Adding an external flash memory "MX25R6435F" to my arduino nano 33 ble board

[gizgiz]

I want to use the external flash memory MX25R6435F to run my code which has a big size on my arduino nano 33 ble but I'am a newbie and I don't find supports that can help me,

Is that possible and should I add some configuration ?

[Vidar Berg]

Hello,

I don't see any external memory ICs on in the schematics: https://docs.rs-online.com/9208/A700000006519068.pdf. Have you connected it via the pin headers?

Best regards,

Vidar 

[gizgiz]

Yes I want to use the pin headers to use the external flash memory MX25R6435F

[Vidar Berg]

If you have already connected the flash IC to your board, then the next step I would suggest is to try the JEDEC SPI-NOR Sample to verify if you are able to interface with the chip (if the connections are good, etc).

Below is a devicetree overlay you can use to add support for the macronix flash (can be placed in /spi_flash/boards along with the other board specific config/overlay files).

arduino_nano_33_ble.overlay

And config file to enable the QSPI driver:

arduino_nano_33_ble.conf

[gizgiz]

Thank you for your answer, but I want to use "nRF5" vs code or Segger without zephyr, is that possible?

[Vidar Berg]

Yes, you can try the QSPI Example from nRF5 SDK 17.1.0 instead then.

[gizgiz]

Can I run my program directly in the external flash memory?

[Vidar Berg]

It is possible, but I wouldn't recommend running an entire application from external flash on the nRF52840 because of errata and performance limitations. It may make more sense if you run certain parts of the application from external flash. For instance, if you have some library code that is does not need to run too often.

Do you know what's taking up so much space in your application? You can fit a lot of code on a 1MB device like the 52840..

[gizgiz]

I want to add artificial intelligence application using edge impulse and I found that the code libraries request a lot of memory space, have you an idea about that?

[Vidar Berg]

I would recommend you consider the Edge impulse samples included in our nRF connect SDK instead of trying to integrate all this into a nRF5 SDK based project.

https://developer.nordicsemi.com/nRF_Connect_SDK/doc/2.0.0/nrf/applications/machine_learning/README.html

[gizgiz]

But I want my project to be independent of internet, and I made project with nRF5 SDK which require a long time

[Vidar Berg]

Internet access is only required when you need to train a machine learning model in the cloud, and that applies regardless of what SDK you use.

[gizgiz]

But in the prj.conf file we should put our model link

And i think that need internet connection

I also had this error when flashing:

C:\wrapper1\prj.conf:23: warning: ignoring malformed line '"'">studio.edgeimpulse.com/.../download

error: Aborting due to Kconfig warnings

CMake Error at C:\Users\ghaaz\ncs\v1.9.1\zephyr\cmake\kconfig.cmake:272 (message):
command failed with return code: 1

I also had a look at the questions posted in this forum, and I found that they couldn't work with this project and they advice them to integrate the model in their project:

Unable to add my Edge Impulse Module to my nRF Connect project. - Nordic Q&A - Nordic DevZone - Nordic DevZone (nordicsemi.com)

Having issues downloading trained module into NRF Connect - Help - Edge Impulse