Fullscreen HTML5Med Res (??MB)HTML4

Circuit Surgery Regular clinic by Ian Bell Interference and noise R ecently, Michael Lamontagne posted a question on the EEWeb forum concerning a possible ground loop problem when measuring an op amp circuit with an oscilloscope. Rather than looking at this specific case in detail (partly because we are not certain enough about exactly how everything was wired up) we will take a quick look at unwanted signals in general, specifically noise and interference. Noise and interference Unwanted signals present in electronic systems get often referred to as ‘noise’, although we can be more precise with our terminology. In audio system it may make its present felt as hiss, hum, buzzes and crackles. In sensor systems, it limits measurement of low-level signals and degrades accuracy of measurement. Noise may already be present as part of an input signal (eg, it may come from a sensor along with the wanted sensor signal), or it may be introduced by the circuitry (eg, amplifier) used to process the signal. For example, all resistors generate random electrical noise – you cannot prevent this, it is part of their basic physics. Circuits also produce nonrandom unwanted changes to signals – distortion, which we will not be looking at in this article. Unwanted signals may also come from outside or elsewhere in the system, coupled or picked up inadvertently and added to the signal being processed – this is often called ‘interference’ to distinguish it from random noise. Crosstalk is interference between multiple channels or signal paths. Random noise Radom noise causes the instantaneous value of a signal to deviate from its ‘true’ value, with decreasing probability for larger deviations. The specific mathematical probability function depends on the type of noise, but may be the well-known Gaussian or normal distribution, familiar to all statisticians. There are a variety of types of random noise generated within electronic circuitry; these include thermal noise, shot noise, flicker noise, and avalanche noise. This noise generated is fundamentally due to the discrete nature of electricity at the atomic level – electric charge in circuits is carried in packets of fixed size (eg, electrons). The waveform in Fig.1 shows a random voltage variation with time. This gives us some simple insight into what noise ‘looks’ Fig.1. Random signal (noise). 54 like, but in general, plotting random noisy signals against time is not particularly useful. When dealing with noise we often need to look at the spectrum of the signal – the variation of signal level against frequency. Unwanted signals may look like random noise (eg, on an oscilloscope), but they can actually have significantly different characteristics. For example, the noise on the power supply of a digital circuit may look random, but a look at the spectrum will show that certain frequencies, related to the system clocks will be dominant. Pure random noise has a smooth continuous spectrum – for example, that shown in Fig.2. Random noise may be classed according to the shape of its spectrum. White noise has the same power throughout the frequency (f) spectrum, whereas 1/f noise (or pink noise) decreases in proportion to frequency. For 1/f noise there is the same amount of noise power in the bandwidth of say 100Hz to 1kHz, as there is in 1kHz to 10kHz, whereas for white noise there would be 10 times as much power in the bandwidth 1kHz to 10kHz as 100 to 1kHz because it is 10-times larger. Amplifiers (and other circuits) typically exhibit a mixture of pink and white noise, with pink noise dominating at low frequencies. The frequency at which the dominant noise component changes between pink and white noise is called the corner frequency or noise corner (see Fig.2). The fact that the components in any electronic circuit or system generate random noise means that there is always a certain level of noise, even with no signal present. This is known as the noise floor, which is important because the circuit cannot meaningfully process input signals that are smaller than the noise floor. As noise floor relates to noise within the circuit, this is different from noise within the input signal. If the properties of the required signal are known then there are techniques which can extract signals that are smaller than noise present within the signal. The difference between the signal and the noise is often important; it is expressed as the ‘signal to noise ratio’ (SNR), usually in decibels (dB) and based on the ratio of noise power (hence the v2 terms in the equation). Larger values indicate better performance. 𝑣𝑣%& 𝑣𝑣% 𝑆𝑆𝑆𝑆𝑅𝑅!" = 10 𝑙𝑙𝑙𝑙𝑙𝑙#$ * & , = 20 𝑙𝑙𝑙𝑙𝑙𝑙#$ . / 𝑣𝑣' 𝑣𝑣' Fig.2. Typical spectrum of amplifier noise. Practical Electronics | January | 2021 points. Careful circuit design and construction can greatly reduce these problems. S ource of interf erence; eg, digital clock line Capacitively coupled interference (see Fig.3) can be reduced using screening, which C ircuit 1 C ircuit 2 effectively grounds the interference coupling capacitance. Screening is implemented using a) coaxial (screened) cable to link (for example) a Ground sensor to a circuit, and by enclosing the sensitive circuits in a grounded screening box. The source of interference can also be screened to reduce A lternative connection C ircuit 1 ground vi a screen; its effect on other circuits. Choice of where the usef ul f or ‘ f loating’ sensors screened cable is grounded may have an effect C ircuit 1 C ircuit 2 C ircuit 1 on circuit performance due to the possibility S creened cab le of creating grounding loops if the screen/signal return path is grounded at both ends (more on b) Ground this a little later). For differential signals, we can also use screened wires – the two signal wires form a twisted pair and are enclosed by the screen Fig.3. Capacitively coupled interference. (see Fig.4). Here, grounding at both ends is less of a potential problem as the ground does not carry the signal. Magnetic interference is worse when physically large loops Here, vs is the rms signal voltage and vn is the rms noise voltage. occur in the circuit (see Fig.5) but can be reduced by avoiding When using or quoting SNR values, the bandwidth (range of such loops – for cables, use of twisted pairs of wires is an effective signal and noise frequencies considered) should be quoted approach. For PCBs, use a ‘ground plane’ on one side of the because noise power is frequency dependent and noise may be board and for ribbon cables each signal can be given an adjacent present well outside the range of signal frequencies of interest. ground wire. Circuits can be shielded against magnetic fields, but this is not used as commonly as shielding for capacitive Interference coupling as it requires special high-permeability materials such External signals may get into your circuit through electrostatic, as Mu-Metal. These materials are expensive and since they may electromagnetic and magnetic coupling. In electrostatic coupling, need to be quite thick the screening may be bulky. a high-impedance part of your circuit acts like one plate of a Power supplies can be a significant source of unwanted capacitor; in magnetic coupling, a loop in your circuit acts like signals. The circuit in Fig.6 illustrates how supply resistance the secondary of a transformer; and in electromagnetic coupling, leads to errors or interference. The supply current taken by parts of your circuit act like antennas. Mains hum signals (at a circuit causes a voltage drop across the supply wiring; so, 50/60Hz) and radio frequency interference from other electronic for example, the ‘ground’ voltage at each subcircuit will not systems such as phones and computers are obvious examples actually be the same (as measured from the same reference of external interference. The amount of external noise a circuit point). Fig.6 is a simplification – every branch of the supply is subject to will vary greatly depending on its location. The wiring will have resistance and hence cause voltage drops. problem will tend to be worse close to things like power lines, Supply and group voltage drops may cause problems if we are electrical machines and transmitters such as mobile phones. trying to accurately process voltage signals. If the supply currents Signals in one part of your circuit can find their way into other are constant, then the error will be an offset (DC error) but the parts of the circuit where they cause problems. A common example ground voltage is not necessarily constant; as the supply current of this is the clock of a digital section of a mixed analogue and of one subcircuit varies then the supply voltage drop and hence digital circuit getting into an analogue section, via the power the error voltage at this and other subcircuits fluctuates (this is supply lines or by capacitive coupling to high-impedance sometimes called ‘ground bounce’). This problem can be very significant, for example, when one subcircuit has a digital clock C ircuit 2 signal that is coupled via the supply into a sensitive amplifier. This situation can also occur in the mains wiring – along the C ircuit 1 equipment power cables and the wiring between supply outlets. S creened C ab le A solution to supply and ground noise is to wire connections Ground separately to a single point rather than using the same pointto-point wire for all the connections (see Fig.7). This approach applies equally to the wiring inside the cabinet of an instrument Fig.4. Screened differential signal. and to the supply connections on an integrated circuit. It may be more difficult to achieve with mains wiring, as it cannot be easily changed and doing so may be very dangerous. C ircuit 1 Magnetic f ield C ircuit 2 Rsupply2 Rsupply1 S upply C ircuit 1 C ircuit 1 C ircuit 2 C ircuit 2 Rground1 Fig.5. Large wiring loops (upper schematic) make a circuit susceptible to voltages generated by magnetic fields. Reducing loop size (lower schematic) helps combat the problem. Practical Electronics | January | 2021 Rground2 S upply Fig.6. Supply wiring resistance causes voltage shifts and noise due to supply currents. 55 Fig.7. Supply wiring to reduce noise. S upply C ircuit 1 S ignal Fig.9. Triaxial connector used with triaxial cables for guarded connections. Guard C ircuit 2 S upply Ground loops When two circuits, sub-circuits, instruments, or other equipment are grounded at two separate points on a ‘ground bus’ we have a situation know as a ground loop (or earth loop) (see Fig.8). The ground bus may be a circuit board track, the chassis of the equipment, point-to-point wiring, or the mains earth connected at different outlets – many people have suffered unnecessary levels of hum in their Hi-Fi systems due to earth loops! The ground loop will pick up magnetic interference, probably mains hum and may also act like an antenna picking up radio frequency interference (RFI). Large loops will make the problem worse. Ground loops are a particular problem when two or more mains-powered systems (such as lab instruments and sensor circuits) are separately earthed and connected together. It is also possible for mains leakage currents to cause currents to flow in the earth (eg, shields) of connections linking equipment. Leakage currents can flow through parasitic capacitances and equipment ground, for example in transformers and EMI filters. The interference causes a current IL to flow in the ground loop, which in turn causes an additional voltage drop (ILRG) across the resistance (RG) of the ground connection between the equipment or subcircuits. The solution to ground loops is to avoid them by using a single grounding point (Fig.8). Use of differential signals, only connecting screens at one end, use of very low resistance ground connections between circuits (reducing RG), and signal isolation using transformers or opto-isolators also help minimise ground loop problems. Power isolation transformers may also help with mains wiring. Again, it is worth pointing out that some potential ‘solutions’ related to mains wiring, such as disconnecting earths, could be lethal. O uter shield/ chassis/ ground/ signal return T riax ial C ab le S ignal S ensor Ground Fig.10. Guarded signal connection. of shielding. The inner shield is connected to a signal of equal voltage to the signal provided by a unity-gain amplifier (see Fig.10). This means that there is a zero-voltage difference between the signal and inner shield, so the leakage currents (and capacitance effects) are minimised. The outer shield is usually grounded and provides interference protection for the guard signal. IM – IL C ircuit 1 IM RS A ) IM–IL RG C ircuit 2 VM IL Signal guarding Signal guarding is concerned with getting the most out of screened cable connections, particularly when connecting very low-level signals from high-impedance sources to high-precision circuits. In such cases, effects such as leakage currents in the cables and cable capacitance can cause significant errors. Signal guarding uses triaxial cables and connectors (see Fig.9), which have an inner conductor, carrying the signal of interest and two layers x1 Guard RS IM IM RC VM IL B ) IM S ignal Guard RS x1 IM VM IL Ground IM C ircuit 1 IM RC 1 C ircuit 2 0 v RS IM I L2 VM Guard RC 2 x1 VM I L2 Ground Fig.8. Ground loops: currents induced in ground loops cause voltage drops which introduce noise (upper schematic). Using a common ground point can eliminate the loop (lower schematic). 56 Fig.11. Guarded resistance measurement: a) non-guarded setup, b) non-guarded equivalent circuit, c) guarded setup, d) guarded equivalent circuit. Practical Electronics | January | 2021 As an example of how guarding works, consider the schematic in Fig.11a, for which an equivalent circuit is shown in Fig.11b. Here we are trying to measure the resistance of a sensor (RS) which has a very high resistance value and therefore leakage through the cable insulation resistance RC is significant. We apply VM and measure IM – this should give the value of RS as VM/IM, but if actually gives us this parallel combination of RS and RC due to the leakage current IL. Using a guard (Fig.11c and Fig.11d) means that the voltage across RC1 between the inner conductor and guard is zero and hence no leakage current flows. The buffer amplifier has no difficulty in supplying the guard-to-ground leakage current IL2 and this does not disrupt the measurement. ESR Electronic Components Ltd All of our stock is RoHS compliant and CE approved. Visit our well stocked shop for all of your requirements or order on-line. We can help and advise with your enquiry, from design to construction. Vibration and chemistry Systems processing low-level signals are also prone to a variety of forms of interference-based noise and errors other than electrically/magnetically coupled signals, including mechanical and electrochemical effects. Movement and vibration of cables can create electric current through the triboelectric effect – charges created due to friction between a conductor and an insulator. Low-noise cables are available for situations where this may be a particular problem. Making sure that cables are well supported and not subject to vibration or large temperature fluctuations helps reduce this effect for any cable. Movement can also generate unwanted signals through the piezoelectric effect, which occurs when mechanical stress is applied to insulators. Unwanted signals due to movement and mechanical stress are sometimes called microphonic effects, because if the signal is listened to, the movement of (for example) a cable will be audible. Batteries create electric current through electrochemical effects. Similar processes can occur if contaminants are present on PCBs and terminals. Variations in humidity can affect sensor systems with very high impedances. Contaminants (handling cables and connectors) can also reduce insulation resistance leading to errors. 3D Printing • Cable • CCTV • Connectors • Components • Enclosures • Fans • Fuses • Hardware • Lamps • LED’s • Leads • Loudspeakers • Panel Meters • PCB Production • Power Supplies • Relays • Resistors • Semiconductors • Soldering Irons • Switches • Test Equipment • Transformers and so much more… Monday to Friday 08:30 - 17.00, Saturday 08:30 - 15:30 Station Road Cullercoats North Shields Tyne & Wear NE30 4PQ Tel: 0191 2514363 sales<at>esr.co.uk www.esr.co.uk Your best bet since MAPLIN Chock-a-Block with Stock Visit: www.cricklewoodelectronics.com O r phone our f riendly kn owledgeab le staf f on 020 8452 0161 JTAG Connector Plugs Directly into PCB!! No Header! Components • Audio • Video • Connectors • Cables Arduino • Test Equipment etc, etc No Brainer! Our patented range of Plug-of-Nails™ spring-pin cables plug directly into a tiny footprint of pads and locating holes in your PCB, eliminating the need for a mating header. Save Cost & Space on Every PCB!! Solutions for: PIC . dsPIC . ARM . MSP430 . Atmel . Generic JTAG . Altera Xilinx . BDM . C2000 . SPY-BI-WIRE . SPI / IIC . Altium Mini-HDMI . & More www.PlugOfNails.com Tag-Connector footprints as small as 0.02 sq. inch (0.13 sq cm) Practical Electronics | January | 2021 Vi s i t o u r Sh o p , Ca l l o r B u y o n l i n e a t : w w w .c r i c k l e w o o d e l e c t r o n i c s .c o m 020 8452 0161 Vi s i t o u r s h o p a t : 40- 42 Cr i c k l e w o o d B r o a d w a y Lo n d o n NW2 3ET 57