FTDI for communication. Additionally, we resoldered a new NRF52833 onto a PCB with a non-working UART, and the UART started working.

My question is:

• Is it possible that the UART on the NRF52833 is being damaged during the production process, such as during soldering, while other functionalities remain unaffected?
• Could ESD selectively destroy only the UART peripheral?

This issue has been observed in only a few units (for example on four out of a production batch of 1000).

[hmolesworth]

Which FTDI model is being used? Should it happen to be a USB serial cable it is possible that it is internally wired with 5V Rx and Tx instead of 3V. I mention this because of a similar problem I saw caused by mixing up several FTDI interfaces (they are not always marked as 5V or 3V).

Quite apart from that consideration note that the FTDI interface must not remain connected to the nRF52833 when power is removed from or not be present on the nRF52833 as (unless there is an isolating circuit in place) the FTDI Tx output will power nRF52833 VDD via the schottky clamp diode and any pull-up on Rx may do the same.

All these scenarios will eventually damage the Rx or Tx pin with no way of predicting when. The only safe way to avoid this automatically is to ensure level-shifters are in place, ie more hardware on the product or a properly designed FTDI interface which monitors nRF52833 VDD.

[Bartosz22]

We use FT232.

[Bartosz22]

So is there any good practice to design PCB with NRF52833 to avoid this problems or is there some sample schematics to do it in right way ?

[ketiljo]

Sounds like the pins are damaged, either by over-voltage or ESD. This can damage a single pin only. 

[hmolesworth]

Have a look at my response to an old post; the VCC of the buffer will connect to nRF52 VDD and the buffer will isolate the FTDI Tx line output from the nRF52 input pin schottky diode, you can probably ignore the Rx line, but if using the direction would be reversed. If no pull-up on the FTDI Rx line the FTDI will spew nulls to the laptop; if using an FTDI Rx pull-up to stop that then a buffer on Rx will be required. There are also twin-rail level-shifters which have two supplies, but they are bigger.

potential-risks-if-voltage-is-applied-to-a-gpio-pin-when-the-chip-is-not-powered-vdd-0-0v