Campbell Scientific CR9000X Measurement and Control System - Manual (Page 293)

Section 8. Processing and Math Instructions
This instruction uses high precision math. A normal single precision float has
24 bits of mantissa. With high precision, a 32 bit extension of the mantissa is
saved and used internally, resulting in 56 bits of precision. Instructions that use
high precision are AddPrecise, Average, AvgRun, AvgSpa, CovSpa,
MovePrecise, RMSSpa, StdDev, StdDevSpa, and Totalize.
This instruction normally should not be inserted within a
For/Next construct with the Source and Destination parameters
indexed and Reps set to 1. In essence this would be performing a
single running average, using the values of the different elements
of the array, instead of performing an independent running
average on each element of the array. The results of this would be
a Running Average of a Spatial Average of the various Source
array's elements.
Running Average Attenuation and Phase Shift
The running average is a digital low-pass filter. As such, its output is
attenuated as a function of frequency, and its output is delayed in time. The
amount of attenuation and time delay depend on the frequency of the input
signal and the time length (which is related to the number of points) of the
running average.
Attenuation: Chart 8-1 is a graph of the signal attenuation plotted against the
signal frequency normalized to 1/(time length of running average). This signal
is attenuated by a Sinc filter with an Order of 1 (simple averaging):
Sin(
π X)/( π X), where X is the ratio of the input signal frequency to the running
average frequency (running avg. frequency = 1/Time length of running
average).
Chart 8-1 Running Average Signal Attenuation
NOTE
8-7