RAK3172-F external flash: extflash.info() fails

[Nicolas.p (Nicolas)]

Hi everyone,

I wanted to play with the external flash on the RAK3172-F, but I can’t even get the info to print. I’m using the example straight from the RUI3 API docs.

SUPPORT_EXTFLASH is defined, so the init code does get compiled in.

I dug into the code a bit: in udrv_extflash_init(), st ends up at 2, meaning uhal_extflash_init() returns EXTFLASH_ERR_TIMEOUT. Because of that the driver never gets flagged as initialized, and every extflash call afterwards fails.

I haven’t gone further than that yet. Can you reproduce it on your side, or is it more likely something wrong in my board setup or code?

Thanks for the help!

[beegee (Bernd Giesecke)]

What RUI3 version?
Are you sure you have the RAK3172- F ?
Please share a picture of the label on the stamp module.

[Nicolas.p (Nicolas)]

Hello beegee,

Thanks for your help !

I am using the 4.2.4 RUI3 version.

I am using a -F variant (as you can see below)

The used code is the following. I tried it with 2 different boards and both couldnt get memory info.

uint32_t extflash_test_addr = 0x0u;
void setup() {
  RAK_EXTFLASH_INFO info;
  int32_t st = api.system.extflash.info(&info);
  if (st != 0) {
    Serial.printf("[FLASH] extflash.info failed, status=%d\r\n", st);
    return;
  }

  Serial.printf("[FLASH] JEDEC: %02X %02X %02X\r\n", info.jedec_id[0], info.jedec_id[1], info.jedec_id[2]);
  Serial.printf("[FLASH] total_size=%lu, sector_size=%lu, page_size=%lu\r\n", (unsigned long)info.total_size, (unsigned long)info.sector_size, (unsigned long)info.page_size);

  static uint8_t rxbuf[16];
  memset(rxbuf, 0, sizeof(rxbuf));

  st = api.system.extflash.get(extflash_test_addr, rxbuf, sizeof(rxbuf));
  if (st != 0) {
    Serial.printf("[FLASH] read failed %d\r\n", st);
    return;
  }

  Serial.printf("[FLASH] READBACK:");
  for (uint8_t i = 0; i < sizeof(rxbuf); ++i)
    Serial.printf(" %02X", rxbuf[i]);
  Serial.printf("\r\n");
}

void loop() {
  api.system.sleep.all(30000);
}

The problem seems to come from HAL_SPI_Receive that reach time out when called in uhal_extflash_init (in udrv_extflash_init).

[beegee (Bernd Giesecke)]

Not sure what is wrong, but if I use your code on a RAK3172-F, it works as expected.
Log output from your code on my RAK3172-F:

[FLASH] JEDEC: C2 23 13
[FLASH] total_size=524288, sector_size=4096, page_size=256
[FLASH] READBACK: FF FF FF FF FF FF FF FF FF FF FF FF AA 01 05 00
Current Work Mode: LoRaWAN.

Do you have anything else conected to the RAK3172-F? Anything on the GPIO’s or external SPI pins?

[Nicolas.p (Nicolas)]

Indeed, i have an IMU connected to the SPI bus, its CS pin is the pin 18 (SPI1_NSS).
Does the SPI bus need to be free to allow the Memory use ?

[beegee (Bernd Giesecke)]

No, it should be no problem, the internal flash is connected like this:

Flash Pin	STM32WLE5 GPIO
SPI NSS	PB9
SPI MISO	PC2
SPI MOSI	PC1
SPI SCK	PB10
WP	PC0
RESET	PC3

I will check with R&D if there is any known problem.

Does the IMU use any GPIO’s for interrupts or other functions? If yes, which GPIO’s does it use?

[Nicolas.p (Nicolas)]

If I’m right, all the pins you listed are internal to the module: none of them are exposed on the RAK3172-F pads ?

Here is how the module is wired on my board:

In the previous version of my PCB, the module was a RAK3172-T. That’s why the pins specific to the -F are not here, they are left unconnected.

The IMU has an interrupt on IO5 (PA15), but the test sketch doesn’t configure any of my pins: it only contains the code I posted above.

[beegee (Bernd Giesecke)]

Do you have an option to test the module without peripherals, or at least without the external SPI device?

I connected an external SPI FLASH module and run your code without problems.

[Nicolas.p (Nicolas)]

I will try removing my peripherals one by one.

Could you share your compiled .hex please ?
I want to make sure that my compilation toolchain isn’t the problem.

[beegee (Bernd Giesecke)]

Here is the compiled hex file RAK3172-F-Flash-Wear.ino.hex (497.7 KB)

This is my Flash wear test application, but the Flash chip check is exactly the same code as in your application.

[Nicolas.p (Nicolas)]

Good news !

The problem seems to be with my compilation. Here is the result with your .hex :

Flash wear test started.
Flash JEDEC: C2 23 13
Flash total_size=524288, sector_size=4096, page_size=256
valid1 0 valid2 0 block# 9
valid1 FF valid2 ff block# 8
valid1 FF valid2 ff block# 7
valid1 FF valid2 ff block# 6
valid1 FF valid2 ff block# 5
valid1 FF valid2 ff block# 4
valid1 FF valid2 ff block# 3
valid1 FF valid2 ff block# 2
valid1 FF valid2 ff block# 1
valid1 AA valid2 0 block# 0
Valid block found 0
Flash init success
Read block 0
Valid data read: 
mySettings.counter: 1
mySettings.value: 0.00
mySettings.is_active: 1
Last block used 0
Available blocks = 10
Current Work Mode: LoRaWAN.
Data written to block 1
>> Counter: 1
>> Value: 0.010000
Data read back from block: 1
>> Counter: 1
>> Value: 0.010000
Data written to block 2
>> Counter: 2
>> Value: 0.020000
Data read back from block: 2
>> Counter: 2
>> Value: 0.020000
Data written to block 3

I will check my configuration again.

[Nicolas.p (Nicolas)]

Update :

After redownloading the RUI3 v4.2.4 firmware, the example code is now working perfectly. I’m not sure how I managed to corrupt my previous firmware, but that seems to have been the issue!

Thanks for your help!