CC3220SF: I2C bus clear

Hi Jiří,

Is your question about how to get the I2C bus unstuck?

These E2E threads might have some relevant info:

https://e2e.ti.com/support/wireless-connectivity/wi-fi-group/wifi/f/wi-fi-forum/829053/cc3220s-launchxl-changing-between-i2c-and-gpio-mode

https://e2e.ti.com/support/wireless-connectivity/wi-fi-group/wifi/f/wi-fi-forum/811025/cc3200-cc3200-i2c-communication-failure

Regards,

Michael

Hi Michael,

thank you for your reply. I have checked both links and there is no information I am looking for.

I have understood: I have to implement my own "bus clear" function. Currently I am calling this function from init, before my FreeRTOS starts
and before TI SDK takes I2C pins under control. So I have no issue with driving pins. But I do have problem with timing. See my question above regarding UtilsDelay(..).

static void busClear(void)
{
	uint8_t u8L_sclPulses = 8+1; /* 8 pulses on SCL line to clear I2C bus, the ninth one is stop condition */

	gpio_set(BOARD_PIN_I2C_SCL, 1);
	gpio_set(BOARD_PIN_I2C_SDA, 1);

	UtilsDelay(I2C_BUS_HALF_PERIOD_US);

	/* Toggle SCL */
	do
	{
		gpio_set(BOARD_PIN_I2C_SCL, 1);
		UtilsDelay(I2C_BUS_HALF_PERIOD_US);

		gpio_set(BOARD_PIN_I2C_SCL, 0);
		UtilsDelay(I2C_BUS_HALF_PERIOD_US);

		u8L_sclPulses--;
	} while (u8L_sclPulses > 0);

	/* Generate STOP condition */
	gpio_set(BOARD_PIN_I2C_SDA, 0);
	UtilsDelay(I2C_BUS_HALF_PERIOD_US);

	gpio_set(BOARD_PIN_I2C_SCL, 1);
	UtilsDelay(I2C_BUS_HALF_PERIOD_US);

	gpio_set(BOARD_PIN_I2C_SDA, 1);
	UtilsDelay(I2C_BUS_HALF_PERIOD_US);
}

It would be also good to have some detection of stuck SDA line, but I am not sure how to do that.

Regards
Jiří 

PS:
Off-topic: section 3.1.16 in the I2C specification ( UM10204 I2C-bus specification and user manual) is not very clear.
There is no information about timing constrains and there is also nothing about the need to have STOP condition
part of 9 pulses (actually the ninth pulse). But my colleague gave me some links where people are using this.
Very confusing. :-(

Hi Jiří,

Your code to clear the I2C looks fine.

I am not the most familiar with third-party I2C documentation so I cannot comment on that unfortunately.

Regards,

Michael

Michael,

thank you regarding I2C bus clear procedure.

Will you advice regarding UtilsDelay(..) function? What is the time duration for UtilsDelay(1)?
I expect that input clock related to this function is 80MHz for CC3220MODASF.

Jiří

Hi Jiří,

Yes, the delay length of the UtilsDelay() function is linked MCU clock of 80MHz.

Specifically, UtilsDelay() takes 3 clock cycles to complete. As each clock cycle of the ARM Cortex M4 takes (1s/80MHz) = 0.0125us, UtilsDelay(1) should result in a delay of 0.0375us.

What is the delay you are measuring?

Regards,

Michael

Hi Michael,

thank you for info, I will use GPIO_setConfig() based on advice in the link above to get more precise results.
I will come back with my results as soon as I will switch back to relevant task in my project.

Jiří

I believe the timing is correct and the 75ns (which is double of expected) is caused just be measurement. As for pin toggling I need at least 1 instruction. ;-)