XIAO nRF52840 does not sleep after writing to flash

I am using an nRF52840 with SX1262 as a remote sensor. It achieves approximately 20 μA in normal use. I added a boot counter using the Adafruit LittleFS and InternalFileSystem libraries and can read and write files to the flash without problem. However, if I write to or delete a file, I can no longer put the nRF52840 to sleep: its current consumption rises to 660 μA and the red LED turns on. I assume something to do with flash is generating interrupts following a write.

Any ideas on how to debug this and shutdown the interrupts?

Hi there,

SO , looks like you’re close,
Red LED state: The red LED turning on is a massive clue. Check your board definition or pin mappings (LED_BUILTIN or the red user LED on the XIAO nRF52840). Is the filesystem library or boot counter inadvertently toggling or holding a GPIO pin high, or is a peripheral error state triggering an error handler that drives the LED and blocks sleep?

Recommended Next Steps for Debugging:

  1. Isolate the write: Write a minimal test sketch that boots, mounts LittleFS, performs one write, closes the file, and immediately enters a deep sleep state without running any BLE or sensor code. Check if the current still spikes to 660 µA. If it does, it confirms the LittleFS/InternalFS driver on the nRF52840 is leaving a clock or peripheral enabled.

HTH
GL :slight_smile: PJ :v:

Here is some code that demonstrates the problem. On a bare board with no external peripherals or SX1262 connected, I measure 7.5 μA when sleeping, but jumps to 650 μA following writing a byte to flash.

#include <Arduino.h>
#include <Adafruit_SPIFlash.h> 
#include <RadioLib.h>
#include <InternalFileSystem.h>
#include <Adafruit_LittleFS.h>

const unsigned long kSerialSpeed = 115200;

/// other pin connections
const u_int32_t kPinBatEnable = 14;     // Send LOW to enable voltage divider for battery monitor
const u_int32_t kPinBatADC = PIN_VBAT;  // ADC pin connected to XIAO on-board battery voltage divider

const char kPathTest[] = "/testfile";       // LittleFS path to store sensor id

using namespace Adafruit_LittleFS_Namespace;

Adafruit_FlashTransport_QSPI flashTransport;

// task handle for direct to task notifications
// - faster than a semamphore
static TaskHandle_t xTaskToNotify = NULL;

bool ReadWriteFlash() {

  if (!InternalFS.begin()) {
    // unable to initialise InternalFS
    return false;
  }

  uint8_t test_data[]{9};

  Adafruit_LittleFS_Namespace::File file(InternalFS);

  // open file for reading
  file.open(kPathTest, FILE_O_READ);
  // if file exists, read 1st byte and close
  if (file) { // file exists
    file.read((uint8_t*) &test_data, 1);
    file.close();
  }
  
  /*
  // write byte to file
  file.open(kPathTest, FILE_O_WRITE);
  if (file) {
    file.seek(0);
    file.write((uint8_t*) &test_data, 1);
    file.close();
  }
  */
  
  InternalFS.end();
  
  return true;
}

/**
 * Power down the flash memory.
 * (Saves approximately 10 µA.)
 */
void PowerDownFlash(void) {
  flashTransport.begin();
  flashTransport.runCommand(0xB9); 
  delayMicroseconds(5);
  flashTransport.end();  
}

/**
 * Activate battery voltage measurement by connecting one end of the voltage divider to ground.
 * NB although this increases total current consumption by several µA, this is necessary to
 * avoid over-voltage of the ADC by the connected battery.
 */
void EnableVoltageDivider(void) {
  pinMode(kPinBatEnable, OUTPUT);
  digitalWrite(kPinBatEnable, LOW);
}

void setup() {
  // necessary to prevent over-voltage of ADC pin
  // - increases quiescent current by Vbat / 1.5E6
  // - for 4.5 V supply : 3 microamps
  EnableVoltageDivider();

  // test read/write to flash storage
  ReadWriteFlash();

  // Use DC-DC converter to save power - best improvement is on high duty cycle
  NRF_POWER->DCDCEN = 1;   

  // saves a few microamps
  PowerDownFlash();

  xTaskToNotify = xTaskGetCurrentTaskHandle();

  while(true) {
    // istop here and go to sleep until woken by interrupt
    ulTaskNotifyTake(pdTRUE, portMAX_DELAY);
  }
}

void loop() {
  // never gets here
}

The schematic shows the red LED is connected to P0.26, which is not connected to anything else. I suspect the library flashes the LED during writes. I wonder whether the fact that this particular board uses low side switching for the LEDs so that sending the pin LOW turns the LED on has caused this bug. I’ll have a dig through the libraries at the weekend.

Anyway, the schematic shows a 2.2K series resistor, which with a 2V voltage drop gives around 600 μA of current draw for the LED, so it all checks out.

Hi there,

Yes All xiao LEDS are active LOW.. :+1:

GL :slight_smile: PJ :v:

The board variant file configures LED_BUILTIN and PIN_LED as LED_RED, so I suspect this is why the red LED is getting driven.

´´´
/ LEDs
#define PIN_LED (LED_RED)
#define LED_PWR (PINS_COUNT)
#define PIN_NEOPIXEL (PINS_COUNT)
#define NEOPIXEL_NUM (0)

#define LED_BUILTIN (PIN_LED)

#define LED_RED (11)
#define LED_GREEN (13)
#define LED_BLUE (12)

#define LED_STATE_ON (1) // State when LED is litted
´´´

not sure this is it but the numbers line up too well, 660 minus your normal 20 is about the 600 the red led draws, so it might actually be sleeping with the led left on. id try driving the red led pin high before sleep, since its active low, and see if the current drops back down

… yes some of the led are on when pulled low… i think thi started around ESP32 days… I think pulling led low is easier on the MCU than driving it high…

Hi @RussH,
I tried the code you posted, and it works fine—it reads 7uA. Could you show me the code that reads 650uA?