Hello, I have been using the ADXL335 to measure the noise floor in my laboratory with a power spectral density (PSD) of acceleration. The data sheet says the typical noise at the Z axis is 300 µg/√Hz, yet I believe I am measuring far below this.
To create this plot, I sampled at 1100 Hz (twice the max bandwidth of the Z output), recorded data for about a minute, and computed a PSD in python using Welch's method. The expected noise is the 300 µg/√Hz from the datasheet. I am wondering if there is anything wrong with my method or conversion that may cause this low of a noise floor, or does this PSD make sense for this sensor? I understand that adding capacitors to the outputs can decrease the noise floor, but I am just reading out straight from the Z axis output and computing a PSD after converting to g based on the datasheet.
My setup is as follows:
- This is a borrowed accelerometer, but the person before me had sewn it into a velcro strap and powered it with a 3 V battery. There is outgoing conductive yarn that is connected to the accelerometer outputs and ground, which I use to connect to readout electronics
-I am using a Moku:Go to collect data, and I connect the accelerometer to the Moku analog input by using a BNC to banana plug adaptor, where I can measure one axis at a time by using alligator clips to connect to the conductive yarn.
-I use the Data Logger instrument within the Moku to record data for varying sample rates and amounts of time.
-I then download this data to a csv file which I can use in Python to compute a PSD using the welch function from scipy.signal. I am using the 'density' option in this function, and then taking the square root of the resulting output to get a PSD in units of g/√Hz.
Let me know if I need to clarify anything, any help is appreciated!
