How does Deep Power-Down Work with QSPI and some example code.

[Cheets]

QSPI_Test.ino

Hi,

Just wondered about the following ....

Strange parts of this code ... Or things I don't understand

After the first READ in Setup() it successfully reads the data (Returns 0 = NRFX_SUCCESS), but the status flag
has the top 8 bits set to 0xFF which causes nrfx_qspi_mem_busy_check() to show 17 (Returns 17 = NRFX_ERROR_BUSY).
However masking the STATUS register with 8 reveals the Ready Status = 1, QSPI is ready!
This was why I wrote the QSPI_IsReady().

nrf_qspi_phy_conf_t not visible as a structure if you try and set it like this

    QSPIConfig.phy_if {
        .xxx = yyy,
        .aaa = bbb
     };
 
I don't know what the significance of the 48ms Deep Power-down Mode (DPM) is.
Will it go into DPM if not used for 48ms and then take 48ms to wake up if instructed?

Thanks

Paul

[Cheets]

Hi Vidor,

Thanks for replying so quickly.

The SReg = 0xFF and stays like that until I do a write.  I have pasted the test output from my (slightly modified) program.

Thanks for the datasheet, I was using the nRF52840_PS_v1.1.pdf, I think they are the same (or at first glance they are).

The QSPI flash I am using is the one built into the chip itself.

I can't get it to work with all 4 IO pins as well (Just tried today, it will work with 2 IO pins, reading is faster, writing not so much).

Paul

QSPI Test Output

(Setup) QSPI Initialising ...
(QSPI_Initialise) Wait for QSPI to be ready ...
(QSPI_Initialise) QSPI is ready
(Setup) QSPI initialised and ready
(Setup (After initialise)) QSPI is busy/idle ... Result = 0
(Setup (After initialise)) QSPI Status flag = 0xA (from NRF_QSPI) or 0xA (from my QSPI_Status_Ptr Pointer)
(Setup) QSPI IFCONFIG0 = 0x89
(Setup) QSPI IFCONFIG1 = 0x89
(Setup) QSPI Pins
  (Setup) IO0 = 0x14
  (Setup) IO1 = 0x18
  (Setup) IO2 = 0x16
  (Setup) IO3 = 0x17
(Setup) QSPI is about to be read and then erased. Current busy state is = 0
(Setup (Before read)) QSPI is busy/idle ... Result = 0
(Setup (Before read)) QSPI Status flag = 0xA (from NRF_QSPI) or 0xA (from my QSPI_Status_Ptr Pointer)
(Setup) QSPI took 8ms to read 64Kb ... Read result = 0
QSPI Data : 0x0 0x1 0x2 0x3 0x4 0x5 0x6 0x7 0x8 0x9
(QSPI_Erase) Erasing 64Kb of flash memory
((QSPI_Erase) Waiting Loop Breakout) QSPI is busy/idle ... Result = 0
((QSPI_Erase) Waiting Loop Breakout) QSPI Status flag = 0xFF00000A (from NRF_QSPI) or 0xFF00000A (from my QSPI_Status_Ptr Pointer)
((QSPI_Erase) Finished Waiting) QSPI is busy/idle ... Result = 0
((QSPI_Erase) Finished Waiting) QSPI Status flag = 0xFF00000A (from NRF_QSPI) or 0xFF00000A (from my QSPI_Status_Ptr Pointer)
(QSPI_Erase) QSPI took 0ms to erase 64Kb of flash memory
(Setup (After erase)) QSPI is busy/idle ... Result = 0
(Setup (After erase)) QSPI Status flag = 0xA (from NRF_QSPI) or 0xA (from my QSPI_Status_Ptr Pointer)
(Setup) QSPI read after erase
(Setup) QSPI took 17ms to read 64Kb ... Read result = 0
(Setup (After read and erase)) QSPI is busy/idle ... Result = 0
(Setup (After read and erase)) QSPI Status flag = 0xFF00000A (from NRF_QSPI) or 0xFF00000A (from my QSPI_Status_Ptr Pointer)
QSPI Data : 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF
(Setup) Just before QSPI write
(Setup) QSPI took 352ms to write 64Kb ... Write result = 0
(Setup (After write)) QSPI is busy/idle ... Result = 0
(Setup (After write)) QSPI Status flag = 0xA (from NRF_QSPI) or 0xA (from my QSPI_Status_Ptr Pointer)
(Setup) Just before QSPI read
(Setup) QSPI took 8ms to read 64Kb ... Read result = 0
(Setup (After read)) QSPI is busy/idle ... Result = 0
(Setup (After read)) QSPI Status flag = 0xFF00000A (from NRF_QSPI) or 0xFF00000A (from my QSPI_Status_Ptr Pointer)
QSPI Data : 0x0 0x2 0x4 0x6 0x8 0xA 0xC 0xE 0x10 0x12

[Vidar Berg]

Hi Paul,

There is no QSPI flash in the nRF52840 IC package, so I guess the P25Q16H flash must be placed next the nRF inside the Seeed Studio module.

Cheets said:

I can't get it to work with all 4 IO pins as well (Just tried today, it will work with 2 IO pins, reading is faster, writing not so much).

Where did you find the pin numbers?

Vidar

[Cheets]

Hi Vidor,

From the maual on your site!  nRF52840_PS_v1.1.pdf page 286, which names it as a peripheral.  I got the pins from Seeed XIAO nRF52840 v1.0.pdf (Pins P0.20 - P0.25).

I would attach both, but I cannot find an upload option.

Paul

[Vidar Berg]

Hi Paul,

The nRF 52840 PS recommends some GPIOs for QSPI operation, but I don't know if Seed studio as opted to use all those or not.

What I'm looking for a datasheet or schematic that tells us exactly how the P25Q16H and nRF52840 are connected internally in the module. I can't seem to find it.

Vidar

[Cheets]

Hi Vidor,

Link to your document -

https://infocenter.nordicsemi.com/pdf/nRF52840_PS_v1.1.pdf

Link to SEEED document -

https://files.seeedstudio.com/wiki/XIAO-BLE/Seeed-XIAO-nRF52840-v1.0-SCH.zip

Paul

[Vidar Berg]

Hi Paul,

Thanks for pointing me to the schematics. I went over the pin assignments in your code and checked that they matched those used in the module, so the pinout is fine.

Cheets said:

I can't get it to work with all 4 IO pins as well (Just tried today, it will work with 2 IO pins, reading is faster, writing not so much).

I had a closer look at the P25Q16H datasheet today and see now that the Quad Enable (QE) bit has a different position in the status register compared to the MX25R64 flash IC mounted on our 52840 DKs (9 vs 6) so I think this must be why Quad mode does not get enabled in your case.

Please try to configure the IC for quad mode operation using the function below and see if it works. I don't have a P25Q16H here. Otherwise I would have tried it here.

    // Switch to qspi mode
    uint8_t temporary[] = {0x00, 0x02};
    cinstr_cfg.opcode = QSPI_STD_CMD_WRSR;
    cinstr_cfg.length = NRF_QSPI_CINSTR_LEN_3B;
    err_code = nrf_drv_qspi_cinstr_xfer(&cinstr_cfg, &temporary, NULL);
    APP_ERROR_CHECK(err_code);

Vidar

[Cheets]

Hi Vidor,

Thank you very much for your help, it is much appreciated.  I now have a better understanding of the documentation and how to read it.  Also, your snippet was perfect, the read times are now down to 4ms - 15ms for a 64K block, write times are down from 363ms to 350ms (64Kb).

I am happy for you to close this case now with "Answered"!

And, once again, thank you.

Paul

[Vidar Berg]

Hi Paul,

No problem! Thanks for confirming that it worked.

Vidar