Understanding Noise
In a previous post I covered the fundamentals of test engineering — sensors, continuous and discrete signals, and data quality. This post will cover topics related to electrical noise which affects the accuracy of data gathered from sensors and instruments.
noise — unwanted signals that are coupled with information carrying signals and cause errors in measurement or data transmission.
Electrical noise is the bane of test engineers. Noise can be constant, or variable based on the source and can vary in frequency and amplitude. While some types of noise are present everywhere and unavoidable (called internal noise), there are many external sources of noise which are coupled onto signal wires. This concept of noise superimposed onto an underlying signal is shown in the illustration below.
The challenge of a test engineer is eliminating the noise component without affecting the underlying signal. Elimination or mitigation of noise is critical in designing accurate and robust test systems. Noise can be reduced or eliminated through hardware and software solutions. But before discussing how noise can be managed, it is important to understand why it exists and how it gets coupled onto our signals.
Why does noise exist?
Ask why enough times and you end up in the quantum realm and start questioning the nature of reality. However, engineering is an applied science. We don’t question the nature of the rules governing this universe, we simply exploit them to achieve our practical aims, such as designing test systems. What follows is a brief summary of various types of internal and external noise, and the ways in which external noise is injected into measurement systems.
The main types of internal noise are:
- Thermal noise
- Shot noise
- Flicker noise
- Burst noise
Thermal noise (Johnson noise): Johnson noise is the electronic noise generated by the thermal agitation of the charge carriers inside an electrical conductor at equilibrium, which happens regardless of any applied voltage. What this means is that a simple resistor sitting on a table has voltage across its terminals! The mean value of this thermal noise voltage is zero, but the RMS value is non-zero.
The formula for Johnson noise is given below.
T= temperature, R=resistance, F=frequency band of the measuring instrument. A more detailed explanation of the formula and calculations can be found in a textbook somewhere. The intent here is to understand the phenomena and associated variables.
The magnitude of the Johnson noise is on the order of nV to uV which is not insignificant in many sensitive applications. Johnson noise can be reduced by reducing R, and ΔF. Reducing T is not practical in most applications and does not have any affect until cryogenic temperatures are reached.
Johnson noise is irrelevant in DC applications (such as measuring voltage using a resistor divider) since it's a zero-mean signal. More on Johnson noise: https://physicsopenlab.org/2016/10/07/johnson-noise-and-boltzmann-constant/
Shot noise: Shot noise arises because charge and energy are quantized. Electrons and photons leave sources and arrive at detectors as quanta; while the average flow rate may be constant, at a given time there are more quanta arriving than at another instant.
Although current is described as the ‘flow’ of electrons, the flow is not uniform over time. At one instant 100 electrons might flow past a point, and in another instant, it might be 80 electrons. This fluctuation will manifest itself as noise on the current signal since the magnitude of current is related to the quantity of charge carriers in motion.
Shot noise is only present when a DC current is flowing in the circuit. The magnitude of shot noise is on the order of pico-amps. Shot noise is not relevant in the vast majority of industrial test system applications.
Flicker noise: Flicker noise is one of the mysteries of the universe. I’m not going to attempt an explanation here, but only describe the behavior and effects. It is present in all active and some passive devices and is associated with dc current. It is also called 1/f noise because it is proportional to the inverse of the signal frequency. This means that flicker noise dominates at low frequencies, and op-amps include 0.1–10 Hz plots quantifying flicker noise at the input pin. More details in this excellent blog post: EEVblog #528 — Opamp Input Noise Voltage Tutorial — EEVblog
The magnitude of this noise is typically on the order of a 100nV to 1 uV. Flicker noise cannot be filtered out using line filters or other external devices since it is inherent to the op-amp. If flicker noise is going to dominate in a given application, then specialized equipment and techniques must be used. But there is no free lunch, and flicker noise will be traded off for other kinds of noise or distortions in the signal. More information here: Understanding and Eliminating 1/f Noise | Analog Devices
Burst noise: Unlike other forms of noise burst noise is transient. It consists of sudden step-like transitions between two or more levels. The burst noise steps may be as high as several hundred microvolts, at random and unpredictable times and lasts for several milliseconds. Burst noise is an inherent property of active electronic components and is prevented by choosing high quality components. Transient bursts in the signal can be removed by inline filters.
Internal noise is inherent to electronic components and a consequence of material properties. Internal noise is reduced by using a narrow frequency bandwidth and high-quality components. Effects of internal noise are on the order of nano to micro volts and not relevant in most industrial applications.
External noise: External noise is noise induced/injected into a signal/circuit due to interference from an external source. This is called electromagnetic interference (EMI) or radio frequency interference (RFI). EMI acts on circuits and signals wirelessly, that is no physical connection is necessary for noise to be injected into a measurement system. This is due to the nature of electromagnetic fields which act across a distance.
Typical sources of EMI are devices which produce quick changes in voltage or current or harmonics, such as:
- Large electrical motors
- Fluorescent lighting tubes (because of the fast switching within the electronic ballast)
- Solid-state converters or drive systems
- Large consumers of current such as welders
External noise can couple onto signal wires through the following mechanisms:
Capacitive coupling: A capacitor can be formed out of any two conductors separated by a dielectric. As a quick recap a capacitor is a simple device consisting of two conductive plates separated by a dielectric medium. A capacitor stores energy in the form of an electric field that develops between the two conductive plates with opposite polarities. Capacitance (C) is a measure of a capacitor’s ability to store energy. Larger plate area increases capacitance, since more surface area is available for the electric field flux. Closer plate spacing also increases capacitance since the electric field force between the two plates is higher/stronger.
A parasitic capacitor is formed between adjacent wires in a wire bundle or traces on a PCB if they are carrying different voltages. This parasitic capacitor will act as a voltage divider for high frequency AC signals and induce a noise current into the signal wires. A simple scenario is that of a low voltage dc signal wire being adjacent to the AC mains cable.
Inductive coupling: Similar to capacitive coupling but with magnetic fields instead of electric fields. As a quick recap an inductor is a simple device consisting of a coil of wire. An inductor stores energy in the form of a magnetic field which acts to oppose any change in current through the inductor. Unlike a capacitor an inductor only works in a live circuit. Inductance (L) is a measure of an inductor’s ability to store energy. It is directly proportional to number of turns, and coil area, and inversely proportional to coil length.
A cable carrying AC currents will generate a corresponding variable magnetic field around itself. This variable magnetic field will induce a noise voltage in any adjacent signal wires following the principle of a voltage transformer.
Conducted noise: Noise can be conducted into a circuit over signal wiring. One particularly common problem is ground loops. A ground loop is created when devices are grounded at different points in a system, and also share a common ground, offering multiple paths to ground thus creating a loop. A ground loop is a physical loop created by the ground paths of interconnected devices in a system, as seen in the image below.
Devices 1 and 2 use a common ground wire in their communication architecture (e.g., USB and RS-232). This creates a ground loop because each device also has a ground path internally through its power supply. The problem with ground loops is that they act as antennas and pick up coupled and induced noise. Since the ground loop is low impedance large noise currents will flow in the loop from very small differences in ground potential at various points in the distributed grounding system. Ground loops are particularly troublesome when long runs of cable are involved. This is because resistance increases with wire length, and long runs of wire mean that there will be potential difference between earth ground and circuit/signal ground.
External noise is injected into signals due to improper wiring or electromagnetic coupling. This noise can be eliminated or mitigated through proper wiring and circuit design practices.
Noise suppression strategies will be discussed in a future article.
References:
Breaking ground loops to reduce transmission errors — Electronic Products
