nrf52 disable debug by reset with eeprom locked

[puck]

Hello,

I have a PCB with a NRF52 powered by batteries. We need to be able to plug the batteries, then program the device and never remove the batteries. We noticed, that currently, we have a consumption of about 1.3mA more than expected, because the debug is still enabled. We are programming using nrfjprog, here is the sequence we send:

nrfjprog --recover
nrfjprog --program OURHEXFILE
nrfjprog --memwr 0x10001200 --val 21 #enable reset pin
nrfjprog --memwr 0x10001204 --val 21 #enable reset pin
nrfjprog --reset
nrfjprog --pinreset
nrfjprog --rbp ALL #locking eeprom read
nrfjprog --pinreset

After reading this question I expected the final pinreset to disable debug and to get the normal power consumption, but it is not the case.

We only managed to do it by removing eeprom lock, this way we could send "normal" reset with nrfjprog -r which actually behave as expected. All other reset (--debugreset, --pinreset and physical pin reset) are not helping with the power consumption. Actually, --debugreset put our device in a weird state that force us to send some command to it to do something and therefore consume the 1.3mA.

But, we need to lock the eeprom and therefore can't do nrfjprog -r.

How can we disable debug to reduce power consumption?

EDIT: Here a summary of test done with associated power consumption:

1. Programmed NRF52, physically trigger a reset by removing power and reapply it, device working as expected, power consumption as expected 0.3mA.

2. Programmed NRF52, keep power plugged, send nrfjprog --pinreset, device working as expected, power consumption is 1.6mA. I.e JTAG debug still enable a draw power.

3. Programmed NRF52, keep power plugged, send nrfjprog --reset, device working as expected, power consumption as expected 0.3mA. Solution not viable because no applicable with eeprom read lock.

4. Programmed NRF52, keep power plugged, send nrfjprog --debugreset, device not working as expected, power consumption is 0.5mA. Device frozen.

5. Programmed NRF52, keep power plugged, unplugged JLink cable, resetting by connecting RESETn to GND, device actually reset, device working as expected, power consumption is 1.6mA

[puck]

@endnode We use the pin RESETn and send a memwr with val 21 to both 0x10001200 and 10001204, and SWD debugger is connected to RESETn, VCC, SWDIO, SWCLK, DBG_RX, DBG_TX (uart interface) and GND. Physically trying to trigger a reset by connecting to pin to GND does trigger a reset of our firmware but does not reduce the power consumption to an acceptable value.

[endnode]

OK and then after grounding RESETn and removing/attaching SWD cable you can or you cannot read that memory? Also can you try how power consumption works if you schedule SW reset from within the FW (e.g. 30s after FW boots the first time) and you program + lock the FW and then remove the cable immediately?

[puck]

@endnode Yes, I can read back the memory after removing/reattaching the cable. I did program + lock, remove the cable immediately and then triggered a reset with external pin, this doesn't change anything.

[endnode]

Just a note: if you can do the read then chip isn't read-back protected (meaning that also SWD functions are not blocked). Still if debugger isn't active power consumption should go down to proper level.

[Sigurd]

It could be that the DIF is not being reset. Try the solution in this post