This document contains notes on David Weston's 2000 book on electromagnetic compatibility. Feel free to expand it to a general document of organized information. == 01 - Introduction to EMI and the Electromagnetic Environment == 1.1 This chapter skims different concepts: - coupling modes - EMI regulations - emissions environments 1.1.1 - EMI can cause issues from minor static, through production delay, to catastrophic malfunction - EMC is the production of equipment which is neither harmed by nor interferes with its electromagnetic environment (exhibits no effects of EMI) - EMC analysis techniques in this book are useful both in design and diagnosis. It is most important to apply them in the design phase. 1.1.2 Coupling Modes - Coupling is the travel of the emission from source into the system. - Two coupling modes are "radiated" or "conducted". - Radiated emission is predominantly magnetic (H) or electric (E) - At distance (far field), radiated H and E are in a fixed proportion. This is not true up close (near field). (->2.2.1) = Close Proximity = - Coupling also called crosstalk - Is via mutual inductance or capacitance, usually one dominates - Conducted path generally a combination of resistance, inductance, and capacitance effects - Often components resonate, with current rising and falling at constant frequencies 1.2 EMI Regulations (->ch 9) - Immunity depends on criticality of equipment and environment in question. - Requirements often provide different immunity classes. - There are many groups of regulations and bodies which provide them. 1.2.1 Military - It sounds like military requirements are hard to meet, and often exceptions are provided. - Each environment is different; solutions are tailored to the environment. 1.2.2 Commercial - Inflexible limits with very rare exceptions. - Only countries of the EU require immunity testing. 1.2.3 Unregulated - EMC may be desired for customer satisfaction, safety, or legal protection - EMI environment choices can be based on a model or an existing standard. - Specified limits are unfortunately not always close to a realistic worst-case environment (->1.3) 1.3 Environment - EMI sources are either natural, or manmade (much stronger). - Most unintentional emissions are broadband, whereas signals are narrowband. (->ch 9) - E field strength is measured in V/m (->2.1), or dBμV/m - Broadband field strength is often measured in dBμV/m/MHz or dBμV/m/kHz, with reference measurement bandwidths of 1 MHz or 1 kHz - I believe a broadband measurement is taken over the bandwidth around a given center frequency. Dividing by the bandwidth normalizes the units. See "Measurement Units of Emissions", below. - MIL-STD 461 RE02 will be used as an example regulation on broadband emission (fig 1.4, ->9.4) 1.3.1 Radiated Environment - Natural Sources - from cosmic bodies or mostly thunderstorms - thunderstorms can send noise thousands of km via ionosphere - complex random behavior over time, characterized by pulses, spikes, and continous background - fig 1.5 shows atmospheric noise increasing as frequency decreases, ranging from 0.5 to 250 μV/m at 100 kHz. Blackbody and cosmic radiation dominate above 10 - 100 MHz, at around 0.1 μV/m, with blackbody intensity rising with frequency. Blackbody radiation is model for thermal emanatains of earth surface. - Upper-limit atmospheric noise is near the RE02 limit at 100 KHz! 20 dB below at 10 MHz. Cosmic noise is 34 dB below RE02 @ 30 MHz. - Cosmic noise sources: - sky background ionization - daily variation of sky background synchrotron radiation - solar radio noise, dramatically increases with solar activity - additionally, moon, jupiter, and cassiopeia-A - An additional source of EMI is nearby static discharge. A number of avoidance methods are mentioned. 1.3.2 Radiated Environment - Manmade Noise - Emissions from a number of sources are discussed in detail - Arc welders, electric heaters, industrial-scientific-and-medical manufacturing sites, power lines, traffic, flourescent lighting etc (incandescent were insignificant), microwave ovens, hospitals, residential homes - Broadband and Narrowband units interpretation is discussed, summarized in "Measurement Units of Emissions" below (ch 9, ch 3) - Susceptibility of equipment depends on a wide variety of factors, including resonance effects, bandwidth and interfaces of equipment, and transient response of capable and structure. - Magnetic field shielding discussed in 6.2 1.3.3 Intentional Emitters - caused by radio, television, radar transmitters - interferes both with tuned receivers and equipment not intended to receive at all - may spuriously open a garage door, or crash an aircraft - Narrowband, but a single source emits on many frequencies - harmonics of fundamental continuous wave (CW) - sidebands if wavev are modulated - local oscillator frquency - broadband noise from different transmitter stages - Non-fundamental noise typically at least 70 dB "down from" the fundamental (less than?) - Strength of radiated fields from antenna depend on many factors - proximity - output power - antenna directivity - antenna height relative to measurement - proximity of reflecting or intervening material - Emissions discussed - pg 14: typical ambient field strengths in major urban center, maximum measured fields during surveys (highest examples, 4 V/m distance, 55 V/m near, geneally MHz range, some Ghz) generally much lower - pg 15: maximum typical E field at 100 m from transmitter - An RF ambient survey is often made before installing equipment. - Strong EMI may affect equipment even when the passband is far from the interference frequency (ch 12) - Low-frequency component falls more rapidly with distance than VHF,UHF (mechanism discussed ch2) - EMI prediction involves: - distance from potential sources - presence of conductive structures near emission source (walls, building, ground topology) may either reflect or shield depending on location (shielding of structures & soil discussed ch 6) 1.3.4 Conducted Noise on Power Lines - Noise is not only fields but also conducted on a transmission medium - must be accounted for (device may be well shielded but interference conducts in easily on wire) - predominantly on power lines, but also signal interfaces and grounds - susceptibility prediction requires data on extreme expected behavior on power supply - pg 16: maximum conducted noise currents measured in hospitals - conducted noise data is minimally available - conducted noise may be predicted from susceptibility test levels (ch9) or regulations calculation of this briefly described - chapter ends with references used on pg 17 = Measurement Units of Emissions = (->2.1, ->1.3, ->1.3.2, ->3, ->9) - NOTE this can be used to normalize attenuation recordings over frequency this seems to be pretty much the same thing as spectral density - E field strength is measured in V/m or dBμV/m (a narrowband unit) - Broadband field strength is measured in dBμV/m/MHz(or kHz), using a reference bandwidth of 1 MHz or 1 kHz. - Logarithmically measured fields are done as the square of the field strength. Hence dB value = 20 log10(F ratio) = 10 log10(F^2 ratio) - Converting to broadfield field strength requires knowledge of the bandwidth of the measuring instrument -- it is a unit that may be plotted over frequency, as a moving average. - The magnitude of a broadband field is invariably higher in broadband units than in a narrowband unit. - Narrowband measurements are always taken at bandwidths < 1 MHz. - A narrowband measurement will capture only one spectral line; a broadband measurement will capture many. - Narrowband and broadband measurements cannot be compared together. - Details in ch 9 Candidates for high-priority chapters: ch 6 - shielding ch 9 - measurements ch 2 - basics, one section is simple antennae construction == 09 - EMI Measurements, Control Requirements, and Test Methods == 9.1 Introduction - EMI is measured either to meet requirements or find a problem source. Problems may include failure due to self-compatibility or interference externally. - Manufacturers may not have specialized test equipment and need to measure with conventional tools. - This chapter deals with both common and specialized test equipment and errors in their use. - It is easy for measurement itself to introduce a wide variety of very significant emi errors - antenna location, ground connection, radiated pickup in leads, undamped shielding (>30 dB), antenna orientation, agc of tool (20 dB) - great care must be exercised. measurement is not intrinsically more accurate than prediction. - measurement should be used to _validate_ design, not guide it 9.2 Test Equipment 9.2.1 Oscilloscope - Common in electronic labs - Limited in frequency domain - complex waveforms must be converted to compare with frequency domain regulations - single frequencies may be measured if low harmonic distortion - in the face of many frequencies, it may be impossible to measure weak high frequencies due to oscillation of strong low frequencies; one may then use a spectrum analyzer - spectrum analyzer has much lower impedance on input, hence smaller diff voltage (50-75 ohm vs 1-10 M ohm) - or, voltage drop across shield impedance reduced becasue input impedance is across shield impedance - however, low input impedance of spectrum analyzer may load the signal when source impedance is high; additionally, input is DC coupled, so decoupling capacitor is needed on high DC voltages - Excellent for measuring transient noise in time domain - Inadvertently includes common-mode noise - converted into differential across high-impedance single-ended input - Test for this: connect both ground and probe to signal ground at same point - common-mode current on shield and center conductor develops voltage across input impedance - voltage developed across impedance of shield, and appears across the input - test will also include radiated pickup from loop formed by ground and probe - Single-ended inputs on oscilloscope and spectrum analyzer connect signal return to enclosure of tool, and thus to AC safety ground - drastically alters measured noise levels and susceptibility characteristics - useful when required to simulate the input of a similar piece of equipment, correctly terminating the input while measuring it - Wave shape and amplitude measured are changed by resonance of tool - inductance of probe ground wire and input capacitance of probe form resonant circuit - may "cause ringing on a transient", which may be falsely attributed to circuit under test - forms a low-pass filter and attenuates high-frequency signals - Reduce this: minimize ground wire inductance by making it as short as feasible, and locating it as closely as possible to the probe Or use a field effect transistor (FET) input probe with a low input capacitance - Solve many of these issues: use a differential scope probe - a solution for undesirable single-ended input characteristics - differential plug-ins do not provide highest level of common-mode noise rejection - poor balance between two inputs partly caused by use of separate test leads - scope probe locates differential input at tip! - lower pickup on test lead from radiated ambient - lower loading and resonance effects - Emulate differential input if you don't have one: combine two scope channels - two channels A and B may be put in A - B or inverted A-plus-B mode - measures differential if connected to voltage and return - Remove ground loop problem inherent in single-ended inputs: - connect ground clips together at the problem but not to ground or enclosure - removes ground loop problem inherent in single-ended inputs - common-mode voltage not totally removed due to poor balance between inputs - Check common-mode contribution: - connect the tips of both channels to the signal ground - adjust compensation trimmer capacitors in the probes for minimum level - Reduce radiated pickup by twisting probe leads together - reduces pickup loop area and tends to cancel field-induced currents - Measure noise current with current probe - differentiate between common-mode and differential-mode noise sources - isolate oscilloscope ground from signal ground - currents rather than voltages create radiation noise in some instances - especially when circuit impedances are low - often true for current flow in chassis ground - will probably need a preamplifier for scope to show current probe output - Measure relative E field with length of wire on probe tip - Use small calibrated loop and other antennas (ch 2.2.6) - combine with 50 ohm terminated input or 50 ohm external termination - will probably need a preamplifier - Average scopes are limited in upper-frequency response - Modern digital scopes (but also less so older analog) have issues with high C/M voltage - seen in CS01 and CS06 series injection test on AC lines (120-220-V) - provide erroneous measurements - Potentially solve this: place injection in neutral line - much lower C/M voltage referenced to chassis/safety ground (doesn't actually solve the issue) 9.2.2 Spectrum Analyzer - v useful for diagnostic tests, gaining acceptance in requirements and certification - similar to EMI receiver, but important differences exist - CRT display makes SA superior, EMIR generally far inferior - displays emissions in short-duration sweeptime - can see short-duration changes in amplitude - IF filter components capable of fast charge and discharge, gives Gaussian shaped filter - receiver disallowing fast rate has rectangular-shaped filter - greater selectivity due to reduced bandwidth below 3-dB down point (i think this means much less frequency smearing in output) tech details top of pg 480 - under assumptions, 60 dB bandwidth typically 200 Hz for EMIR, 1.4 kHz for SA - displays out-of-calibration if sweeptime set too short - must deal with compression and overload (same as described in preamplifier section) - Check for this & avoid: - SA and EMIR have built-in variable-input attenuator - search over frequency range to find maximum input level - adjust input attenuator so that displayed amplitude is at maximum, or lower - if magnitude changes on adjusting, tool is compressing - beacuse it automatically adjusts displayed level regardless of attenuator setting - effective gain of front end changes - HP provides useful booklet "EMI Measurement Solutions Using the Spectrum Analyzer Receiver" - Block diagram on pg 480 - it is a "swept tuned super-heterodyne receiver" - local oscillator sweeps frequency range, mixed with signal - bw captured in IF filter, peak detector identifies - some allow switching in or attaching external quasi-peak detector - it looks like the amplitude of different frequency components compared to a ref level is used to affect the video output as it scans from left to right, like oscilloscope output but in frequency domain - input frequency range typically 10 kHz to 1.3 GHz - may be as wide as 100 Hz to 22 GHz - controls allow selection of parameters - frequency span - sweeptime - resolution bandwidth / IF bandwidth - narrower gives lower noise floor, better snr - widening increases amplitude of noise but leaves isolated signals at same amplitude - shows displayed power is total in bandwidth envelope (integration) and not average - determines selectivity - IF filter shape is traced out as tuning past the signal - emissions hence are displayed with the IF filter shape - width and individuality of spectral lines changes with bandwidth - HP specifies "3-dB bandwidth" - width over which amplitude is no more than 3 dB down on max - this is comparable to what is differentiable on the display - video bandwidth - reference levels - may also have marker to read off frequency and amplitude of component - some have counter for accurate frequency measures - accuracy of display results from resolution bandwidth and frequency span - error is typically +-3% of span and resolution bandwidth - use minimum frequency span and resolution bandwidth for maximum accuracy - may have maximum or peak hold and storage of sweeps - enable capturing short-duration emissions and comparing to profile - narrowband (NB) and broadband (BB) specifications - one definition of impulsive noise refers to resolution bw of measuring instrument - when both NB and BB specified, typical bandwidths must be agreed upon - tables 9.20,9.21 (pg 594,595 9.6.1) provided for typical bandwidths when not defined - resolution affects NB or BB characterization of impulsive noise - however, they must be differentiated - FCC allows 13-dB relaxation in conducted emissions if they are broadband - many requirements specify different emissions for broadband and narrowband - type may aid tracing emissions source - e.g.: strong narrowband emission may exist in multiple lines from broadband source this contributes to broadband measurement but is from some individual traceable oscillation (power supply, circuit instability, resonant circuit) - narrowband means only one spectral component within filter bandpass :. spectrum analyzer displays each spectral component individually, and frequency may be read off of X coordinate of display - impulsive noise may also appear as spectral emissions lines - may be classified as narrowband when bandwidth set to narrowband value - pulse repetition frequency (PRF) is the frequency spacing between lines ==> pg 483