Connecting two sensors to the same SPI bus - possible contention

[ELask.1]

Hi,

In our system there's an LSM6DSL accelerometer and LPS22HH barometer connected to the same SPI bus.

When nCS of the accelerometer is asserted (0), the nCS of the barometer is de-asserted (1) and vice versa.

HOWEVER, as nCS=1 (for both sensors) also means that the components works in I2C mode, the following condition occurs (as an example):

• nCS_Acc = 0, nCS_Baro = 1 : an attempt to communicate with the accelerometer
• MCU (STM32F412) communicates over SPI - it drives MOSI & SCLK with the relevant acceleromter command
• Concurrently to the accelerometer, the barometer which is now in I2C mode, receives the MOSI & SCLK signals on its SDA & SCL lines
• The accelerometer responds as expected on SDO towards the MCU MISO
• Now, if the barometer interpret the SPI communication to its peer as a valid I2C address byte it ACK on the SDA line, (which is shared with MOSI) and make a bus contention and mess up the communication for both sensors....

My questions are:

• Do I understand correctly or did I oversee something?
• How can we avoid a such potential contention/communcation mess up?

Thanks for any advice

[Eleon BORLINI]

Hi @Community member​ ,

you are in theory right, there could be some undesired cross talk between the sensors. But it is a remote possibility.

Namely, please note that:

• The I2C and the SPI are different protocols, often running at different clock speeds, and the first one needs a precise salve address and acknowledges to be effective (i.e. to trigger the digital communication block in the sensors' ASIC).
• You can however prevent a "false-positive" I2C running setting the device as follows:
  • For the LSM6DSL IMU, you can write the I2C_disable bit of CTRL4_C (13h) register (datasheet p.63);
  • For the LPS22HH pressure sensor, you can write the I2C_DISABLE bit of IF_CTRL (0Eh) register (datasheet p.42).

Please let me know if everything is working.

-Eleon

[ELask.1]

Thanks!

This was very helpful.