We ahve an issue with LSM6DSM and LSM6DSV. When their SPI clock input pin 13 is driven by a week 47k pull down resistor, the LSM6D starts consuming abnormally high current of about 2.5 mA.
It seems that LSM6D becomes unstable and begins to oscillate. If however we connect the oscilloscope probe to any of its inputs/outputs, the current drops to about 150 uA above expected normal. This happens when LSM6D is in any operation mode: in a full shutdown mode just after cold start, in a low power mode, any SPI speed, etc.
The reason why we drive SPI clock input pin 13 by a week 47k pull down resistor is that LSM6D suppose to wake up the STM32L496RGT6 MCU from the shutdown mode. In the shutdown mode, the STM32L496RGT6 has only a week pull down option to drive its pins. But the voltage on the LSM6D SPI clock input pin 13 is about 7 mV and hence is well below the required logical 0 level. Documentation does not mention clock driver impednce and neither SPI nor I2C standards require low impedance input drivers.
We found a temporary work around as to connect a 300pF capacitor C5 from LSM6D SPI clock input pin 13 to the ground. The LSM6D consumption current returns to the expected normal (15 uA in our case), but we do not know how reliable this solution will be. For example, for different temperature ranges or other conditions. What value of C5 should be? Should it be 10 pF, 100 pf or 1000 pf? Also, we do not know should we connect similar capacitors C7 and C8 to the MOSI and NSS pins? C6 also sometimes helps.
Our PCB design is as good as it can be. We have dedicatad and uninterrupted ground and power planes, decoupling capacitors as close as physically possible to the LSM6D, etc. This happens to both LSM6DSV and LSM6DSM.
Please see attached schematic PDF.
