Immunity Testing: Practical Aspects of Basic Standards

EMC testing of immunity against electrostatic discharges as defined in the "old" standard, IEC 801-2, edition 1:1984, concerns air discharges occurring when the charged tip of an ESD generator is brought close to a test object. The rise time from the generator is nominally 5 ns, but due to influences on the air gap during the flashover, the actual rise time can vary from 5 ns up to 30 ns. It may also happen that a high voltage level like 8 kV might create a series of partial discharges, all with much lower amplitude. These kinds of variations can elicit quite nonreproducible results. This is why the discharge principle was modified by IEC 801-2, edition 2:1991 to include an air gap inside the generator, making contact discharges possible. To ensure that the discharges will be insensitive to atmospheric influences, a "discharge switch" is mounted in a closed space with air of a given composition. The test generator is nowadays commonly called an ESD pistol.

The discharge voltage is referred to a reference ground plane (RGP) on the floor, to which the handheld ESD pistol is grounded by a wire. When testing tabletop equipment, this should be placed on a wooden or plastic table covered with a metallic coupling plane. This in turn is to be connected to the RGP by a discharge return cable, with a resistance of about 1 M. A thin layer of insulation is used to isolate the EUT from the coupling plane. Due to the fast rise time, it is very important that the test setup is done very carefully to obtain reproducible test results; for example, the pistol shall be held perpendicular to the tested surface with the return cable kept at least 0.2 m from the EUT. Note that two discharge resistors (2 x 470k) should be used for the discharge return cable, one at each end of the cable. The cables, in between each resistor and its contact point at the reference plane and the coupling plane respectively, should be as short as possible.

According to the new standard IEC 61000-4-2 (IEC 801-2:1991), testing shall start with contact discharges toward reachable metal surfaces, then as air discharges toward insulating surfaces. Note that when testing an insulated test object (e.g., plastic chassis) and objects without safety earth connection (Class 2) this has to be manually discharged between successive tests. Besides contact discharges to metal surfaces on the EUT itself, the revised standard also describes how to generate indirect pulsed fields by having the ESD pistol discharged against a vertical coupling plane placed close to the EUT (see Figure 1).

The revised standard defines a very short rise time, about 1 ns, when discharging the pistol to a target with low resistance (noninsulated parts of the EUT). An oscilloscope is used for verification of the pulse waveform; the rise time should be within 0.7–1 ns and peak current as defined in the standard within +10%. Note that for measuring the 0.7-ns test pulse, the oscilloscope used must have at least 1 GHz real-time bandwidth. To realize how short this rise time is, compare it to the speed of light propagating one foot during one ns.

The test level is often 4–8 kV contact or 8 kV air discharge, which is fairly realistic for most practical situations (a contact discharge of 6 kV corresponds to about an 8-kV air discharge). Tests should be carried out with the amplitude increased in binary steps.

The standard prescribes that the pistol has to be fired at least 10 times toward each test point. Experience shows, however, that interference in a digital system may show up on the 50th firing when the ESD pulse happens to coincide with a specific clock pulse! So when trying to find weak points during the development stage, it is recommended that one perform a lot of discharges at each point.

The basic instruments required for this test are an RF signal source, a power meter, broadband power amplifier(s), broadband antenna(s), and measuring probe(s). The most commonly used antenna types are biconical antennas or strip lines for frequencies up to about 200 MHz, log-periodic antennas for higher frequencies, or combined antennas of the bilog type to cover 20 MHz–1 GHz. Horn antennas are used at GHz frequencies. Dipoles are mainly used for calibration and control.

A homogenous electromagnetic field can also be generated in a guided-wave TEM cell, which has two flat conducting surfaces to generate a traveling electromagnetic wave in transfer electromagnetic mode. Such a TEM cell can be used for very low frequencies up to about 200 MHz. One type of open TEM strip line is described in CISPR 20/EN 55020, where it is used for immunity testing of audio and broadcast receiving equipment (radio & TV) up to 150 MHz. A GTEM cell can be seen as an alternative to the use of antennas in an anechoic chamber. The GTEM cell is a special kind of enclosed transmission line device. It is a pyramid-shaped line terminated with absorbers and resistive loads in the far end. It can be used up to the GHz region (GHz TEM).

Figure 2. Testing of immunity against radiated fields for large test objects requires a big anechoic or semianechoic chamber (SEMKO EMC Center).

Immunity testing by illuminating the EUT with radio frequency electromagnetic fields usually has to be performed in a shielded enclosure in order not to disturb radio communications in the surroundings. Tests can be performed in a cell, a shielded tent, or in a shielded chamber. Any metallic enclosure, however, gives rise to reflections that would make it impossible to control the field. It is therefore necessary to use some kind of lining with absorbers (see Figure 2). A fully anechoic room is preferred, but a semianechoic room (with reflecting ground plane) may also be used if, during the test, the floor between the antenna and the test object is covered by absorbing material (ferrites, foam absorbers, or a combination). There is one exception, the stirred-mode reverberation chamber. This is a shielded enclosure with moving reflectors, where the RF field is scattered around the room in order to achieve the worst possible situation for the EUT. (Refer to the draft basic standard IEC 61000-4-21.)

The test object has to be illuminated from several sides. In a shielded room, testing is therefore usually performed by placing the EUT on a turntable, manually or remotely controlled. Using a turntable in a GTEM cell might, however, cause some problems. In light of that difficulty, another interesting approach might be to keep the EUT at a fixed position and circulate the electromagnetic field.

The antenna is placed in one or several positions in front of the EUT in such a way that the whole test object can be illuminated by the field, usually of 3, 10, or 30 V/m amplitude.

The power needed to generate a certain field strength, E [V/m], depends on the power at the antenna input, P_a[W], the distance to the EUT, r [m], and the antenna gain (factor) G:

Note that the gain G of a broadband antenna varies with frequency; this explains why the power has to be adapted for each frequency step. Broadband antennas, like the biconical type, usually have a high VSWR at low frequencies and reflect much of the power back to the amplifier, which has to be taken into account.

Testing as defined in the "old" standard, IEC 801-3:1988, can be performed in a nonanechoic shielded room, by closed-loop control from a measurement probe placed 20–40 cm beside the EUT. Two probes may be needed in case of large equipment (mean value or the lowest value is then used for the control). This method, however, is ambiguous. Modern testing according to IEC 61000-4-3:1996 refers to a substitution method, based on calibration in an empty test chamber (without the EUT). Calibration is usually performed by measuring, at each frequency, the feed-forward power needed to generate the defined field strength. The results from the empty-chamber calibration can then be used during testing to control the amplifier output power or the forward power (with correction for the reflected power).

The electromagnetic field is measured with a field strength meter (probe), which has a typical accuracy of about +1 dB. A triaxial dipole probe is preferred, to make it possible to measure fields of any polarization without changing the probe direction. A power meter (typical accuracy about +0.1 dB) is usually connected to the power amplifier output via a probe. The probe includes a measuring head with a direction coupler, making it possible to measure the feed-forward power as well as the reflected power.

The antenna shall be placed at a fixed height above the floor, at a suitable distance from the EUT. A distance of 3 m is recommended, but a shorter distance is often used when illuminating small test objects. Tests shall be performed with the antenna in a vertical and horizontal position respectively. Tabletop equipment is placed 0.8 m above the floor. The platform used for this is usually a wooden or a plastic table without metallic parts (plastic usually shows better transparency than wood). Floor-standing equipment should be placed on a nonconducting support of 0.1 m thickness. If a turntable is used when testing a system, special arrangements may be needed to turn around all the cabling.

The RF frequency is usually varied from 27 or 80 MHz up to 1 GHz or more. The dwell time, the time during which the EUT is continuously illuminated at each frequency, must be long enough for the EUT to complete a given working sequence and for monitoring eventual reaction. If the frequency is swept over the frequency range, the maximum sweep rate should not exceed 0.0015 decades a second. When stepping the frequency, each step should not exceed 1% from the previous frequency. This involves 231 frequency steps [1/log (1+1%)] per decade. To step over the frequency range 80 MHz to 1 GHz with a dwell time of 3 s thus takes about 20 minutes (including amplifier settling time). A complete testing with illumination of vertical and horizontal fields against four sides of the EUT usually takes a couple of hours, depending on the software used.

Modern standards require tests to be performed with a modulated field. This is usually performed by adding a 1-kHz sinusoidal modulation signal (amplitude modulation) or by keying the carrier (pulse modulation), as shown in Figures 3a, 3b, and 3c. Modulation with keyed carrier is defined in the temporary European standard ENV 50 204.

Figure 3a. The unmodulated carrier.

Figure 3b. Illustrates amplitude modulation of the RF carrier by adding a 1-kHz sinusoidal at 80% modulation depth.

Figure 3c. Modulation with a keyed carrier simulating a typical GSM, DECT, TD-CDMA, or other digital mobile communication systems.

The power requirements usually in-crease significantly at the low end of the fre-quency spectrum, due to the fact that the effective antenna gain usually decreases significantly for frequencies under 80 MHz. For generation of homogenous fields in a smaller room, 150 W might be needed up to 80 MHz, but 30 W might be sufficient for higher frequencies. For a large anechoic chamber, 1 kW might be needed for illuminating a large test object, as the antenna has to be placed at some distance from the EUT. Since broadband, high-power amplifiers tend to become rather costly, two amplifiers are often used (each can then be matched to the frequency range and power requirements for the antenna used for that range). Modulating the RF carrier implies that the amplifier must be linear and capable of producing enough peak capacity. Modulation at 80% increases the instantaneous power requirement by a factor of 5.2 dB as to an unmodulated signal.

The same antenna that is used for emissions measurements can be used for immunity testing, usually a biconical and a log-periodic or a combined bilog antenna. Since the power handling capacity of the antenna is limited due to the fact that some power will end up in heat, care must be taken not to blow the balun (a transformer built into the antenna to adapt the balanced antenna to the unbalanced coax cable input).

The test object is supposed to be illuminated by a plane electromagnetic wave with impedance Z₀ = 377 () (Z₀ = E/H). In order to achieve this, the antenna must be placed sufficiently distant from the EUT, in the far field region for example. The lobe from the antenna in the near field region has a complex shape, and the amplitude is very dependent on the distance (the field strength close to the antenna decreases as 1/r³ as to the far field, which decreases as 1/r). The transition in between the near field and the far field region is often represented in EMC literature by the distance /2 because this is where the field factors representing 1/r³ become equal in amplitude. However, in order to be able to approximate the field as a plane wave, a minimum distance of /2 is recommended. The need for a long distance is of course also due to the size of the antenna as well as the test object and its cabling. Modern standards, which usually specify the nominal frequency range for EM field testing as 80–1000 MHz, prescribe that measurements at 3 m take precedence (in case of dispute).

The new standard requires that the field is homogenous within a certain vertical area in which the EUT is to be placed. Since it is impossible to establish a uniform field above a reference ground plane, this has to be covered by absorbing material in between the antenna and the EUT. For this purpose, ferrite tiles mounted on plates on wheels are handy to use in conjunction with a semi-anechoic chamber used for emission testing (when a RGP is needed). In the case of tabletop equipment, the uniform area is required at a height of 0.8–2.3 m above the floor. The required vertical area of the uniform field represents a grid of 16 points within 1.5 x 1.5 m.

Control of the uniformity shall be accomplished as an empty-chamber calibration without the EUT. Calibration shall be repeated in frequency steps no greater than 1% of the start frequency for both horizontal and vertical polarization. From knowledge of the input power and the field strength, the necessary forward power for the required test field strength can be calculated. The recorded "empty-chamber file" can therefore be used for controlling the amplifiers during the test. A probe is, however, normally used during the tests for supervision. In the case of a large EUT, illumination may have to be performed with different antenna positions covering part of the EUT. In the case of a small EUT, the uniform area might be decreased (1 x 1 m may be used).

Of the 16 measured points, at least 12 should represent a uniform field within 0–6 dB of the nominal value (see Figure 4). This is sometimes done by first eliminating four extreme values and keeping 12 reference points to represent a field strength within 0–6 dB. However, other methods are also used. Note that first calculating the mean value and then eliminating extreme points, which differ more than ±3 dB, may not lead to the same remaining reference points. Algorithms for empty-chamber calibration are usually included in commercially available software packages; note, however, that these might be based on different methods for defining the homogeneity.

Figure 4. Field uniformity is measured at 16 points of a vertical grid (an empty-chamber calibration without EUT). At least 75% of these points („12) must be within a 6 dB spread.

Figure 5. Testing of immunity against an induced RF current according to IEC 61000-4-6.

Immunity Testing with Induced RF Disturbances
(EN 61000-4-6/IEC 61000-4-6)

This standard defines RF immunity testing by inducing radio frequency disturbances into mains, cables, and into cables connecting the EUT to auxiliary equipment (AE) (see Figure 5). Testing is usually performed by applying a voltage via a coupling device to one cable at a time while keeping the remaining cables terminated. In case of a screened cable with multiple lines, the RF disturbance may be applied directly on the screen as bulk current injection.

The disturbance signal is usually applied to the cable connected to the input or output side of the EUT through a coupling network or a clamp. In order not to allow the disturbance signal to interfere with the AE, there is also a need for a decoupling device, consisting of filters with series inductors or a separate ferrite tube. Coupling/decoupling impedances will of course affect the induced currents, depending on all other impedances in the system (input/output impedances as well as the balanced impedances to the reference ground plane). An extra ferrite tube (absorber) can be used to enhance the decoupling at the auxiliary end to improve reproducibility.

Injection can be performed by three different coupling methods. Voltage injection via a combined coupling/decoupling network is the recommended test method. Note that isolated cables may have to be deisolated or cut, which, of course, might be a disadvantage. Unscreened power supply lines are injected through a resistor and a capacitor in series (Figure 6a) (the capacitor is needed to prevent the line voltage from affecting the disturbance source). For screened and coaxial cables, the disturbance signal is directly injected via a resistor (Figure 6b). In case CDNs cannot be used, injection by clamp may be used as an alternative. The first choice is then an EM clamp, which is a ferrite clamp providing both E- and H-field coupling to the cable. A ferrite tube on the auxiliary side is recommended as mentioned above. The second choice is a clamp-on current probe, which, like the EM clamp, is easy to use (nonintrusive method) but requires more power for the same stimulus due to higher insertion losses. When using a current clamp, a separate current transformer is usually used for measuring the current.

Figure 6a. Injection by a couple of different coupling/decoupling networks, CDNs. Unscreened power supply lines are injected through a resistor and a capacitor in series.

Figure 6b. For screened and coaxial cables the disturbance signal is directly injected via a resistor.

The different injection methods do not produce identical results. In order to make tests reproducible, it is important to stick to the injection method recommended. Furthermore, it is important to keep track of the impedances. The power applied to the EUT will depend on the common-mode impedance at the EUT port. A low impedance may imply over-testing if a voltage source is used and under-testing if a current source is used (vice versa for a high impedance).

The signal is, as for RF field testing, produced with a signal generator connected to a linear power amplifier. The test generator should have a 50-() internal impedance. Injection by CDNs is usually done through a 100-() impedance (see Figure 6b). The far end of the EUT input/output is usually terminated by 50 () in series with 100 () to create a common-mode impedance of the coupling/decoupling system of 150 . Tests should be performed with the test generator connected to each of the CDNs in turn, while the nonexcited input ports of the coupling devices are terminated by a 50- load. Calibration can be performed by monitoring the voltage appearing across the decoupling network, normally 1, 3, or 10 V, depending on the severity level.

The output power from the amplifier, and the voltage across the 50-() termination connected to the output of the coupling/decoupling network, is measured with a power meter and a couple of probes. Calibration is performed by recording the power needed to give the defined voltage over the whole frequency range. In order to reduce reflections due to mismatched impedances, it is recommended to use a power attenuator (6 dB), placed as close to the coupling network as possible.

The injection technique requires that the common-mode impedance at the far end of the cable from the EUT is defined. Each cable therefore requires a common-mode decoupling network (including an isolating inductor or common-mode choke) at its far end, to ensure the impedance and to isolate any ancillary equipment. The coupling and decoupling devices can be combined into one box, the so-called coupling/decoupling network (CDN). Injection on different cables therefore re-quires a set of different CDNs. The standard specifies different CDNs for specific unscreened cables, for example CDN-M1/-M2/-M3 for unscreened supply (mains) lines, CDN-AF2 for unscreened nonbalanced lines, and CDN-T2/-T4/-T8 for unscreened balanced pair(s). CDN-S1 is used for screened cable (see Figure 6b).

The test object is to be placed 0.1 m above an RGP. All cables must have relevant CDNs placed in contact with the RGP at a given length. Cables connected to the AE, at the far end from the EUT (that is, they are not connected to the EUT) should also be terminated at a short distance from the AE. This can be done with CDNs.

The test system should be connected to the power line by an isolation transformer. This is necessary for keeping the impedance to the reference ground correct. If the EUT has a part that is to be manually controlled by hand, an artificial hand should be used (refer to CISPR 16).

The test method illustrates the situation where long cables pick up electromagnetic energy via antenna effects at lower frequencies. Subunits that are part of the equipment to be tested, which are connected by short cables < 1 m, may be considered part of the EUT (no separate testing on these cables is needed).

Injection is often used as an alternative to radiated field testing in the lower part of the RF frequency range, usually 150 kHz–80 MHz, sometimes up to 230 MHz. The RF signal is varied over the frequency range. Frequency step, modulation, and dwell time are the same as for IEC 61000-4-3/EN 61000-4-3.

The standard defines an idealized burst (15 ms) of short pulses repeated every 300 ms. Pulse rise-time and width are typically 5 and 50 ns respectively, and the pulse repetition frequency a few kHz (5 or 2.5 kHz). Real transients, occurring when breaking inductive loads, might have another shape and cause higher repetition frequencies.

The fact that a modern test generator gives reproducible bursts is of primary concern when testing for legal purposes. There have been discussions on the use of higher repetition frequencies, up to 100 kHz, but since the test method is well established by now, it will not be popular to have to exchange (or rebuild) all burst generators in use. Old generators with a spark gap, however, are not recommended for use since they give less-reproducible test pulses than modern generators.

Control of the pulse shape is easily performed with an oscilloscope (bandwidth >=400 MHz) connected by an attenuator to the output of the generator. The repetition frequency shall lie within 5 kHz ± 20%, and the burst length shall be 15 ms ± 20%.

The test signal is to be applied to power supply ports, protective earth, and signal and control ports. I/O lines shorter than 3 m, however, need not be tested according to the standard. The transients are injected through a coupling/decoupling network or a capacitive clamp, and the length of the signal and power lines between the coupling device and the EUT shall be 1 m or less. The test setup is designed for 50-() loads. For power lines, injection is to be performed in both common and differential mode, with the signal applied using a capacitor (33 nF) from a 50-() source. For I/O cables, the injection is performed using a distributed capacitance along the cable. This is usually achieved through a 1-m-long capacitive clamp.

Tests have to be performed with the EUT placed on a reference ground plane. As the rise time is very fast, it is important that the test generator is connected with a short, wide strap to the reference ground. The test setup, with necessary insulating supports and ground connections as described in the standard, should be followed in detail to get a reproducible result.

Figure 7a. Common-mode (Ua-c and Ub-c) and differential-mode (Ua-b) disturbances, respectively.

The standard surge is defined as a "single pulse with a double exponential waveform." This is usually simulated with a hybrid generator dimensioned to generate a voltage surge having a front time and pulse width of 1.2 and 50 µs, respectively, at open circuit conditions and a current surge of 8/20 µs into a short circuit. The generator has an output impedance of 2 W, which means that a 2-kV open-circuit voltage injection generates a short-circuit current of 1 kA max. By adapting the impedance of the injection circuit, tests can be performed in different modes on different types of lines. The surge is injected through a coupling/decoupling network. When testing ac/dc lines, line-to-line (symmetrical/differential mode) coupling is performed via a capacitor, and in case of line to ground (asymmetrical/common mode), via a coupling capacitor in series with a resistor (see Figure 7a). When testing unshielded symmetrically or unsymmetrically operated lines, e.g., telecommunication lines, arrestors are placed in parallel with the coupling capacitors.

Figure 7b. True common mode in a three-phase system can be seen as injection on all wires with reference to the ground wire or injection on all wires, including the ground wire with respect to the cabinet or reference ground plane.

In some applications, instead of common-mode testing on one wire at a time, "true common mode" testing is used to better mirror surges as they may appear in real life. This involves application of the surge on all wires at the same time. "True common mode" in a three-phase system can be performed as a surge injection via parallel coupling devices on all wires with reference to the ground wire (GND), or injection on all wires including the ground wire with reference to the cabinet or reference ground (RGP) (Figure 7b). The latter may be seen as the most realistic case when simulating a surge caused by a lightning strike.

A regular check is recommended to control the generator voltage and current, e.g., a daily check of 1 kV +10% open voltage output, and current 500 A +10% (with a short circuit plug on the output).

The pulses cannot usually be generated with a repetition rate of more than 1 pulse per minute due to the fact that the power otherwise can destroy components like varistors in the equipment under test. When testing the power supply inlet of the EUT, the pulse is to be synchronized to the line frequency and shall be injected at the following phase angles: 0°, 90°, 180°, and 270°, with 5 positive and 5 negative pulses at each phase angle.

The open circuit test voltage is usually in the range 0.5–4 kV +10 %). The level to be used is usually defined in an applied generic or product family standard. As an example: +2 kV line-to-earth and +1 kV line-to-line, respectively, for equipment intended for use in a residential or light industrial environment. The test procedure shall also consider nonlinear current-voltage characteristics of the EUT. The test voltage therefore has to be increased by steps up to the defined test level.

If not otherwise specified, the interconnection line between the EUT and the coupling/decoupling network should be 2 m or shorter.

This test method describes immersion of the EUT by a continuous LF magnetic field, obtained by a current of power frequency flowing in an induction coil. The test equipment needed for this test includes a current source, the induction coil, and auxiliary test instrumentation. The test setup requires a nonmagnetic ground plane under the EUT of at least 1 x 1 m. The equipment is to be placed on the ground plane with an insulating support.

Depending on the size of the EUT, induction coils of different dimensions may be used. The dimensions recommended are to produce a magnetic field in the whole volume of the EUT within +3 dB. For small EUTs, like terminals and instruments, a square-shaped or circular coil of 1 m side or diameter is recommended. Double coils of the Helmholtz type could be used in order to obtain a homogeneous field better than +3 dB or to extend the volume of testable EUT.

Testing is performed by changing the ac voltage level supplying the EUT during a short period of time.

"Short interruptions and voltage dips," as defined in the basic standard, are to be performed with the following test levels in percentages of the rated voltage: 0% (100% dip = interrupt), 40% (60% dip), 70% (30% dip). The change of these values with the load shall be maintained within 5%.

The dips are applied as abrupt voltage changes occurring at zero crossings of the mains voltage +10%), unless otherwise specified in a relevant product standard (other angles might be considered more critical for certain products). The duration of a dip is usually specified as a number of periods of the mains frequency, ranging from half a period up to 1 second or more.

Voltage variations are gradual changes of the supply voltage to a higher or lower value than the rated mains' voltage. Voltage variation testing is usually optional.

The generator shall have a low output impedance, predominantly resistive (< 0.5 ). During an interrupt, the generator shall be able to act as a low impedance source capable of giving an inrush current with a peak value up to 500 A for 220–240-V mains, or 250 A for 100–120-V mains. For dip testing, the generator must be capable of carrying 40 A at 40% of rated voltage for a duration of up to 5 seconds.

Tests are performed with the test generator connected to the EUT with the shortest power supply cable, as specified by the EUT manufacturer. If no cable length is specified, 1 m should be used.

When referring to one and the same basic standard, generic or product family standards may define specific requirements concerning immunity parameters. For example, voltage interruption according to the generic standard is defined as a 5-second interval, while the product family standard for household equipment defines this interval as 0.5 period (10 ms at 50 Hz).