Home | Glossary | Books | Links/Resources |
| EMC Testing | Environmental
Testing |
Vibration Testing |
AMAZON multi-meters discounts AMAZON oscilloscope discounts It is clear that in-situ testing can only be performed by experienced organizations that have a wide knowledge of both testing, use of the radio spectrum and judgment of test results in the light of equipment use at several locations. If in-situ testing is allowed in general it may create situations that can be considered dangerous. There are a number of differences between testing equipment for EMC at a test laboratory, and testing systems or installations in their place of use, i.e. in situ. A fair amount of information (including the applicable international standards) is available to describe the former case, but much less has been written to cover the special requirements of in-situ tests. This section will briefly cover the standard test methods and will concentrate on those aspects and problems that distinguish their in- situ variants. The important issue of the test plan has already been discussed. There is some debate as to whether testing in situ can be used to represent the compliance status of systems which are not tested on a test site. CISPR 11 ~ states explicitly that Measurement results obtained for an equipment measured in its place of use and not on a test site shall relate to that installation only, and shall not be considered representative of any other installation and so shall not be used for the purpose of statistical assessment. In contrast, CISPR 16-2 suggests that where a given system has been tested at three or more representative locations, the results may be considered representative of all sites with similar systems for the purposes of determining compliance. The US FCC Rules have a similar condition. But in any case, for compliance with the EMC Directive via the Article 10.2 route, a manufacturer may write his TCF around whatever degree of in-situ testing he wishes, if his chosen Competent Body agrees. EmissionsCISPR instrumentation and transducers The measuring receiver The fundamental instrument for measuring radiated and conducted RF emissions is the measuring receiver. The requirements for this instrument are defined in CISPR 16-1, "Specification for radio disturbance and immunity measuring apparatus and methods: Part 1: Radio disturbance and immunity measuring apparatus". The CISPR specification includes:
These parameters are summarized below, which section the requirements vs. frequency. Four frequency bands are defined. Bands A and B (9kHz to 30MHz) are usually applied to conducted emissions, and bands C and D (30MHz to 1000MHz) are usually for radiated emissions; but this is not universally fixed, and there are circumstances where radiated emissions are measured below 30MHz and conducted above this. At the time of writing, no CISPR requirements have been published for emissions measurements above 1GHz. This will change in the future, as the subject is being actively considered and various draft requirements have been circulated for comment. Other standards, particularly for military and aerospace applications, have been requiring tests above 1GHz for some time. ----- CISPR measuring receiver requirements Band A, Band B, Band C, Band D, Frequency range 9kHz- 150kHz- 30MHz- 300MHz- 150kHz 30MHz 300M Hz 1000M Hz ..... 6dB bandwidth 220Hz, 9kHz, 120kHz , Sinusoidal input accuracy +2dB ..... Input impedance and VSWR 50~, < 2:1 with no attenuation, < 1.2:1 with > 10dB --- Two characteristics which are important for any receiver, sensitivity and frequency accuracy, are not defined in the CISPR specification except indirectly. The sensitivity requirement is stated as a maximum error (1 dB) that may be introduced by background noise or spurious responses. No specification is given for frequency accuracy, although in practice good quality receivers will have more than adequate accuracy for most testing purposes. The high cost of measuring receivers is at least partly due to the unusually severe specification that is placed on their linearity and amplitude accuracy. An EMC test is inherently a measurement of a quantity with unknown parameters; particularly, the bandwidth and nature of modulation of the signal(s) being measured are unknown. They may be narrowband, stable, single frequency signals such as microprocessor clock harmonics, or they may be broadband impulsive noise that spreads over a very wide frequency spectrum such as from DC motors. In between is quasi-impulsive noise with periodic frequency components such as from switch mode power supplies, digital data signals or variable speed inverter drives, or mains-related broadband pulses such as from rectifiers and thyristor switches. A real EUT is likely to have a mixture of several of these, as also is the environment in which it is measured (if not in a screened test chamber). The receiver must be able to measure all of these components, simultaneously if necessary, without degradation in its accuracy or linearity. This demand is not placed on any other type of radio receiver. Two particular aspects of its specification are peculiar to a CISPR emissions measuring receiver: its bandwidth, and its detectors. Bandwidth ----- the preferred CISPR bandwidths for each of frequency bands A-D. The bandwidth specification is important because, for an interference signal whose spectrum is wider than the measurement bandwidth, the indicated signal is proportional to the measurement bandwidth. Since the interference spectrum is generally unknown, control of the measurement bandwidth is needed. Military EMC tests historically have dealt with this issue by requiring tests at different bandwidths and consequent reporting of results as "narrowband" and "broadband" emissions. CISPR takes the opposite approach of defining a single bandwidth for all measurements in a particular frequency range t. The bandwidth is actually defined as the difference in frequency between upper and lower-6dB points in the receiver's frequency response. --- Measurement bandwidth Detectors CISPR 16-1 defines three principal detector types, peak, quasi-peak and average. For the present, only the latter two are used for final compliance tests, although the peak detector is often used in practice for speed of response. A continuous, unmodulated signal will give the same indicated value on all three. However, many interference sources are pulsed or modulated. CISPR has found it necessary to weight the measurement of such interference according to how severe its disturbing potential is, and this has resulted in the specification of the quasi-peak and average detectors. The time constants of the quasi-peak detector are given and the characteristic response that these produce. From this you can see that impulsive noise will give a lower indicated output as its repetition rate is reduced. Maximum output is reached as the impulse repetition rate approaches the measuring bandwidth. The average detector simply gives the average value of the pulse which, for a very narrow pulse, is directly proportional to repetition rate, whereas the peak detector gives an output which is independent of the pulse or modulation characteristics of the signal - provided that the signal appears for long enough in the measuring bandwidth for its peak value to be captured. CISPR emissions standards are structured so that the limit with the average detector is 10-13dB below that with the quasi peak. This means that the average limit is more significant for continuous or high repetition rate pulsed interference, while the quasi peak limit effectively applies to low repetition rate pulsed interference. This applies only to conducted emissions below 30MHz - there is no average limit for radiated tests in either CISPR 11 or 22. ----- CISPR QP time constants --- Relative output of CISPR detectors The justification for weighting of pulsed interference is that such noise, when applied to broadcast reception, is subjectively more disturbing the higher its repetition rate, and therefore low repetition rate noise can be treated more leniently. The assumption of a particular weighting curve does not hold for digital broadcast or data transmission, and a new weighting function needs to be developed to cover this. The spectrum analyzer A frequently-used alternative to the measuring receiver is the spectrum analyzer. Several suppliers produce an EMC-specific version of their standard spectrum analyzers, that is the instrument includes the required CISPR bandwidths and the quasi peak detector (average detection can be implemented, with care, in any spectrum analyzer simply by reducing the video bandwidth). Measuring receivers normally use on-board or external software control to make best use of their facilities. A spectrum analyzer can be used stand-alone, or an external software package can control it. The major advantage of the analyzer is that it can show a large part of the spectrum in a single sweep, in near real time, whereas a conventional receiver requires software control and a sweep of several seconds to achieve this (top- of-the-range receivers may combine both functions). This can be especially useful in diagnostic and pre-compliance work although its value in full compliance testing is less marked. A spectrum analyzer has a number of disadvantages compared to a measuring receiver which have a bearing on in-situ testing. These revolve around the extreme sensitivity of its input, and its lack of discrimination of high-level wideband input signals. Input destruction In order to achieve the capability of near-instantaneous sweep across GHz, the analyzer has no tuned filtering at its input. If no attenuation is switched in, the input connector is directly connected to an extremely sensitive semiconductor mixer diode. This device can be destroyed very easily by quite low-level transient impulses, certainly within the scope of what might appear on a direct supply connection. The analyzer front-end is therefore not robust, and should never be connected directly to the output of a conducted emissions transducer such as a LISN or voltage probe. A transient limiter is an essential accessory which must be interposed between the two. This is particularly important when making in-situ conducted measurements, since direct connection to a mains supply via a voltage probe is often needed, and no isolation from mains-borne transients is provided by this connection. The measuring receiver is more robust in this sense, since it has front-end selectivity which reduces the energy of transients, as well as protection devices of its own. Considering the cost of receivers though, many test engineers will use an external limiter even with a receiver. Input overload Allied to the issue of lack of selectivity, the spectrum analyzer can also be overloaded by wideband pulsed noise, or by other high-level signals on its input. This results in an artificially low indicated value, and can also generate spurious signals through intermodulation in the analyzer’s signal processing chain. If the test operator does not realize that the analyzer is being overloaded, the measurements may be logged as if they were real. Since the high-level signals that are responsible for the overload can appear outside the frequency range under observation - pulsed noise below 150kHz is a common phenomenon- their presence may often remain unknown. The use of tuned input circuits to provide selectivity in the measuring receiver overcomes this problem, and to some extent this can also be applied to the analyzer, through a device known as a "pre-selector", which provides a degree of external selectivity. The problem is more acute for conducted emissions rather than radiated emissions testing, since wideband pulsed noise is more common on mains ports of all kinds of apparatus, and so measuring receivers are normally to be preferred for these tests. It is probably also true that with a good quality spectrum analyzer an unusually high level of pulsed noise would be needed to cause overload, and most AMN/LISNs (see next section) include high-pass filters which remove the worst offending part of the spectrum, below 9kHz. The AMN/LISN Conducted emissions measurements are most commonly made on the mains supply port of the equipment under test (EUT), principally because the mains supply offers the most direct route for conducted disturbances to reach other mains-powered apparatus, and because mains supply wiring is inherently a relatively efficient radiating antenna at the frequencies of the conducted tests. The third edition of EN 55022 (CISPR 22), published at the beginning of 1998, includes a requirement for a conducted emissions test on telecommunication ports but these have yet to be widely applied. A set of widely harmonized mains port limits have been published; the Class A and B levels (the thicker lines) are common to most CISPR-based standards. --- CISPR limits for conducted mains emissions These limits assume a defined RF impedance presented to the EUT's mains connection by the test set-up. In general, the mains impedance can vary over a wide range depending on the circumstances of the individual test location. Since we are measuring a voltage across this impedance, and since the EUT has a source impedance which is unknown but may be quite high, in order to make a test which is repeatable between different laboratories the mains impedance must be stabilized. This is the function of the Artificial Mains Network (AMN, to give it its CISPR term) or Line Impedance Stabilizing Network (LISN, which was originally a US term applying to several different types of network). The requirements of the AMN/LISN are defined in CISPR 16-1. Several variants are presented but one in particular has become the norm for standard tests: this is known as the 50 -ohm/501.tH + 5 -ohm AMN/LISN. The principal requirement is that it should present an impedance between each line and the ground/earth reference point which is equivalent to 50 -ohm in parallel with 501.tH, for the frequency range from 150kHz to 30MHz. For tests below 150kHz, a 5.0. resistance appears effectively in series with the 50t.tH inductance. --- the basic circuit of such a network -- the impedance-frequency response given in CISPR 16-1 to which it is calibrated. An identical network referenced to ground/earth is needed for each line, hence two networks are needed for a single phase supply, three or four for a three-phase supply. The high-pass filter shown in the circuit is not mandatory, but most commercial AMN/LISNs implement it; it is helpful in reducing low-frequency (50Hz and harmonics) feedthrough to the measuring receiver and consequent overload. The AMN/LISN must also couple the signal to be measured, with a known and preferably low insertion loss, to the input of the measuring receiver; and it should attenuate signals that are coupled into the test set-up from the mains supply. These functions are carried out by the other components in the network. --- Circuit of the CISPR AMN/LISN --- AMN/LISN impedance versus frequency CISPR 16-1 also defines a 50 -ohm/5gH + 1 -ohm network, whose impedance curve is also shown. This is said to be suitable for currents up to 500A and so might be relevant for in-situ testing. However it is not specifically referenced in any of the usual CISPR standards, it is not widely available, and consequently it is rarely if ever used in commercial tests. The most important aspect of using the AMN/LISN is that its ground/earth reference point must be solidly bonded to the ground plane which forms part of the test set-up. This is covered in greater detail later. You will also notice that, according to the standard circuit, each network presents about 12gF capacitance between the phase and ground/earth. In the case of the live line, the full 230V AC at 50Hz appears across this capacitance, and this causes a current of nearly 0.9A to flow in the ground/earth conductor. This current can easily be fatal if it is not properly returned to mains safety ground/earth, and therefore secure and reliable safety ground/earth bonding is vital when using an AMN/LISN. A regular testing regime for the security of the ground/earth bond is recommended (with records kept), the equipment should be marked by the manufacturer with the mandatory safety warning about high ground/earth leakage currents, and the ground/earth should never be disconnected without first isolating the AMN/LISN from all poles of its mains supply, including neutral. Use of an AMN/LISN should be restricted to appropriately safety-trained and competent staff, and all others should be prevented from touching or using it. A secondary result of this ground/earth current is that the mains supply to the AMN/LISN cannot be protected by an ground/earth leakage or residual current circuit breaker (RCD). If an RCD-protected supply has to be used, or personnel safety cannot be ensured by adequate bonding, then the mains supply should be passed through an isolation transformer. This will not affect the RF properties of the test, though it will of course limit the current that can be supplied, which may be a problem for power convertors (such as the DC power supplies found in most modem apparatus) as they may saturate the isolating transformer with the very large peak currents they draw each half-cycle. This can distort the mains waveform and may lead to erroneous EMC test results. Consequently, where an isolating transformer is to be used to supply an EMC test, it may need to be rated for a very much higher VA than would at first seem to be necessary. (Power-factor-corrected supplies that meet EN 61000-3-2 draw substantially sinusoidal supply current and so will not cause this problem). The voltage probe Of particular interest for in-situ tests is the alternative option provided in CISPR 16-1 to use a voltage probe. The AMN/LISN discussed above is designed to pass the full mains supply current; it must be inserted in series with the mains supply. This can be inconvenient and sometimes impossible, if: ++ the supply current drawn by the EUT is greater than the rating of the AMN/LISN, or: ++ the mains supply cannot easily be interrupted, either electrically or physically, to insert an AMN/LISN for the test. In either of these cases it is necessary to use the voltage probe. This is a very straightforward and simple device, consisting of a series resistor, a 50Hz AC blocking capacitor and an inductor to provide a low impedance to 50Hz at the receiver connection. The resistor gives a typically 30dB insertion loss which must be corrected for to arrive at the proper measured value. The probe is connected between the measurement point (phase or neutral connection) and a local reference ground/earth point, discussed further. The voltage probe is inserted across the mains connection rather than in series with it and therefore is unaffected by mains current. Its disadvantage is that it cannot stabilize the RF impedance; its own impedance is high enough not to affect the measured circuit, but it does not substitute for the actual (unknown) RF impedance of the supply at the point of measurement. Therefore, strictly speaking, its only use should be for in-situ measurements where the measurement results are relevant only to that particular installation, and should not be taken to represent the same EUT used at different locations. --- The voltage probe --- CISPR radiated emission limits Antennas Radiated field tests require a measuring antenna to convert the incident field strength to a voltage which can be measured by the receiver or spectrum analyzer. The limits for radiated emissions are quoted in terms of dB~tV/m (electric field strength) for the common frequency range of 30-1000MHz; those few standards which require magnetic field tests below 30MHz quote their limits in dB~A/m. EMC antennas must be as broadband as possible in order to cover a wide frequency range in a single sweep. The antennas which are now commercially available and widely used fall into three types: ++ loops, for the range 9kHz-30MHz (Bands A and B) ++ biconicals, log periodics or the hybrid BiLog, for the range 30MHz-IGHz (Bands C and D) ++ horns, for above 1GHz (some log periodics or BiLogs can be stretched to 2GHz) The antenna factor is the most important property for EMC measurements because it determines the voltage at the antenna terminals for a given incident field. Each calibrated broadband antenna will be supplied with a table of its antenna factor versus frequency. This will apply for an incident field in the same plane of polarization as the antenna, and with the antenna terminals loaded with its specified impedance (usually 50 -ohm). --- Antenna types for radiated tests Note that in virtually all circumstances a cable is used to connect the antenna to the measuring instrument. The attenuation of this cable (which is also frequency- dependent) must be allowed for in calculating the field strength from the actual measured voltage. Modem software-controlled spectrum analyzers and receivers allow for the antenna factor plus cable loss to correct the measured reading before it is displayed, so that the display can be calibrated directly in terms of field strength. The antenna factor AF and cable attenuation A are added to the indicated voltage V to give the true field strength value: E (dBpV/m) = V (dBpV) + AF (dB/m) + A (dB) --- Using the antenna factor The loop The loop antenna is a coil which is sensitive to magnetic fields and is shielded against electric fields. Measurements below 30MHz are in the near field for most of the frequency range, and magnetic field measurements give greater repeatability than electric field, since the magnetic field is largely undisturbed by nearby objects (including the antenna). The magnetic field component perpendicular to the plane of the loop induces a voltage across the coil which is proportional to frequency according to Faraday's law. The voltage is also proportional to the area of the loop - the larger it is, the more sensitive is the antenna. Few standards require a loop measurement for apparatus. It can though be very useful for in-situ measurements to determine where particular sources of interference are located and their area of influence. The loop is quite directional (it has been used for direction-finding since the early days of radio); a sharp null in response occurs in a direction orthogonal to the axis of the loop. By contrast to the loop, both CISPR 11 and CISPR 22 require radiated emissions testing from 30MHz to 1GHz. These tests can use either a combination of biconical and log periodic, or a single BiLog, to cover the whole range. These antennas are a sophisticated variation on the electric dipole, modified to extend their frequency range. The biconical The standard antenna for the 30-300MHz frequency range is the biconical. This is a development of the simple half-wave dipole in which the elements have been formed into a conical shape, which dramatically increases the operating bandwidth. The usual form is a skeleton construction which maintains the antenna's electrical properties while minimizing weight and wind resistance. The directivity and plane of polarization remain the same as the dipole. Different versions of the biconical, with minor variations in construction, cover either 30-200MHz or 30-300MHz. The sensitivity is greatest at the half-wave resonant frequency, around 75MHz, at which the AF is 5-6dB/m. Either side of this the AF increases rapidly to 15-20dB/m, making the measurement less sensitive. Because the antenna is balanced, it must be interfaced to the unbalanced 50 -ohm coax cable via a balun (balance-to-unbalance transformer) mounted at the centre of its elements. Adequacy of the balun is one of the most important characteristics of an accurate measurement antenna. Any imbalance results in considerable sensitivity of the antenna to nearby conducting objects, especially the ground plane and the antenna feed cable. The log periodic The log periodic structure (so called because the element dimensions and spacing are logarithmically periodic with frequency) is commonly used for the higher frequencies up to and beyond 1GHz. With this structure a small amount of directivity is available, which gives some gain and hence a lower AF than would be the case for a simple dipole. The structure is essentially an array of tuned dipoles which is fed via a transmission line formed by the twin booms, which removes the need for a separate balun. The directivity means that the antenna factor for signals impinging from off axis, such as from ground plane reflections, is slightly different from that for direct signals (CISPR 16-1 allows no more than 1dB difference). Also, the operational part of the antenna (known as the "phase centre") shifts with frequency as different elements become active, so that the effective distance to the EUT also changes with frequency. For the sake of uniformity, distance measurements are always taken to a reference point in the middle of the boom. The effect of a shifting phase centre is most marked at close separation distances, e.g. 3m. Most commercially-used log periodics cover the frequency range 200-1000MHz or 300-1000MHz. The type can be designed to cover a very wide range, e.g. 80- 1000MHz or 200MHz-5GHz, but it becomes unwieldy if extended to lower frequencies. The BiLog The BiLog was designed in 1993. It is essentially a combination of a biconical and log periodic with the element dimensions being substantially the same, the biconical elements having been collapsed into a flat "bow- tie" shape. This approach means that the elements must be driven via a balun at low frequencies and via the boom transmission line at high frequencies. The BiLog's benefit is that a single antenna covers the whole frequency range from 30 to 1000MHz. This is extremely valuable to test houses as the full compliance measurement can be performed in one sweep, saving time spent changing antennas, with the associated introduction of unreliability due to changed set-ups, damaged connectors and so on. It is larger and more cumbersome than either biconical and log periodic, for which reason it is less popular for in-situ measurements. However a more recent development is the X-wing BiLog, whose biconical section is extended to give an improved low frequency response. This has allowed a more compact version to be produced which is as portable as the other types. --- Typical antenna factors versus frequency (Schaffner-Chase EMC) Conducted test methods To understand the issues surrounding conducted emissions tests, it helps to be familiar with the equivalent circuit. This section a generalized circuit which is applicable to any type of equipment under test, whether it is a component, apparatus or system. There are four significant parts to this circuit: the AMN/LISN, the EUT, the mains cable between the two, and the ground/earth reference. In the case of a test in a test laboratory, the ground/earth reference is a conductive plane whose minimum dimension must be substantially larger than the EUT. For in-situ testing, the ground/earth reference arrangement may be more flexible. --- The equivalent circuit for conducted tests The EUT can be represented as having three distinct equivalent noise sources. These are: ++ the differential mode source VDM across live and neutral: caused for example by harmonics of the power supply switching current, due to rectifiers or semiconductor switches; ++ the common mode source VCM 1 between the structure of the equipment and the mains connection; this will normally be capacitively coupled through CCM, since the mains is isolated from the chassis, and represents the noise sources within the equipment that can be contained by shielding and mains filtering; ++ the common mode source VCM 2 which represents noise sources that are capacitively coupled to the environment by leakage through the shielding or through lack of shielding, shown as CS2. The EUT may have a direct ground/earth connection, in which case VCM 1 and VCM 2 are isolated from each other and only VCM 1 is measured. If it does not, or if the ground/earth connection has a high impedance, then the metalwork is coupled to the ground/earth reference via CS1 and the green-and-yellow safety ground/earth in the mains cable (assuming it is safety Class I apparatus). In this case, both common mode sources are relevant; the amplitude of the voltage due to VCM 2 is affected in a complex manner through C S 1, CS2 and the mutual inductances within the mains cable. The mains cable itself can also affect the measurement of common mode voltages. If it is run close to the ground/earth reference then C_cable can be large enough to affect the load impedance presented by the AMN/LISN at higher frequencies, or it may introduce its own resonances when combined with the cable's inductance or other inductances in the EUT. The differential mode voltage VDM is largely unaffected by the stray capacitances or by variations in the mains cable. It can be measured quite successfully without serious uncertainties being introduced by layout issues. Unfortunately, there is no way of isolating it from the other sources with a conventional AMN/LISN, and experience suggests that there are very few types of EUT in which it is the dominant source across the whole frequency range. Important issues in the conducted test Armed with this understanding of the equivalent circuit, we can now explore some of the aspects which are important to achieve a successful and repeatable test. The features which most affect the coupling of the interference are the ground/earthing of the equipment under test, the stray capacitance between the EUT and the ground/earth reference, and the layout of the mains cable. These aspects are all closely defined in the test set-up for laboratory tests, particularly in CISPR 22, but there is much less definition for an in- situ test. It is also vital that the AMN/LISN should make a good low-inductance connection to the ground/earth reference. Clearly the ground/earth reference is a crucial factor in all of these. In test laboratories it is defined by the ground plane as part of the test set-up, but in-situ ground/earthing has to reflect the existing situation. CISPR 16-2 recommends as follows: The existing ground at the place of installation should be used as reference ground. This should be selected by taking high frequency (RF) criteria into consideration. Generally, this is accomplished by connecting the EUT via wide straps, with a length-to-width ratio not exceeding 3, to structural conductive parts of buildings that are connected to ground/earth ground. These include metallic water pipes, central heating pipes, lightning wires to ground/earth ground, concrete reinforcing steel and steel beams. In general, the safety and neutral conductors of the power installation are not suitable as reference ground as these may carry extraneous disturbance voltages and can have undefined RF impedances. If no suitable reference ground is available in the surroundings of the test object or at the place of measurement, sufficiently large conductive structures such as metal foils, metal sheets or wire meshes set up in the proximity can be used as reference ground for measurement. There is no hint as to what size of structure is "sufficiently large", and this is usually the difficulty when it comes to implementing a test ground plane on site. Consideration of the equivalent circuit suggests that it needs to be large enough to stabilize C S 1, Cs2 and C_cable, in other words that it should underlie those parts of the EUT which are relevant for these stray capacitances. Where these are not obvious the largest practicable structure is used by default. However, there is also a prohibition in some documents against using a test ground plane at all, if it is not a permanent part of the installation. A proper choice of ground/earth reference is necessary both for the AMN/LISN test and the voltage probe test; in either case the reference connection of the transducer must be bonded via a short, wide strap to the chosen reference point. Lengths of green-and- yellow wire are not adequate! Choice of the point of measurement will be dictated by the circumstances of the test, but the normal choice will be at the boundary of the system where the power supply connects to it; typically this would be the terminals of a power outlet or a supply transformer that is dedicated for the system. If the route of the mains Cable leading to the point of measurement is well defined, e.g. run in a fixed conduit, there is no problem. Sometimes this is not the case and the mains cable is loosely bundled or coiled on the floor when the test is undertaken. As should be clear, this is not the optimum situation and some steps should be taken to fix and record the layout of the cable, at least for the duration of the test, in what is hopefully going to be representative of its operational state. Because the stray capacitance CS2 is often dependent on other (signal) cables, the layout of these should be fixed and recorded similarly. Test procedures Practically none of the test standards provide adequate guidance for in-situ testing, so site-specific test plans have to be developed. Many decisions have to be taken by test engineers on the spot, taking into account the local situation and specific aspects of the EUT. Nevertheless, some basic practices which apply to conducted tests in the laboratory will also apply here: ++ wherever possible, an ambient scan is taken at the chosen test point with the EUT switched off or quiescent. Situations where it is impossible to eliminate the EUT from the measurement (usually because it cannot be switched off) are severely disadvantaged; ++ a peak detector sweep is made with a reasonably fast scan speed, taking into account the EUT cycle time, to identify parts of the spectrum in which EUT emissions are most significant, and a list of these frequencies is created; ++ these frequencies are re-tested individually with quasi peak and average detectors to make the comparison with the limits, modifying the EUT's operation to maximize the emissions if this is relevant; ++ this procedure is repeated for all phases at each location to be measured. Radiated test methods Conventional radiated emissions compliance tests as described in CISPR 11 and 22 for performance at a test laboratory have the following features: ++ the test is performed on an Open Area Test Site (OATS) with a defined site attenuation characteristic ++ the measurement antenna is placed at a fixed distance from the EUT ++ the test site includes a ground plane between the EUT and the antenna ++ the antenna must be scanned in height to find the height at which the measured emissions are maximum ++ the EUT must be rotated in azimuth to find the direction of maximum emission ++ the EUT set-up (particularly cable layout) and operating mode must be varied to find the configuration that gives maximum emission ++ all test must be performed for both horizontal and vertical polarization of the antenna. Each of these features is now discussed in turn, relating to the methods for in-situ tests. A further standard with some relevance, "Radiated emission testing of physically large telecommunication systems". Although this does not refer directly to in-situ testing, some of its requirements may be applied. A draft amendment relating to user installation testing was circulated in late 1998, and some of the comments in this section have regard to that document. Another standard explicitly for in-situ testing is under consideration, but is not sufficiently mature to be reviewed here. Test site and measurement distance Site attenuation is the overall loss between two antennas on a measuring site, spaced at the measurement distance. It can be obtained theoretically, and theoretical figures versus frequency are given in CISPR 16-1 Annex G for the two polarizations, three common measurement distances and two different types of antenna. Normalized site attenuation (NSA) is the figure obtained when antenna factors are corrected for. --- the theoretical curves for 10m and 3m distances. The criteria for acceptability of a measurement site is that the measured NSA should not depart from the theoretical value by more than +4dB across the useable frequency range. ---Theoretical normalized site attenuation Imperfections which cause these departures are related to irregularities in the ground plane, inadequate size of the ground plane, and reflecting or absorbing objects within close proximity of the site. While a laboratory test site can control these factors, an in-situ test cannot; it is very rare to find a site where the absence of reflecting or absorbing objects is ensured. So, NSA is not a measure which has any real relevance to in-situ tests. Measurement distance Radiated emissions limits are set for a specific measurement distance, normally 3m, 10m or 30m, between the reference point on the antenna and the boundary of the EUT. 30m tests are rarely used in practice because of the size of the site needed and because of ambient noise considerations. Limits for 30m could be corrected to 10m by increasing them by a factor of 9.5dB, assuming a linear 1/d relationship for the radiated signal. Similarly, 10m limits could be corrected to 3m in the same way. This practice is allowed by CISPR 22 but not by CISPR 11, which allows closer measurement distances but does not permit correction of the limit values - although a recent amendment to CISPR 11 (A1:1999 to the third edition) has changed this. As may be imagined, this restriction in CISPR 11 is the source of some controversy. It has its roots in the observation that the linear 1/d relationship does not hold in many circumstances, the closer in to the EUT the measurement is made. This is due to near- field effects (at 30MHz the near field/far field transition distance is 1.6m, and near field coupling may extend out some distance beyond this) or, with large EUTs, the problem that a close-in antenna is "seeing" only a portion of the whole EUT. But most in-situ measurements are severely restricted in the measurement distance they can employ, because of local features. Although CISPR 11 envisages measurements at the boundary of a premises at any distance between 30m and 100m for Group 2 Class A equipment (CISPR 22 allows 10m), in most circumstances in-situ measurements have to be made inside a building, at much closer ranges to the EUT. A practical way around this problem would be to make cross-checking measurements of some particular emissions at different distances, to see whether the 1/d relationship holds in the situation of interest. If it does, then the rest of the measurements might have corrected limits applied. Though not strictly in accordance with CISPR 11, and of questionable technical merit since the same distance law does not necessarily hold for all emissions, it may be acceptable as part of a self-certification exercise if the manufacturer's EMC expert agrees, and it may also be acceptable to a competent body assessing a TCF. It is always recommended to discuss such issues with the appropriate experts or competent bodies before carrying out the testing. Ground plane and height scan The OATS features a ground plane as an integral part of its layout. The EUT emits radiation in all directions: some reaches the measuring antenna directly and some is reflected from the ground. The purpose of the ground plane is to stabilize the reflected wave. If the ground between the two were left untreated, variations in reflectivity between test sites and on a daily basis at the same test site, due e.g. to rainfall, would cause variability in the measurements of the same EUT under otherwise identical conditions. The alternative, of testing an EUT in free space, has not (until recently t) been considered practical. Placing a metal plane on the ground between the antenna and EUT ensures that the amplitude of the reflected wave is known- assuming no losses in the plane, the reflected wave is slightly less than the direct wave, because of the path difference. Provided that otherwise the open area is truly open, i.e. no other reflecting objects exist, only the direct and ground-reflected waves are received by the antenna and the measurement set-up is stable and repeatable. Note the significant difference in function between the ground plane for radiated tests and the ground/earth reference plane for conducted tests. The ground plane acts only as a reflector for incident waves. Its electrical connection is irrelevant; except for subsidiary requirements of electrical safety, it could float in space and still fulfill its purpose. The ground/earth reference plane is an integral part of the conducted test circuit and must be connected directly to the AMN/LISN reference terminal in order to work. There is now a proposal within CISPR to standardize an alternative method to the OATS using a fully anechoic room (FAR), which approximates to free space. At the time of writing this proposal has not matured into a published standard. --- Reflection from the ground plane The height scan The disadvantage introduced by the ground plane is that the relative phases of the direct and reflected waves vary with frequency. At some frequencies, they will be in antiphase and the received signal will suffer a null. To get around this, the antenna is scanned in height, from 1 to 4m (for 3 and 10m measurement distances) and the maximum indicated value is taken for the measurement. The effect of the height scan is to vary the differential path length over more than L/2 for most frequencies. This will then ensure that nulls are eliminated. For in-situ tests, a ground plane is rarely practical, and because of the existence of other reflecting objects, implementing one would not significantly improve measurement repeatability. It can also be argued that an in-situ test should measure the actual emissions rather than an artificially modified situation. If, then, a ground plane is not used, then neither is a height scan needed for the purpose of removing nulls. However, since the EUT radiation pattern will vary with height, it is still preferable to make a height scan at each measurement position to find the maximum. Any restrictions on the achievable height scan should be noted in the test report. The combined effect of ground plane and height scan is to result in a measured value which is 5-6dB higher than the direct wave alone. It could be said that if only the direct wave is being measured, without a ground plane, then the limits that are applied should be reduced by 5dB to take this into account. This suggestion has not been seriously pursued and is not found in any CISPR documents that directly specify in-situ tests. EUT azimuth When equipment is tested on an OATS, it must be rotated in azimuth (that is, in the horizontal plane) in order to find the maximum direction of emission at each frequency. The same principle applies with in-situ testing, except that the EUT cannot normally be rotated. Therefore the measurement antenna has to be moved around the boundary of the EUT instead. CISPR 11 requires that the number of measurements made in azimuth are "as great as reasonably practical", with at least four measurements in orthogonal directions, and in the direction of any existing radio systems which may be adversely affected. The four orthogonal directions should be viewed as the absolute minimum, with more being taken where the local topography allows it. ETS 300 127, which refers to laboratory testing of large telecom systems where a turntable is impractical, requires eight equally spaced radials; but it also allows the measurement to be performed in one direction only, if the EUT meets the specified limits minus 10dB, and provided the chosen direction is that which is likely to produce the worst case emissions. Justifying this latter condition would be difficult for an in-situ test. In practice, questions of accessibility will determine what directions are actually possible. It is not necessary to maintain exactly the same measurement distance in each direction. EUT configuration On a test site, the EUT has to be configured in a representative build state and operated in its typical operating mode. The configuration should then be varied, keeping it representative, in order to find that which gives the maximum emissions. For in-situ measurements, this requirement is relaxed, because the measurement is of an actual operating system that is installed in a particular configuration in its place of use. Any variation of configuration that may be necessary therefore depends on how flexible the equipment is. This has to be a matter for the experience of the test engineer and the requirements of the client. However, it is usually necessary to explore the effect of operating mode ( e.g., the operating profile of motors supplied by variable speed drives) in order to be sure that the most disturbing state has been found. Antenna polarization In contrast to all the variables discussed above, the simple instruction to repeat the measurement for both antenna polarizations is quite straightforward. Usually this will be done in the initial stages at each chosen measurement point, and if there is a marked difference between the results then the polarization with the higher result at each frequency is investigated fully. Practical aspects of in-situ emissions tests The basic feature of an emissions test is that the EUT is operated in its normal operating mode and its interference emissions are measured across a frequency spectrum, and using measuring instrumentation and methods, defined by the standard that is being applied. The differences that must be addressed when an in-situ test is being performed Practical aspects of in-situ emissions tests are: -- can the standard measurement method be applied successfully and correctly; if not, what are the implications of necessary departures from the standard method? -- can the emissions that are due to the EUT be successfully distinguished from those that are caused by the background environment? These issues can be dealt with relatively easily when a single product is tested at a test laboratory, but they assume much greater significance when a system or installation is being tested in the field Departures from the standard method The previous few pages have cataloged the major areas in which emissions tests in situ might differ from emissions tests on a test site. In many cases these differences are authorized by the standards themselves. If they are not, it is up to the test engineer to justify the course of action taken in a particular case, and record it in the test report. The greater the departure from the standard method, the more detailed and thorough should be its justification, and the more understanding of basic principles of EMC coupling and testing is demanded of the person doing the justification. For example, a common issue is the choice of position of the measurement antenna around the EUT for radiated tests. The determining factor is usually accessibility, as has been noted above. But the test engineer must appreciate the impact of variations in the standard measuring distance, proximity to metallic objects both of the antenna and the EUT, the effect of the flooring, and the issues surrounding antenna height and polarization, in order to be able to assess the impact of a particular position on the test result. All these factors, along with their influence on the final decision, must be noted in the report. Similar issues apply to the choice of transducer and measurement point for conducted emissions. The problem of ambients Ambient signals are those from other transmitters or unintentional emitters such as industrial machinery, which mask the signals emitted by the EUT. On any open site they cannot be avoided, and for in-situ measurements they can be especially problematic. Continuous narrowband signals can be identified and tagged as such in the test report; transient signals may require multiple measurement sweeps to eliminate them. Continuous broadband noise above the limit is the most difficult to deal with, since you can neither work around it nor wait for a break in it. --- Ambient signals in a radiated measurement Emissions standards recognize the problem of ambient signals and in general require, rather optimistically, that the test site ambients should be below the limits by at least 6dB. When they exceed this, CISPR 22 recommends testing at a closer distance such that the limit level can be increased -- as discussed earlier, CISPR 11 does not allow this. This is usually only practical in areas of low signal strength where the ambients are only a few dB above the limits. CISPR 22 also suggests interpolating between adjacent readings, but this is a dangerous practice if the emissions are narrowband. CISPR 11 includes a formula in its annex C which assumes that the ambient signal can be subtracted from the EUT emission, but this is clearly stated only to be valid where the unwanted signal is from a stable AM or FM transmitter with a total amplitude up to twice the EUT emission, and if the EUT emission is itself stable in amplitude and frequency- a rare combination. A proposed amendment to CISPR 16-2 attempts to address the problem of ambients from another angle. This distinguishes between broadband and narrowband EUT emissions in the presence of broadband or narrowband ambient noise. If both the ambient noise and the EUT emissions are narrowband, a suitably narrow measurement bandwidth is recommended, with use of the peak detector. The measurement bandwidth should not be so low as to suppress the modulation spectra of the EUT emission. If the EUT noise is broadband, the measurement cannot be made directly underneath a narrowband ambient but can be taken either side, and the expected actual level interpolated. When the ambient disturbance is broadband, bandwidth discrimination is not possible, but a narrowband EUT emission may be extracted by using the average detector with a narrower measuring bandwidth that maximizes the EUT disturbance-to- ambient ratio. The average detector should reduce the broadband level without affecting the desired EUT narrowband signal, as long as the EUT signal is not severely amplitude or pulse modulated; if it is, some error will result. Broadband EUT disturbances in the presence of broadband ambients cannot be directly measured, although if their levels are similar (say, within 10dB) it is possible to estimate the EUT emission through superposition, using the peak detector. ---Ambient discrimination through use of detectors and bandwidths The in-situ test engineer's kit The practical demands of site testing- and particularly the expensive consequences of inadequate preparation- mean that an engineer who specializes in this activity soon develops a "travel kit" which goes on all site visits. Hofmann has described such a kit and the following list is derived in large part from his paper. Apart from the basic measurement instruments such as spectrum analyzer or measuring receiver and laptop computer control system, the following "accessories" can be considered vital:
Next: In situ EMC testing--Immunity Prev: |
Home | Glossary | Books | Links/Resources |
| EMC Testing | Environmental
Testing |
Vibration Testing |
Updated: Monday, 2012-11-05 0:24 PST