Common errors have been seen in EMI testing over a great number of years and the guidelines used for successful EMI Testing
GENERAL ERRORS APPLICABLE TO MOST TEST
Inadequate testing methods and flagging all possible failures. Reference 3 9.4.11.2.1 details the correct method of functional testing the Equipment Under Test.
The shielding of power cables must remain unshielded. Aerospace and avionic equipment power input cable shielding are often prohibited. MIL-STD 461F and above prohibits the use of shielded cables unless the platform on which the equipment is to be installed shields the Power Bus from the point of origin to the load. However, MIL-STD-461F specifically states that input (primary) power leads, returns, and wire grounds shall not be shielded.
Not including 2m of power cables and interconnecting cables routed parallel to the front edge of the setup. This requirement is not applicable if the cables in the EUT installation are shorter than 2m in which case the installation length shall be used. See reference 3 for other notes on cables 9.6.4 and general test guidelines.
Forgetting to apply the required modulation to the test level in a susceptibility/immunity test.
Having poor or non-existent grounding of the 10uF capacitor, LISN, and the EUT. This may reduce common mode currents and allow the EUT to pass whereas with proper grounding it may fail.
Many of the tests require a system check prior to testing and sometimes these tests are omitted.
If an EUT fails a susceptibility/immunity test failing to reduce the test level to determine the level at which the EUT passes as this provides an indication of how much improvement is required. It is surprising how often this step is missed but it is very useful in solving the problem.
COMMON ERRORS IN RADIATED EMISSION MEASUREMENTS MADE IN A CHAMBER
1) The ambient inside the shielded room is too high due to RF common mode current flow on signal interconnections or primary power cables. The ambient must be at least 6dB below the applicable limit.
This is typically sourced by functional test equipment or Electrical Ground Support Equipment (EGSE) located outside the room. The source may also be due to a high electromagnetic ambient outside the room coupling to cables entering the room. The ambient may be due to high-power TV or radio transmitters or wireless internet. Solutions include better shielding of the interface cables, the addition of signal filters, and the application of ferrite beads (baluns,) as shown in references 1 and 3. Unshielded cables would typically need a filtered connector, depending on the signal level. For low-level signals, such as analog, a filtered connector as well as a shielded cable may be required.
Although ferrite baluns on cable can very effectively reduce the RF current flow, when the source of the current is an RF field incident on the baluns then it may result in the cable becoming a more effective receiving antenna and the current instead increase.
Reference 3 describes these steps in more detail. Sometimes the power amplifier is located in the room, to reduce the length of and thus the attenuation of the cable between the antenna and the power amplifier. In this configuration, feedback between the antenna and the signal input to the power amplifier can result in maximum output from the power amplifier. Therefore first test the setup with the EUT outside of the room using an extended frequency range. Use the monitoring antenna/field sense device to measure the level of the radiated field. If feedback occurs use a cable with more shielding effectiveness or add a braid over the existing cable with the braid clamped to the connector shell at both ends.
One good solution to coupling through the chamber wall is the use of filtered D-type adapters on bulkhead-mounted D-type connectors. These adapters may also be mounted on the connectors on the test equipment/EGSE, although the connection in the shielded room is almost certainly better.
It is also possible to mount filter components between back-to-back connectors.
If filter connectors are not possible, e.g. cables coming through a stuffing tube, then the shields should be terminated on the wall of the stuffing tube with ferrite cores (baluns) between the external test equipment/EGSE and the termination of the shields to the stuffing tube.
Reference 1 describes the application of the baluns and filter adapters. The stuffing tube is typically designed to be a waveguide below cut-off and will work as such with fiber optic cable or non-metal reinforced rubber or plastic air or water hoses. However, it is a mistake to believe that the waveguide below the cut-off function works with shielded cables or it contains internal conductive wires. Even cables that contain tap water will conduct RF currents into the room due to the conductivity of tap water. To keep RF currents out of the room they must be shunted to the room wall at the point of entry. Another source is a noisy power supply, typically switching, outside of the room which is not sufficiently attenuated by the power line filters mounted on the shielded room wall. Use a linear power supply.
2) Use of incorrect measurement bandwidth at a change in the frequency measured, missing a repeat measurement with a different antenna polarization, exceeding the calibration frequency range of antennas or preamplifiers.
3) Non-representative test setup, use of a different type of cable than the delivered type, non-representative EUT, different software than the delivered, addition of aluminum foil, copper tape, or braid over cables that are not to be shielded.
4) The argument that cable-generated emissions are not the concern of the manufacturer of the EUT when the cable harness is supplied by the customer. Regardless of the type of cable (shielded or unshielded), it is the EUT that injects the noise onto the cable and it is the manufacturer of the EUT who must reduce emissions at the source or persuade the customer to change their cable type. If a manufacturer believes the emission limit will not be met with a customer-supplied unshielded cable, and circuit modification will be a problem, then the customer may allow a deviation or waiver. In equipment designed for use in space, if meeting requirements means that the equipment is removed from the spacecraft and the frequency and level of emission will not result in EMI, then a waiver is often allowed.
5) Software-controlled measurements have a number of potential problems. The majority of these programs do not allow multiple sweeps using peak or maximum hold, but instead take data over a single sweep, a snapshot in effect. Emissions often vary in amplitude significantly from one sweep to the next. With a snapshot of the data, there is no guarantee that the worst-case level has been captured. The argument has been made that this has become an industry standard, but it rarely results in a worst-case measurement, which is the goal. If broadband (BB) and narrowband (NB) emission measurements are required, then emissions are characterized to determine if the BB or NB limit applies. Four different techniques may be used to determine this, and the software normally only applies one which may lead to an incorrect assessment. When confronted, some manufacturers admit to the deficiency in their programs.
6) Room resonances. If a chamber with the minimum absorber as required in MIL-STD-461 or DO-160 is used, then reflections and resonances will still occur. Thus measurements made at one facility may differ strongly from those made at another. Reference 2 shows the variation seen from one facility to another. Explaining the discrepancy to a customer, who is less familiar with this issue, can be a problem as the assumption may be made that the customer’s measurement is the correct one. Ideally, measurements made in a chamber will have a good correlation to those made on an Open Area Test Site (OATS) which has met the OATS requirements (see errors in ATS measurements). Reference 3 describes the design of a room that does correlate well to an OATS, and reference 2 shows the type of absorber that is most likely to achieve a good correlation.
7) Not excluding transmitting antennas. Even when not transmitting the fundamental broadband noise generated by the transmitter may result in a failure in low-level emission limits. Transmitting antennas should be replaced by a shielded load resistor.
COMMON ERRORS MADE IN RADIATED EMISSION MEASUREMENTS ON AN OPEN AREA TEST SITE (OATS)
The correlation between OATS is often extremely poor. The correlation between two test sites with measurements made on the same EUT with the same software, and cable orientation was up to 10.4dB from 212 to 216MHz, references 4 and 5.
In one case, a consistent 26dB of difference was seen between one site and another. Only after the site with the higher level measurement was asked to use a signal generator to check the test equipment was the problem found. The instrument had a switchable 26dB preamplifier, and the inclusion or otherwise of the preamplifier was not indicated on the front panel.
This illustrates the importance of making the test facility perform a sanity check on the antennas, cables, and measuring equipment. A number of times a cable instead of going open circuit had a significantly high attenuation which could be easily missed. In an OATS measurement, the use of the known levels from local FM radio stations is a useful check on the measurement system.
Emission close to and masked by an ambient; the frequency of emissions from a EUT should be characterized in a shielded room before measurements on an OATS. If an ambient seems to be at one of the characterized EUT emissions frequencies, narrowing the frequency sweep may enable the display of the emissions to be separated. Using a narrower measurement bandwidth will reduce the displayed width of the emissions and help in the differentiation. If the ambient is broadband and the EUT narrowband narrowing the measurement bandwidth may reduce the broadband level to below the narrowband level.
An emission that is harmonically related (from the same source) should be measured and the measurement bandwidth reduced to that used in the initial measurement. Although not ideal, if the emission amplitude of the harmonic is reduced then the amplitude of the out-of-specification emission can be corrected accordingly.
The configuration of the EUT should be varied to maximize emissions. Interconnecting cables should be of the type and length which will be delivered. However, it makes no sense to orient cables in a configuration that will never be used. The correct configuration is shown in the test requirement. One test facility incorrectly had cables oriented vertically above the test then horizontally above the EUT and then back to the table which is incorrect.
COMMON ERRORS MADE IN CONDUCTED EMISSION MEASUREMENTS
Many of the errors described for radiated emissions apply to conducted emissions.
Most of the measurements are made with a Line Impedance Stabilization Network (LISN), which supplies a controlled and calibrated impedance. A separate LISN may be used for the power and the power return or a LISN supply both. Measurements are made at a measuring port. When the measuring instrument is connected to one port, then it is important to remember to terminate the unused port with a 50-ohm termination. When the test limit is in dBuA then the measurement is of the noise current measured using a current probe. It is important that the current probe is at the location described in the test requirement. For MIL-STD 461 CE01/CE03 it is at the 10uF capacitor, which is used instead of the LISN. For DO-160 it is 5cm from the backshell of a connector on the EUT. The current probe should not be contacting the metal ground plane on the table but rather be lifted up on an insulator.
Leaving the current probe around a power cable in a conducted emissions test and switching on or off the power supply, as the resultant spike can damage the measuring instrument. For this reason, try to avoid measurements with 0dB input attenuation.
Not grounding the measuring instrument to the ground plane with the use of an isolation transformer. Following this rule will keep ground currents, typically at the harmonics of the AC power, from corrupting the test.
If a preamplifier is used, make sure that it is not in compression due to a high-level emission. This emission may be outside of the measurement bandwidth, e.g. 50Hz, 60Hz, or 400Hz power line frequencies. Add a 6dB attenuator to the input of the preamplifier. If the measured level does not reduce by 6dB then the preamplifier is compressing and may not be needed.
COMMON ERRORS IN MIL-STD-CS101, DO-160 AUDIO FREQUENCY CONDUCTED SUSCEPTIBILITY, DIFFERENTIAL MODE, SECTION 18, OR VOLTAGE SPIKE, SECTION 17
All these tests use an injection transformer inserted in series with the power line to inject the test level. Most EUTs have a switching power supply at the input power line which generates transient currents. The secondary of the injection transformer has an inductance of 1.55mH. If this inductance were seen by the EUT then a transient may set up a resonance due to this inductance and the capacitor at the input of the EUT power input. This can result in very high voltages in the filter and possible damage.
To avoid this the following steps should be followed to maintain a low input impedance at all times.
To power up:
With the EUT unpowered,
1. Remove the signal from the input of the audio power amplifier.
2. Short transformer secondary.
3. Power on the power amplifier and the EUT.
4. Remove the short at the transformer secondary.
5. Connect a signal generator and proceed with testing.
To power down:
1. Disconnect the signal generator at the input of the power amplifier.
2. Short transformer secondary.
3. Power down EUT and power the amplifier.
As the frequency is scanned it will hit the resonant frequency of the EUT input power line filter and very high currents may flow, to the extent that components may heat up and become unsoldered. Although the test method may specify a maximum input power level for the test this may not be sufficiently low enough to protect the EUT. It is known that at least two commercial test facilities avoid dwelling at the resonant frequency. It is important to inform the customer of this possibility and that even though the test destroys the EUT power input the test has been correctly performed.
This warning should be given to the customer for all susceptibility tests.
A consensus on the pass/fail criteria of the test may not have been decided by the customer. For example, a few white spots on a display or some minor distortion may be acceptable, whereas multiple images or major distortion is almost certainly not. Similarly, some bit error rate in data communication may be acceptable whereas total loss in communication is not. The pass-fail criteria should be agreed on prior to testing.
In conducting susceptibility tests on input power, monitor the test level at the wrong location. It must be across the EUT. It must not be across the injection transformer primary, secondary, or at the output of the power amplifier.
COMMON ERRORS MADE IN RADIATED SUSCEPTIBILITY TESTING IN A CHAMBER
1) Chamber resonances can result in either very low or very high electric fields at the EUT. If the field at the EUT is low then the power amplifier may not have sufficient power to generate the specified field. Or if the field is achieved at a location away from the EUT such as down the length of the cable then the field may be very high at this location. In the section on common errors in radiated emission measurements described chambers that have very uniform fields. Although a field uniformity test may be a requirement. Some of the test locations are deleted and these may be at very high levels.
2) A consensus on the pass/fail criteria of the test may not have been decided by the customer. For example, a few white spots on a display or some minor distortion may be acceptable, whereas multiple images or major distortion is almost certainly not. Similarly, some bit error rate in data communication may be acceptable whereas total loss in communication is not. The pass-fail criteria should be agreed on prior to testing.
3) The test frequency range may cover 14kHz to 18GHz or above. Although the test levels and field strength should be tailored to the electromagnetic ambient the EUT will operate in this may not be known or EUT will be operated at several locations. If the EUT contains receiving antennas the EUT will not be operated in locations where very high levels of in-band frequencies exist. For this reason, and to avoid damage to the front end of the receiver the in-band and close-to-in-band test frequencies should not be included in the radiated susceptibility test, or the antenna be replaced with a termination.
4) Personnel should not be allowed to view displays, indicator lights, or meters during a radiated susceptibility test due to the potential danger. Mirrors may be set up so that the EUT can be viewed through the waveguide below the cut-off. Alternatively, a camera that is immune to the test field and is connected via a fiber optic to a display outside of the chamber. If a voltage has to be viewed then a meter may be enclosed in a metal enclosure with a wire mesh face for viewing and short wires to the EUT However connecting the meter to the EUT may result in an incorrect measurement.
5) Coupling between the signal generator input cable and the power amplifier output cable before it enters the chamber or between the transmitting antenna and the input cable to the power amplifier when the amplifier is located inside the chamber may result in positive feedback and the generation of very high E fields. Always monitor the E-field level outside of the test frequency, as well as, within it to detect this feedback.
6) Non-representative test setup, use of a different type of cable than that delivered with the EUT, the addition of aluminum foil, copper tape, or braid over cables which will not be delivered with these.
7) Missing a repeat measurement with a different antenna polarization, exceeding antenna or power amplifier calibration frequency range. All of these problems can be avoided by the use of a checklist of the test, with tick boxes to ensure a test has not been missed.
8) The argument that cable-induced susceptibility is not the concern of the manufacturer of the EUT when the cable harness is supplied by the customer. Regardless of the type of cable, shielded or unshielded, it is the EUT circuit/s that are susceptible to the noise induced on the cable and it is the manufacturer of the EUT who must harden the circuit to make it immune or persuade the customer to change his cable type.
COMMON ERRORS MADE IN THE CS06 TRANSIENT AND OTHER TRANSIENT TESTS
Not applying the relaxation on the CS06 test level for equipment and subsystems which contain Varistors, Transorbs, or similar transient protection devices. The relaxed test level is at the maximum safe level for the device. However, the maximum transients possible on the power line should be analyzed or measured to ensure that these are not above the device’s safe level.
Conducted susceptibility spike test; When testing ac power the spike position shall be moved over a full 180 degrees of the AC waveform on either side of the zero crossing point.
REFERENCES
1. EMI testing; errors, misinterpretations, and cautions. Interference Technology Magazine May 2023
2. Determining Semi-Anechoic Chamber Resonances as a Source of Radiated Emission Variation Between Chambers and Comparing to OATS Measurements. David. A. Weston Interference Technology Magazine November 15, 2017.
3. David A Weston. Electromagnetic Compatibility: Methods, Analysis, Circuits, and Measurements. CRC press 2017.
4. David Weston NARTE News. Volume 18 Fall 2000 Number 3. Lack of Standardized Testing Leads to Widely Varying Measurements at Different Commercial Test Sites.
5. Identification and study of influential parameters in CISPR25 Radiated Emission Measurements and Simulation Combined Analysis. Frederic Lafon, Renaud Dupedent, Josseln Davalon, Candice Chevrian, IEEE Transactions on Electromagnetic Compatibilitty. VOL 58. NO5. OCTOBER 2016.
