Clarification about SPIM - SS pin management for nrf9160

RE: Clarification about SPIM - SS pin management for nrf9160

kgarland789 — Wed, 11 Nov 2020 01:35:56 GMT

Hi Kenneth,

What does your prj.conf look for this example? When I try to instantiate a spi_cs_control struct, I get a "storage size of cs_pin isn't known" error.

Thanks,

-Kyle

RE: Clarification about SPIM - SS pin management for nrf9160

frax84 — Thu, 30 Apr 2020 13:34:10 GMT

So the software actually pulls CS line LOW and HIGH automatically, without direct gpio write. Nice, this allow to write more compact and error-proof code.

RE: Clarification about SPIM - SS pin management for nrf9160

Kenneth — Thu, 30 Apr 2020 11:14:03 GMT

I was thinking the CS logic was handled in hardware, but by experimenting with it I can see it's handled in software by the driver. So it is possible to use the CS logic in the driver, I made an example below.

Best regards,
Kenneth

devzone.nordicsemi.com/.../main.c

RE: Clarification about SPIM - SS pin management for nrf9160

frax84 — Wed, 29 Apr 2020 12:47:01 GMT

Hi Kenneth,

ty for the answer. Did you test what you said? Because, from Zephyr documentation, i read:

structspi_cs_control
#include <spi.h>
SPI Chip Select control structure.

This can be used to control a CS line via a GPIO line, instead of using the controller inner CS logic.

It says this is an alternative to "controller inner CS logic". I understood that "controller inner CS logic" is the one that is not supported in this case, while i thought that spi_cs_control structure had some kind of effect. Unfortunately due to COVID i have not my scope to test the behaviour of the CS line with or without using this structure, but maybe you already tested it

Thanks,

Frax

RE: Clarification about SPIM - SS pin management for nrf9160

Kenneth — Wed, 29 Apr 2020 12:31:24 GMT

[quote user=""]Have the struct elements "operation" (it has cs_active_high option) and "spi_cs_control" have any effect on the behaviour of the ss-pin? [/quote]

No, likely this could be used if the hardware supported it, but it doesn't.

So you end up with the spi_dev_write() you show.