I2C2 and UART interference issue

[elsalahy]

Summary:

The device (V0.1) halts when we are transmitting data via UART and I2C2.

Steps to Reproduce:

1. Add print functions around I2C2 transmissions
2. Observe the console output and the device behavior

What do you see now?

System halt due to a SW or HW issue.

What do you want to see instead?

System functions without any halt.

How do you propose to implement this?

Identify if it is a HW issue, or a SW issue. In case it is a HW issue, we will need to adjust the schematics of V0.2 design In case it is a SW issue, we will need to fix the driver causing the issue

Environment:

Baremetal

What can you do yourself and what do you need help with?

All

[elsalahy]

@azerimaker as per our discussion, I opened this issue so we can share info on behavior

Also here is another test case scenario where I switch pins assignment as seen in the code below to try to pinpoint the issue.

Test scenario: I changed the Pins assignment in STNODE_bsp.h as seen below

#define SENSOR_I2C1                  I2C2
#define SENSOR_I2C1_TIMING           0x00707CBBU  //Evaluates to 10 KHz bus frequency
#define SENSOR_I2C1_TIMOUT           100U //Read and write operattions timeout in ms
#define SENSOR_I2C1_PERIPH_CLK      RCC_PERIPHCLK_I2C2
#define SENSOR_I2C1_SOURCE_CLK      RCC_I2C2CLKSOURCE_SYSCLK

#define SENSOR_I2C1_CLK_ENABLE()    __HAL_RCC_I2C2_CLK_ENABLE()
#define SENSOR_I2C1_CLK_DISABLE()   __HAL_RCC_I2C2_CLK_DISABLE()

#define SENSOR_I2C1_SDA_GPIO_CLK_ENABLE()      __HAL_RCC_GPIOA_CLK_ENABLE()
#define SENSOR_I2C1_SCL_GPIO_CLK_ENABLE()      __HAL_RCC_GPIOA_CLK_ENABLE()

#define SENSOR_I2C1_FORCE_RESET()              __HAL_RCC_I2C2_FORCE_RESET()
#define SENSOR_I2C1_RELEASE_RESET()            __HAL_RCC_I2C2_RELEASE_RESET()

#define SENSOR_I2C1_SCL_PIN                    GPIO_PIN_12
#define SENSOR_I2C1_SCL_GPIO_PORT              GPIOA
#define SENSOR_I2C1_SDA_PIN                    GPIO_PIN_11
#define SENSOR_I2C1_SDA_GPIO_PORT              GPIOA
#define SENSOR_I2C1_SCL_SDA_AF                 GPIO_AF4_I2C2

#define SEC_ELM_I2C2                  I2C1
#define SEC_ELM_I2C2_TIMING           0x00707CBBU  //Evaluates to 10 KHz bus frequency
#define SEC_ELM_I2C2_TIMOUT           100U //Read and write operattions timeout in ms
#define SEC_ELM_I2C2_PERIPH_CLK      RCC_PERIPHCLK_I2C1
#define SEC_ELM_I2C2_SOURCE_CLK      RCC_I2C1CLKSOURCE_SYSCLK

#define SEC_ELM_I2C2_CLK_ENABLE()    __HAL_RCC_I2C1_CLK_ENABLE()
#define SEC_ELM_I2C2_CLK_DISABLE()   __HAL_RCC_I2C1_CLK_DISABLE()

#define SEC_ELM_I2C2_SDA_GPIO_CLK_ENABLE()      __HAL_RCC_GPIOA_CLK_ENABLE()
#define SEC_ELM_I2C2_SCL_GPIO_CLK_ENABLE()      __HAL_RCC_GPIOA_CLK_ENABLE()

#define SEC_ELM_I2C2_FORCE_RESET()              __HAL_RCC_I2C1_FORCE_RESET()
#define SEC_ELM_I2C2_RELEASE_RESET()            __HAL_RCC_I2C1_RELEASE_RESET()

#define SEC_ELM_I2C2_SCL_PIN                    GPIO_PIN_9
#define SEC_ELM_I2C2_SCL_GPIO_PORT              GPIOA
#define SEC_ELM_I2C2_SDA_PIN                    GPIO_PIN_10
#define SEC_ELM_I2C2_SDA_GPIO_PORT              GPIOA
#define SEC_ELM_I2C2_SCL_SDA_AF                 GPIO_AF4_I2C1

and I get this log

This is only for testing purposes and it shows that when I2C2 bus is being used with UART printing the device halts.

If this is a HW issue, it would be due to shortening of the I2C2 bus and the node UART Tx If it's a SW issue, it will be mainly because of ST HAL drivers

Also it is good to keep in mind that I tested the following scenarios 1) Removed all UART prints and the device functions as expected 2) Increased stack and heap sizes and the problem still occurs

[azerimaker]

Thanks @elsalahy for a detailed FW review. Since I2C1 works properly, I suspect crosstalk is the culprit here. Here's the screenshot of the lower layer, as we can see there's a section of about 13mm, where these lines run alongside each other and it might be the reason for such behavior. One way for fix this is to decrease the pull-up resistor values (currently 4.7K Ohm). I'll play with smaller values and get back to you.

[azerimaker]

Lowered the pull-up resistors to 1.8K Ohm (don't mind the through-hole resistors :sweat_smile: ), still the same behavior.

[KrishnaIyer]

Sorry I had to: