How to ensure an nRF52840 GPIO pin is disconnected at powerup via the DeviceTree board overlay file?

[Ken58]

I would like to ensure that a particular GPIO pin is disconnected when an nRF52840 is powered up via the initial configuration specified in the DeviceTree board overlay file.  I am using nRF Connect SDK v3.2.1 with an nRF52840 Dongle.  The pin will be used later as an output, but I must ensure there are no spurious signals on the pin until that time.  What is the recommended method of doing this?

Thank you,

Ken

[Vidar Berg]

Hello Ken,

On startup after a power-on-reset, all GPIOs are configured to their reset state (“disconnected input”), so you should not need to do anything to avoid spurious signals at startup. If this is not what you are seeing, perhaps it could be that the pin may be assigned to a peripheral or a driver in the device tree and is being reconfigured during driver initialisation?

(PIN_CNF[0])

"Wakeup from System OFF mode reset" is the only reset source which will not reset the GPIO registers.

Best regards,

Vidar

[Ken58]

Hi Vidar,

I can run my application on both the nRF52840DK and nRF52840 Dongle although I am using different pins (P1.06 and P0.13 respectively) to connect with the rest of my electronics.  Using the debugger and the DK, I have checked with that my PIN_CNF register value for the GPIO pin I am using is set as you show:  Input and Disconnect.  I checked DeviceTree as well.  Monitoring the DK pin with a scope, turning the DK on via the SW8 power switch, the GPIO pin remains stable and at 0V through the power on reset and boot.  On the other hand, when plugging the Dongle into a USB cable, the GPIO pin shows a spike and oscillating voltage as shown in the attached picture.  I think this is the root of the problem I am trying to solve.  Why is the Dongle producing this GPIO pin voltage during the power on reset and boot?  I have seen no evidence of this from the DK.

Best regards, and good to talk with you again,

Ken

[Vidar Berg]

Hi Ken,

Thank you, likewise. With the DK board switches in their default positions and the USB cable connected to the J2 port, the nRF52840 will be supplied with 3 V regulated from the USB bus in normal voltage mode while the dongle is powered directly from the USB in high voltage mode. Perhaps this explains the different behaviour. Could you please try supplying the nRF52840 DK from the nRF USB to see if you get the same result?

https://docs.nordicsemi.com/bundle/ug_nrf52840_dk/page/UG/dk/power_source.html

Best regards,

Vidar

[Ken58]

Hi Vidar,

I did the following with the nRF52840 DK:  Set SW9 to USB and plugged the USB cable into J3.  I did not modify the application software.  Powering the board on and off via plugging/unplugging the USB cable into my PC USB port does not produce anything weird on the GPIO pin I've been monitoring.  It seems well behaved as before.  I did note that VDD is now about 1.8V vs. 3V when powered as before.

Also, a couple more screen shots from my scope showing the weird pulse seen on GPIO P0.13 when plugging in the nRF52840 Dongle, zoomed in on the second pic.  I'm assuming the 60Hz noise is expected.

Ken

[Vidar Berg]

Hi Ken,

I'm wondering if you are simply seeing noise because the inputs are high-z. To be honest, I am not entirely sure what to expect here, but I know that USB ports can be quite noisy. Have you tried probing other unused pins on the dongle to see if they exhibit similar behaviour?

Ken58 said:

I did note that VDD is now about 1.8V vs. 3V when powered as before.

Yes, default is 1.8. It is controlled by the REGOUT0 register. The pre-programmed FW on the dongle sets it to 3v to be able to drive the on-board RGB LED.

Vidar

[Ken58]

Probing other GPIOs shows similar pulses.  Looking around the internet, I see indications that this is not uncommon on MCUs in general, though I haven't found a specific instance of Nordic SOCs.  Could you ask one of your EEs to comment on this?  If the pulse cannot be eliminated through setup within the application software, what would be the recommended method of connecting a GPIO port to other electronics - say a simple case of interfacing to a transistor operating as a switch?

I'll add a couple more pics.  As above, the yellow trace is the nRF52840 Dongle GPIO pin (P0.13).  In these new images, blue is the 5V supply from my electronics going into VBUS on the dongle.  The second pic is zoomed in on the area of interest.

Ken

[Vidar Berg]

Hi Ken,

I discussed this with one of my more hardware oriented coworkers who suggested adding an external pull-down resistor between the GPIO pad and the transistor gate to ensure the signal is not floating on startup.

Vidar

[Ken58]

Hi Vidar,

I had tried the pull-down resistor before using a rather high value (10k) without success.  Using a lower value resistor (2.2k) seems to allow dissipation of the startup pulse without switching on the transistor.  This becomes a balancing act to keep the "on" circuit current minimized and still pulling enough current on startup to pull down the pulse.  Never the less, this conversation helped me ensure there was nothing else in the software I could or should do to prevent the pulse.

A related question or 3 from our discussion:   You said "The pre-programmed FW on the dongle sets it to 3v to be able to drive the on-board RGB LED".  Where does this happen and can I easily set REGOUT0 to a different value within a Zephyr application on the nRF52840 Dongle?  Is there a system initialization function where this could be done, or DeviceTree parameter used by Zephyr?  What would be your recommended method?

A sidebar to that question:  I have a particular nRF52840 Dongle whose output voltage running the same application as we have been talking about is 3.5V.  That's a little weird!  Any idea on what might cause that?

I will probably have a related question or 2 concerning a custom board, but I'm not ready to ask yet.

Thanks,

Ken

[Vidar Berg]

Hi Ken,

REGOUT0 is a non-volatile register and you need to erase the UICR section  (ERASEUICR) before you can update it to a different setting. In the case where the dongle is not running the pre-programmed FW, the REGOUT0 register will be updated by the board init function here: https://github.com/nrfconnect/sdk-zephyr/blob/f80011480a2208ddbfb2decad3e3d9dab078c548/boards/nordic/nrf52840dongle/board.c#L12

Ken58 said:

A sidebar to that question:  I have a particular nRF52840 Dongle whose output voltage running the same application as we have been talking about is 3.5V.  That's a little weird!  Any idea on what might cause that?

Yes, this does not make sense. The REG0 regulator cannot output more than 3v3 according to the specification. Have you made any changes to the dongle's supply configuration by any chance (https://docs.nordicsemi.com/bundle/ug_nrf52840_dongle/page/UG/nrf52840_Dongle/hw_power_supply.html)?

Best regards,

Vidar

[Ken58]

Hi Vidar,

Thanks on the board_early_init_hook() function info.  Now, the 3V makes sense.

On the 3.5V dongle:  All of my nRF52840 dongles are ancient V1.0.0 hardware.  None have been modified for the supply configuration.  I don't know if that matters, but they are missing resistor R7, a 47 ohm series resistor between VBUS and VDDH.

In board_early_init_hook(), if REGOUT0 is not set with the DEFAULT value, it will not try to set REGOUT0 to 3V.  Furthermore, since board_early_init_hook() does not erase the UICR, it will successfully set REGOUT0 a maximum of twice for any particular nRF52804 Dongle.  What if back in 2018 for example, REGOUT0 was set permanently with option value 6, the missing value in the list of possible values.  Was there an 3V5 option?

Is there any code available in the SDK that would allow me to easily erase the UICR, and then execute code like in board_early_init_hook() to set REGOUT0 to 3V on my 3V5 Dongle?

Best regards,

Ken

[Vidar Berg]

Hi Ken,

The only scenario in high voltage mode where I can see it getting close to 3.5V is if REGOUT0 is configured to 3v3  and you add the +5% worst case accuracy deviation.

Ken58 said:

What if back in 2018 for example, REGOUT0 was set permanently with option value 6, the missing value in the list of possible values.  Was there an 3V5 option?

3v3 has always been the maximum and our SDK code (both new and old) has been setting the register to 3.0v.

(https://docs.nordicsemi.com/bundle/ps_nrf52840/page/power.html#ariaid-title74)

Ken58 said:

I don't know if that matters, but they are missing resistor R7, a 47 ohm series resistor between VBUS and VDDH.

This is interesting. I was not aware of this. Maybe this can explain why you didn't observe the same spike when testing with the DK. The series resistor was added in revision 2.0.0 of the DK.

Also from the "Reference circuitry" chapter of the Product specification:

• The schematics include an optional series resistor on the USB supply for improved immunity to transient overvoltage during VBUS connection. Using the series resistor is recommended for new designs.

Ken58 said:

Is there any code available in the SDK that would allow me to easily erase the UICR, and then execute code like in board_early_init_hook() to set REGOUT0 to 3V on my 3V5 Dongle?

There isn't any existing code for this unfortunately. But maybe you can read out the register before making any attempts on this. 

#include <soc.h>

...

    LOG_INF("NRF_UICR->REGOUT0: %d", NRF_UICR->REGOUT0);

Best regards,

Vidar