E1C01 09/14/2010
15:40:34 Page 9
Noise and Interference
Just how extraneous variables affect measured data can be delineated into noise and interference.
Noise is a random variation of the value of the measured signal as a consequence of the variation of
the extraneous variables. Noise increases data scatter. Interference imposes undesirable deterministic trends on the measured value. Any uncontrolled influence that causes the signal or test
outcome to behave in a manner different from its true behavior is interference.
A common interference in electrical instruments comes from an AC power source and is
seen as a sinusoidal wave superimposed onto the measured signal path. Hum and acoustic
feedback in public address and audio systems are ready examples of interference effects that are
superimposed onto a desirable signal. Sometimes the interference is obvious. But if the period
of the interference is longer than the period over which the measurement is made, the false
trend may go unnoticed. So we want either to control the source of interference or to break up
its trend.
Consider the effects of noise and interference on the signal, y t
ð Þ ¼ 2 þ sin 2pt. As shown in
Figure 1.7, noise adds to the scatter of the signal. Through statistical techniques and other means, we
can sift through the noise to get at the desirable signal information. But interference imposes a trend
onto the signal. The measurement plan should be devised to break up such trends so that they appear
as random variations in the data set. Although this will increase the scatter in the measured values of
a data set, noise can be handled by statistics. It is far more important to eliminate false trends in the
data set.
With this discussion in mind, recall the boiling point example earlier. Barometric pressure
caused interference in each individual test. The barometric pressure did not change over the conduct
of any one test. But we could discern the effect only because we showed the results of several tests
taken over a period for which the value of this uncontrolled variable did change. This is a form of
randomization in that as barometric pressure changed between tests, its effect was entered into the
data set. Randomization methods are available that can be easily incorporated into the measurement
6
5
4
3
2
1
0
0.0
0.5
1.0
2.0
1.5
Time (s)
Signal + interference
Signal + noise
Signal: y t
( ) = 2 + sin (2 t)
π
Signal
y(t)
Figure 1.7 Effects of noise and
interference superimposed on
the signal y(t) ¼ 2 þ sin 2pt.
1.3 Experimental Test Plan 9
15:40:34 Page 9
Noise and Interference
Just how extraneous variables affect measured data can be delineated into noise and interference.
Noise is a random variation of the value of the measured signal as a consequence of the variation of
the extraneous variables. Noise increases data scatter. Interference imposes undesirable deterministic trends on the measured value. Any uncontrolled influence that causes the signal or test
outcome to behave in a manner different from its true behavior is interference.
A common interference in electrical instruments comes from an AC power source and is
seen as a sinusoidal wave superimposed onto the measured signal path. Hum and acoustic
feedback in public address and audio systems are ready examples of interference effects that are
superimposed onto a desirable signal. Sometimes the interference is obvious. But if the period
of the interference is longer than the period over which the measurement is made, the false
trend may go unnoticed. So we want either to control the source of interference or to break up
its trend.
Consider the effects of noise and interference on the signal, y t
ð Þ ¼ 2 þ sin 2pt. As shown in
Figure 1.7, noise adds to the scatter of the signal. Through statistical techniques and other means, we
can sift through the noise to get at the desirable signal information. But interference imposes a trend
onto the signal. The measurement plan should be devised to break up such trends so that they appear
as random variations in the data set. Although this will increase the scatter in the measured values of
a data set, noise can be handled by statistics. It is far more important to eliminate false trends in the
data set.
With this discussion in mind, recall the boiling point example earlier. Barometric pressure
caused interference in each individual test. The barometric pressure did not change over the conduct
of any one test. But we could discern the effect only because we showed the results of several tests
taken over a period for which the value of this uncontrolled variable did change. This is a form of
randomization in that as barometric pressure changed between tests, its effect was entered into the
data set. Randomization methods are available that can be easily incorporated into the measurement
6
5
4
3
2
1
0
0.0
0.5
1.0
2.0
1.5
Time (s)
Signal + interference
Signal + noise
Signal: y t
( ) = 2 + sin (2 t)
π
Signal
y(t)
Figure 1.7 Effects of noise and
interference superimposed on
the signal y(t) ¼ 2 þ sin 2pt.
1.3 Experimental Test Plan 9