Dr. Kevin Craig  
Professor of Mechanical Engineering  
Hofstra University
and        Dr Edward H. Currie
Associate Professor of Engineering
Hofstra University

 Previous - Accuracy and Repeatability

Signals and noise often occur together and it’s important to be able to distinguish between them. Some define signal as ...“that which we want ...” and noise as “ ...that which we don’t want ...”.
While this may seem a cavalier distinction, it does clearly delineate between the two. Usually. noise makes itself known after a system has been assembled and therein lies the problem.To quantify the relationship between a signal and the associated noise the signal to noise ratio is defined as
                                  SNR = Psignal/Pnoise                                                                                             (1)
where P refers to power.
However, the dynamic range of signals is often quite high, SNR is often expressed in terms of a base ten logarithm, as follows:
                                 Psignal,dB = 10 log10(Psignal)                                                                            (2)
                                 Pnoise, dB = 10 log10(Pnoise)                                                                             (3)
And therefore,
                                 SNRdB = 10 log(Psignal/Pnoise)                                                                          (4)
Decibel Table
Figure 1: Decibel Table.

Because it is seldom, if ever, possible to completely eliminate noise in a system, the SNR is an important figure of merit to characterise the relative amount of noise present in the system. There are a number of technologies that are available to assist in the reduction of noise but it should be borne in mind that one does not generally speaking ever completely remove all the noise in a system, but rather minimizes noise to a level at which if not acceptable it is at least tolerable. With that in mind, we turn our attention to various methods and technologies with which to wage war against noise.

Typical noise sources

A large percentage of noise sources involve current flow, although thermal effects can produce unwanted potentails in the nano-microvolt a result of random electron motion, e.g., a 1kΩ resistor produces about 40μV of average noise over a bandwidth of 100 Mhz. Even diodes with a current of 1 mA will produce a noise current of 20 pA /Hz. Common sources of noise include:
• Switch Mode Power Supplies (SMPS)
• Wireless devices
• Lightning and other electrostatic discharges (ESD)
• Power Line Fluctuations
• Electric motors/generators/solenoid actuators
LEZ 1 H Frame
Custom Cartesian Robots
Quick and Easy

Click for a QUOTE!

• Computers/Peripherals
• Vehicle ignition systems
• Neon and Fluorescent lights and fixtures
• Electromagnetic radiation
• Crosstalk
• Magnetic field fluctuations
• High speed logic circuits
• High voltage circuits
• Thermal noise in capacitors, resistors and inductors.1
• Mechanical and solid state switches, e.g., relays, toggle switches, SCRs.
• Fans
• Mechanical switches, keyboards, toggle switches, push buttons, etc

Some Basic Noise Reduction Guidelines

The first rule in dealing with noise is “...don’t let it in, in the first place...”. Once noise has found its way into a system, it becomes problematical to eliminate it. The reality is that it may not be possible to remove a system’s noise entirely because efforts to do so can, in fact, introduce new noise sources. The reality is that ultimately one may have to be content with reducing noise to an “acceptable level”, meaning to a level that doesn’t interfere in a significant way with the desired functioning of the system. Fortunately, careful design and judicious application of best practices can minimize the introduction of noise, and help minimize inherent noise in the system, e.g.,

1. The input signals should be asserted using shielded cables, appropriately grounded, and isolated from power lines and
power supply distribution connections. Coax cables or any shielded cable that is used for transmission of signals should be grounded on one end,
but not both as shown in Figure 2.
Proper grounding of coax.
Figure 2: Proper grounding of coax.

2. The power supply should be carefully analyzed to ensure that its output voltages are protected from passing line
transients through to the output(s) of the supply This can be accomplished by a) judicious application of disk capacitors,
b) filtering of the line voltage supplying (e.g., using inductive chokes), the power supply, increasing the filter capacitance
of the output, etc. (Connect bypass capacitors, e.g., disk capacitors of the type shown in Figure 3, between the supply
voltage rails for each component and local ground.)
A typical disk capacitor
Figure 3: A typical disk capacitor

3. Physical isolation of signal lines from the line and power supplies based on careful attention to the associated grounding circuits

4. Using “Faraday cages” for sections of the system that are particularly sensitive to time varying electric and magnetic fields. Shielding from static magnetic fields is more difficult but one can employ magnetic shielding materials, e.g., μ metal, magnetic shielding film (MCF), etc.
A smple Faraday Cage PCB Faraday cage
Figure 4 (a) A smple Faraday Cage Figure 4 (b) PCB Faraday cage..

5. In addition to employing shielded enclosures, use, shielded connectors, e.g., BNC, SMA and type N connectors, and shielded
Popular shielded connectors
Figure 5: Popular shielded connectors.

6. Proper handling of unused analog and/or digital inputs. Typically all inputs are either grounded, or connected to an
appropriate supply voltage. Unused logic inputs can serve as sensitive antennae and can introduce pulses so fast that
they may be difficult to detect. Taking care of unused inputs is often accomplished by using pull-up or pull-down resistors as shown Figure 6
Pull up resistor should be added to unused inputs
Figure 7: Pull up resistor should be added to unused inputs

7. Ferrite beads, also referred to as ferrite chokes, can be applied to DC power supply lines to minimize high frequency noise, e.g., switching transients, while allowing DC currents to pass without unimpeded. Ferrite cores are often used on USB and power cables to suppress incoming transients.
A ferrite core is used to protect a USB input and power inputs from transients Ferrite beads are wired in series with supply lines
Figure 8a: A ferrite core is used to protect a USB input and power inputs from transients. Figure 8b: Ferrite beads are wired in series with supply lines.
8. Static fields and discharge can give rise to noise and in some cases destroy sensitive solid state devices.Switching transients must be suppressed. As shown in Figure 9, a single pulse is capable of introducing broad speactrum noise into an environment

2μ second pulse with 1 ns rise/fall times The spectrum of the pulse shown in Figure 8
Figure 9a: An example of a 2μ second pulse with 1 ns rise/fall times. Figure 9b: The spectrum of the pulse shown in Figure 8.

9. Signal paths must be chosen such that ”crosstalk”, coupling, e.g., inductively and/or capacitively, of one signal path into another is minimized. Adequate spatial separation and if possible, shielded signal paths can be very effective, in this regard.

10. As shown in red, in Figure 10, clock signals are rich in harmonics that may find their way into other parts of a circuit and manifest as noise. A magnetic scope probe, or an ordinary probe connected to a high gain broad bandpass scope can be used to ”probe” a circuit, when searching for harmonic noise intrusion into other parts of a system. Signal paths conveying a clock signal should be carefully shielded from the rest of the system. Clock sugnal lines should be carefully shielded.
red - clock signal blue - harmonics of clock signal
Figure 10: Clock signals are rich with harmonics.
red - clock signal
blue - harmonics of clock signal

11. Input signals are often connected to a differential amplifier stage, sometimes with unity gain, in order to reject any common mode signals. Twisted pair(s) can also be used over short distances and frequencies as high as 400+ MHz in order to provide common mode rejection.
OpAmp differential amplifier Transistorized differential amplifier
Figure 11 (a) OpAmp differential
Figure 11 (b) Transistorized differential amplifier

12. If signals are to be transmitted in otherwise unprotected/unshielded environments, optical fibers can be used as shown in Figure 12,
Fiber optics cables
Figure 12: Fiber optics cables can be very effective in shielding signals against electrical noise

13. Careful attention with respect to establishing adequate grounding and the avoidance of ground loops is a critical consideration. Ideally, ground lines would consist of high conductivity, copper. It must borne in mind however, that copper does have some resistance and any currents flowing in the ground lines can cause a potential drop and result in so-called ground shift that may in fact cause parts of the ground system to be at different potentials within a system. Copper braid, although offering great flexibility, has capacitance and inductance that at sufficiently high frequencies may reduce the significantly effectiveness of a ground system

14. All resistors have some capacitance/inductance and produce some noise. All capacitors have some inductance/resistance (ESL/ESR) and all inductors have resistance/capacitance, Therefore, transients which can contain very high frequency components have the potential to interfere with analog and digital circuits when they encounter resistance, capacitance or inductances, or permutations thereof.

15. Vibration can also result in the production of spurious signals and should be minimized for circuitry, connectors, etc.

16. Thermal gradients and ambient temperature can cause solid state devices to exhibit anomalous or other unwanted behavior in a system that may result in noise generation.

17. Figure 13 illustrates the use of a choke (inductor) to minimize high frequency signals from being introduced in a DIP (Dual-Inline Package)package via the power supply pin. In this example the capacitor is used to keep lower frequency noise produced from within the DIP from being introduced into the power supply circuit connections.
Inductor and capacitor
Figure 13:Inductor and capacitor can be used to filter low frequency noise at DIP power supply pin

18. Energy stored in inductors sometimes results in substantial noise production. The basic rules to remember are:
- Current cannot change instantaneously in an inductor
- Voltage cannot change instantaneously in a capacitor
Solenoid activated devices can be a substantial source of noise and associated transients when activated/deactivated. The current wants to keep flowing  through a solenoid coil, even when the switch is opened and that causes a voltage spike, i.e.noise, when the energy has no way of being dissipated. A diode, installed as shown in Figure 14, should be used to facilitate the discharge of the stored energy. Such diodes, sometimes referred to as ”snubbers”, allows the stored energy to be dissipated while reducing the associated noise.
diode to reduce noise
Figure 14: Using a diode to reduce noise resulting from changes in the inductor’s magnetic field.

19. Multiple bypass capacitors of different values connected in parallel from a power supply rail to ground can be very effective at shunting a broad range of unwanted frequencies to ground. Small capacitors (pf range) can be used for very high frequencies while larger capacitors can be used to filter lower frequencies.

20. Digital circuits are inherently noise prone, in part because of the fast rise times of the logic signals involved and therefore special attention should be given to grounding techniques and power supply noise suppression. 

21. Establish a single ground point for the system components and, if possible, do not connect it to earth ground, i.e., line power ground.
Isel AUTOMATION Belt Drive Slides
The LEZ 1 AUTOMATION STICK - up to 4m travel!

22. Use isolation transformers and/opto-couplers to isolate noise generators from noise sensitive components.

23. Switches that involve mechanical parts coming into contcat with another contact often produce noise as a result of the bouncing of the contacts prior to them reaching a stable state, as shown in Figure 15. These effects can be at least minimized and typically removed by either hardware of software techniques, depending on the type of system and the switch contacts used, e.g., push button, toggle switch, reed relays, etc. Mercury-whetted contacts are sometimes used particularly if large currents are being switched on/off. However, solid state switches have largely replaced mercury switches. Software deboucing is accomplished by introducing a delay after the first switching transition sufficient to allow all subsequent ”bounces” to end before signaling a switch transition to a stable state.
Switch contact bounce example
Figure 15: Switch contact bounce example
Analog debounce circuit example using Schmitt Trigger hysterisis Digital debounce circuit example using memory circuit
Figure 16: Analog debounce circuit example.using Schmitt Trigger hysterisis Figure 17: Digital debounce circuit example.using memory circuit

Control of noise is important to avoid modification of the signal by adding/subtracting “information” to the signal. Noise can be continuous or transient in nature and produced by RF signals (RFI), electric motors and generators in the area (EMI), radiation from fluorescent lights, improper grounding, signals in one part of a circuit finding their way into another part of the circuit (cross-talk), improper grounding of signal shields, ground loops, line/switching transients, static discharge, lightning, atmospheric transients, etc. Power cables are well known for introducing 60 cycle noise, sometimes referred to as 60 cycle hum, into sensitive parts of circuits and signal lines. It should also be noted that noise may be produced within a properly designed circuit that fails to include sufficient shielding. Simple components such as inductors, capacitors and resistors are not the “pure” devices they are often believed to be and in fact can be contributory factors in noise production, e.g., resistors are known thermal noise generators (typically, microvolts of noise per volt of applied voltage, for a 1 MHz bandwidth), inductors can serve as a virtual transformer winding, etc. Sub-components that are not designed with noise minimization carefully considered may prove to be noise sources when integrated into a system.

Single-Ended vs Differential Measurement Techniques

Electric potential, measured in volts, is determined by comparison to a reference potential, often assumed to be zero volts and referred to as“ground”. Two basic techniques are employed in making measurements of potential whether DC or AC.

Single ended measurements are made with respect to ground and differential are made by measuring the potential difference between two arbitrary points. nether of which are ground.
Differential measurement in the presence of in-phase noise
Figure 18: Example of differential measurement in the presence of in-phase noise.
Differential measurement in the presence of arbitrary noise
Figure 19: Example of differential measurement in the presence of arbitrary noise.

Differential inputs provide a more stable reading when EMI or RFI is present, and therefore, it is recommended to use differential signal handling whenever noise is generally a problem. This is especially true when measuring thermocouples, strain gauges and bridge type pressure sensor inputs. inputs, since they produce very small signals that are very susceptible to noise. Single-ended inputs are lower in cost, and provide twice the number of inputs for the same size wiring connector, since they require only one analog HIGH (+) input per channel and one LLGND (-) shared by all inputs. Differential inputs require signal HIGH and LOW inputs for each channel and one common shared LLGND. Single-ended inputs save connector space, cost, and are easier to install.

[1] Whitlock, Bill. Understanding, Finding, and Eliminating Ground Loops