Noise
1
Introduction
Noise is a variation in the measurement of a variable that does not reflect real
changes in the variable.
Noise may come from different sources. Whatever the source, noise distorts the
original signal
2
Effects of Noise
1. Reduces precision, accuracy (sometimes) and resolution of the measurement.
Example: Resolution-
a temperature sensor generates 10 µV/°C and a good microvolt meter is
capable of reliably measuring 1 µV.
This, would imply a resolution of 0.1°C
Suppose noise (from all sources) is, 2 µV
Only signals above the noise levels are useful. So, any signal below 2 µV is
useless.
The resolution cannot be more than 0.2 °C.
3
2. Introduces errors in control systems.
To a controller, fluctuation in the process variable caused by noise are
indistinguishable from fluctuations caused by real variable. Noise in a
process variable will be reflected in the output of the controller
4
Types and Sources of Noise
• Many sources and many types of noise
We will distinguish between two broad types
1. INTERNAL: Noise internal to a sensor/amplifier/circuit
2. EXTERNAL: Interference noise (external).
1. Internal noise
elimination is not an option since this type of noise cannot be entirely eliminated
it must be reduced as much as possible
When amplification occurs, noise is also amplified and the amplifier itself can
add its own noise.
5
Internal Noise
Internal noise is generated due to many effects in the circuit elements. Some of them are
avoidable, but some are intrinsic.
a) Thermal or Johnson noise
b) Shot noise
c) Pink noise
a) Thermal or Johnson noise:
One of the basic sources of the noises is the resistor R.
Causes: This is generated due to the chaotic thermal movement of the charges.
6
The noise power density is given by:
V2 k is the Boltzman constant (k=1.38x10-23 J/K),
en2 4kTRf T is the temperature in K,
Hz R is the resistance in
f is the bandwidth in Hz.
Can be reduced by the decrease of R, by limiting of bandwidth and of course by
control of temperature.
Fairly constant over wide range of frequencies. So, it is called a WHITE NOISE
A white noise is a random signal with constant power spectral density over
the entire frequency range.
7
b) Shot noise:
This type of noise is produced in electronic/photonic devices with semiconductor
junctions.
Causes: Produced in semiconductors when charges flows past the potential
barrier by random collisions of electrons and atoms that causes fluctuations in
the real signal current
q is the electronic charge, 1.6x10-19 C
I is the DC current, Amp
f is the bandwidth in Hz.
Shot noise also is the white noise as it is fairly constant over a wide range of
frequencies.
8
c) Pink noise:
These are 1/f type noises (sometimes called low frequency noise, flicker noise or
excess noise), as they are inversely proportional to the frequency.
So, unlike white noise, this type of noise has higher energy at low frequencies.
This is a problem for the sensors that operate at low frequencies.
There are many sources of these noises, most of them unknown.
9
Noise Evaluation
Due to the random nature of the noise, usually they are described not by the
amplitude level, but by the spectral density, S(f)
Vn2
S( f )
f
Often the noises are characterized by the SNR factor –signal to noise ratio.
signal power
SNR
noise power
V
or, ( SNR) dB 20 log10 s
Vn
10
2. External (Interference) noise
The largest sources of noise
Originates outside of the measurement device and is coupled to it.
Sources of interference can be many:
Electrical:
Electrical power supply lines, electrical machines, lighting equipment,
commutating devices, radio communication transmitters, electrical/
atmospheric discharges or cosmic noises etc.
Mechanical:
Vibrations, gravitational forces, acceleration etc.
Environmental:
Temperature variations, ionization sources, humidity variations,
chemical sources etc.
11
Electrical Interference Noises:
This is the most important of all interferences.
In some cases, a noise is easily identifiable.
Example: a common noise in electrical system, especially those that
contain long wires, is a 100 Hz noise, This is due to power lines.
This type of noise is also a good example of a time-periodic noise.
When transient or random- is almost impossible to identify and to
correct.
12
Electrical interference noise can enter the measurement system in four
ways:
Through capacitive or electrostatic coupling
Through inductive coupling
through direct conductive coupling
By radiation (Electromagnetic interference)
13
Interference- Capacitive Coupling
Any two wires, any two connectors will produce a stray capacitance that
can cause coupling.
Capacitances are small - impedances are high.
Capacitive coupling is specially a problem at higher frequencies.
14
Solution:
the sensor must be electrostatically shielded from the sources that might
couple noise.
An electrostatic shield is usually a thin conducting sheet, sometimes a
conducting mesh, which envelopes the protected area and is grounded
In effect this shorts the noise source to ground.
Example:
This also creates a new capacitance between the protected device and
ground.
But, the noise signal is zero, as there is no source is associated with the
shield.
15
Interference- Inductive Coupling
External devices such as power cables and equipment, fluorescent
lighting and circuits operating at audio or radio frequencies generate
noise through inductive coupling.
If measurement cables are close to such external cables or equipment, a
significant mutual inductance M can exist between them which
generates noise.
This is a particular problem between current carrying conductors.
100 Hz noise from power liner usually links to sensors through inductive
coupling
16
Solutions:
At high frequencies, a conducting shield just like the electrostatic shield
should envelope the source. High frequency field will create eddy
current on conducting shield and be grounded.
• The use of coaxial cables is such an example.
If the noise signal is very low in frequency, a magnetic shield is necessary.
• Usually a thick ferromagnetic shield (box) that envelopes the
protected device to guide low frequency (or DC) fields away from
the sensor.
17
Noise due to inductive coupling is also substantially reduced if twisted
pair cables are used.
• In the first loop, wire A is closest to the noise source and has a
voltage V1 induced in it. The wire B has an induced noise voltage V2.
For loop 2, wire B is closest to the noise source and has an induced
voltage V1. now the wire A has an induced voltage V2.
• For these two loops, total voltage induced in wire A and in wire B is
the same, V1+V2. This pattern continues for all the loops and hence
the two wires have an identical voltage induced in them.
18
Interference- Conductive Coupling
A typical example of the conductive coupling of interferences is the current in
the common wire: supplying or grounding.
Especially in the case of connection of two different grounding points the inter-
ground difference of potentials can be dangerous. This is called multiple
earths/grounds
Often the solution would require that the circuit be floating.
19
Electromagnetic Interference (EMI)
Any conductor carrying an ac current is in effect a transmitting antenna.
Any other conductor is in effect a receiving antenna.
If that conductor is part of a loop, a current will be induced in the loop.
Two types:
External EM sources: The EM sources are external to the
measurement system, e.g. radio transmitters and mobile telephony
communication. They are the largest sources of EMI.
Internal EM sources: here, different parts of a high-frequency
measurement system (or even different parts of a chip) act as
transmitters and receivers
20
Solution:
Reduction of this source relies extensively on reduction of lengths of
wires and on reduction of size (area) of loops.
Twisting of the two wires leading to a device together to reduce the area
of the loop they form.
Coaxial cables can reduce or eliminate most radiated interference
In most of the cases, the EMI or RFI can be reduced by filters in hardware
and advanced algorithms in software.
21
Other Sources of Noise
Example: any junction between different metals becomes a
thermocouple and introduces a signal in the path.
This may affect the reading of the sensor and is called
Seebeck noise.
It may not be a big problem in most cases but it can be when
sensing temperature.
The issue of noise is both difficult and ill-defined.
Often finding the source of noise will depend on sleuthing
work and on experimentation.
22
Even After All These Steps, If Noise Is Still a Problem
1. Hardware Signal Filtering: Characteristics of different types of filters are
shown in a previous class.
23
Hardware Signal Filtering
Figure: Passive Low-pass filter Figure 2: active low-pass filter
24
Even After All These Steps, If Noise Is Still a Problem
2. Signal Processing:
It is concerned with improving the quality of the signal at the output of a
measurement system and to attenuate any noise in the measurement signal
that has not been eliminated by careful design of the measurement system
as discussed before. Software implementation.
25
