Post Go back to editing

ADXL355 Noise Performance

Category: Datasheet/Specs
Product Number: ADXL355

Hello all,

My research group and I have had trouble reconciling the specs given in the ADXL355 datasheet for Noise Density.

My understanding of the device so far is that "effective" bandwidth in the ADXL 355 is the ODR/4. We are using an I2C bus with an ODR of 2KHz so 2KHz/4 = 500Hz, and not the expected Nyquist frequency due to an internal LPF. The second is that there is a coefficient present in the calculation for RMS noise based on this blog post:

Using the values from the ADXL355 datasheet, I've computed:

Total Noise(RMS) = ND * sqrt(1.6 * BW)
          = 22.5ug * sqrt(1.6 * 500Hz)
          = 636ug
After running some tests with the device sitting statically on insulating foam in a dark room, we've generated PSD's from 3600 seconds worth of data using the VibrationData Toolbox.

Time-Series Data:
Additionally, computing RMS noise on the dataset using this formula:
and the following Python script:
import numpy as np
import pickle

def rms(x):
    return np.sqrt(np.mean(np.square(x)))

def rmsd(x, mean):
    return rms(x - mean) 

with open('ADXLbenchtest10182022_1020.pickle', 'rb') as handle:
    axisdata = pickle.load(handle)

xs = axisdata['X-Axis']

xout = rmsd(xs, np.mean(xs), len(xs))
Provided a result of  599ug RMS.
We anticipated the bottleneck of our testing to be the lab environment and were surprised to find performance that is seemingly better than that outlined in the spec sheet by a reasonable margin. Is it possible to provide PSD's from the internal testing that can corroborate performance for this product? 

Cleaned up presentation so issue is clearer.
[edited by: mrem21 at 8:33 PM (GMT -4) on 19 Oct 2022]