I have the FlexSPI able to erase, read, write, read successfully on power up - but on subsequent runs (without re-powering the board), erase doesn't seem to erase the FLASH.
After failing to erase, if i cycle power - erase works.
Sending a RESET command to the FLASH part doesn't fix this phenomenon.
I've noticed a few things on the scope:
1. My SCK frequency is around 1.78MHz instead of the expected 104MHz. Could someone provide guidance on properly setting SCK frequency?
2. The SCK level ranges from 0V to 1.68V - almost like the IOs are on a 1.8V pad, but I don't see anywhere that you can specifically set 1.8V vs 3.3 V on the IOMUX CTL bits, is this something configurable? How would I get this to 3.3V?
My code is based off of flexspi_nor_polling_transfer. I'm currently configuring clock (I thought) with:
flexspi_device_config_t deviceconfig = {
.flexspiRootClk = 104000000, // 104 MHz
.flashSize = FLASH_SIZE,
.CSIntervalUnit = kFLEXSPI_CsIntervalUnit1SckCycle,
.CSInterval = 11, //100ns -> 9.61ns @ 104Mhz ,
.CSHoldTime = 2, //see RM fig 24-26: hold CS after last data read: (5 active + 5inactive)
.CSSetupTime = 2, //see RM fig 24-26: CS asserted before data read: (4.6 active + 5 inactive?)
.dataValidTime = 0,
....
};
and I'm not using XIP, so in app.h from the :
BOARD_BootClockRUN() is also untouched (with XIP set to 0).
BOARD_InitPins is the default except for IOMUXC_GPIO_SD_B1_05_FLEXSPI_A_DQS, which is used in internal loopback mode.
Hi kerryzhou,
1. Yes - what I'm finding actually is:
API reads give the same data as AHB reads.
Running repeated erases doesn't fix erase.
Updating the value via FLASH program appears to be successful.
API reads give the same data as AHB reads - gives back what I program, but after erase - reads back the programmed value (not 0xFF's).
2. Test tool is OK - I'm using an oscilloscope to test.
Is the above code I shared the correct place to update SCK frequency??
I had reduced the frequency there to 10MHz and it still showed the exact same 1.78MHz.
Our scope samples at 1Gbps, 100MHz bandwidth - so well within the range to probe 10MHz
3. Our board has a direct connection from the RT1015 to the flash, no pull-ups/pull downs aside from CS:
4. Yes - running from RAM. The end goal is to have code stored in NAND FLASH and the on boot - the boot loader will copy the executable (code/data/ and all) to chip internal RAM.
Thanks!
Hi variable_andrew,
1. You said: Updating the value via FLASH program appears to be successful.
Do you mean, even you erase is failed, then your program still works?
Please test this steps:
erase->read OK or not? (OK)-> write the specific data-> read the write data is ok or not?->erase again OK or not?
If the erase is failed, then read the data, is it contains the previous write data or 0XFF?
2, About the FlexSPI clock
You modify the wrong area, your modified point is the XIP boot clock.
If run from the RAM, please check this code:
FlexSPI clock is 120Mhz.
This is my test result in the FlexSPI_CLK_A GPIO_SD_B1_07pin.
So you can find my FlexSPI clock is correct.
If you are using the same SDK code configuration and run from RAM, please test it like me.
3. Our board has a direct connection from the RT1015 to the flash, no pull-ups/pull downs aside from CS
I am testing the MIMXRT1015-EVK board, adesto flash:
Please check the flexspi_clock_init code in your side, and test it again.
Have a great day,
Kerry
-------------------------------------------------------------------------------
Note:
- If this post answers your question, please click the "Mark Correct" button. Thank you!
- We are following threads for 7 weeks after the last post, later replies are ignored
Please open a new thread and refer to the closed one, if you have a related question at a later point in time.
-------------------------------------------------------------------------------
