6.1.3.5 Filtering to reduce measurement noise
An adjustable filter is applied to analogue measurements, reducing signal components at
selected frequencies. The following figure shows the filter frequency response. Using the first
notch frequency (fN1) parameter, users can select the placement of the filter notches. The first
notch falls at the specified fN1, and subsequent notches fall at integer multiples of fN1.
Commonly, fN1 is set at 50 or 60 Hz to filter 50 or 60 Hz signal components, reducing noise
from ac power mains.
Filtering comes at the expense of measurement time. The time required for filtering is equal to
1/f
N1
. For example, setting fN1 equal to 50 will require 1/50 sec (20 ms) for filtering. As fN1 is
set to smaller values, random noise in the measurement results decreases, while measurement
time increases. The total time required for a single result includes settling + filtering + overhead.
Consult the following technical paper at www.campbellsci.com/app-notes for in-depth treatment
of measurement noise: Preventing and Attacking Measurement Noise Problems.
CR1000X filtering details
The datalogger utilizes a sigma-delta ADC that outputs digitized data at a rate of 31250 samples
per second. User-specified filtering is achieved by averaging several samples from the ADC.
Recall that averaging the signal over a period of 1/f
N1
seconds will filter signal components at f
N1
Hz. The final result, then, is the average calculated from 31250/f
N1
samples. For example, if fN1
is set to 50 Hz, 625 samples (31250 / 50) are averaged to generate the final filtered result.
The actual f
N1
may deviate from the user-specified setting since a whole integer number of
samples must be averaged. For example, if fN1 is set to 60 Hz, 521 samples (31250 / 60 = 520.83)
6. Measurements 53
To Next Page
Loading...
