Noise comparison of three accelerometers, LSM303C, LSM6DS3, LSM6DSL

[Ray Sun]

I have three accelerometers and the circuit is designed according to the datasheet.

I have collected the data of three kinds of accelerometers with the vibration test-bed.

The sampling frequency of the accelerometers is all set to 800 Hz, placed horizontally, and the z-axis is perpendicular to the horizon.

The vibration test-bed is set as the vibration frequency 50Hz, and the vibration amplitude is 0.5g. After that, I analyzed the z-axis data of the three accelerometers with FFT, and found that the noise of lsm303c is smaller than that of the other two accelerometers.

Is that possible?

[Eleon BORLINI]

Hi @Ray Sun​ , the LSM303C full width at half maximum of the peak is however much broader than for the other two devices. Are you sure that the integral noise value is lower? The behavior seems to be related to the anti-alias filtering strategies (signal before the ASIC ADC block): LSM6DSL has to be set in High Performance mode to enable an analog anti-aliasing lowpass filter BW of 400Hz for ODR < 1.6kHz (AN5040 p. 13). In the LSM6DS3 case, the low-pass anti-aliasing filter can be set enabling the XL_BW_SCAL_ODR bit in CTRL4_C register (AN4650 p.25); also in this case, the anti-aliasing filter is enabled in High Performance mode only, and also a low pass filter LPF1 can be enabled after the ADC block. Last case, the LSM303 device has no selectable anti-alias filters, because the filter cannot be enabled or disabled (should be always enabled by default, in any case): this is the reason for which you don't see the spikes in the last FFT. Regards