The EU Directive on Electromagnetic Fields—Practical Experience of Field Measurements

: The EU directive on exposure to electromagnetic fields was published in 2004, but due to some problems it was not introduced in the EU countries before 2016. However, still today, many companies are not aware of the legislation and have not taken measures to comply. This could perhaps be due to some of the practical problems they are facing in trying to comply. Here, we address some of these problems, such as showing compliance with the action levels for non-sinusoidal extremely low-frequency magnetic fields, time-averaging for the initial start-up current of handheld machines and time-averaging of radiofrequency fields when measuring plastic welding machines, including the uncertainty in the measurements. Finally, we discuss some of the problem concerning workers with special needs.


Introduction
The EU directive on minimum health and safety requirements for the exposure of workers to the risks arising from an electromagnetic field (EMF) was adopted as a minimum requirement in June 2013 and introduced in the EU countries in 2016.Still today, many companies are not aware of the legislation and have not taken measures to comply.This could perhaps be due to some of the practical problems they are facing in trying to comply.
In 2009, Hansson Mild et al. [1] wrote a paper on still-open questions and it reviewed some of the problem experienced by this group of authors with expertise in occupational exposure to the EMF.Some of these questions are still of concern and we will here address some of them.
First, the directive only addresses short-term effects and does not concern possible long-term effects or the possible carcinogenic effects, but it is intended to cover all known direct biophysical effects and other indirect effects caused by EMF exposure.
Risks assessment of EMF at the workplace should be performed according to the framework directive [2].Here it is stated that if relevant action levels are exceeded and relevant exposure limit values may be exceeded, an action plan must be implemented to show compliance.
Special attention shall be paid to workers at risks (pregnant, living with implanted medical devices), including individual risk assessment where applicable.
When measurements are needed to show compliance, there are several aspects that need to be considered, many of which are not clearly stated in the legislation.Here, some of these issues are discussed.

Extremely Low Frequency, ELF
One of the most difficult operations to measure from a compliance point of view is non-sinusoidal field exposure.In the EU directive, the Weighted-Peak-Method in Time-Domain is specified as the reference method for the ELF part.The "Non-binding guide for practical implementation of the directive" [3] gives an alternative method, the Time-Domain-Assessment-Method (TDAM).The problem with assessing these kinds of exposures has been discussed by several authors [3][4][5][6][7].The problem with the different methods is also addressed at length in [8].They point out that the TDAM method underestimates the exposure by up to a factor of 22 in certain situations.We will here look at some of the situations where the problem of assessment of non-sinusoidal signals occurs, and where we think the underlying biological knowledge is not sufficient and needs further studies.
Spot welding in industrial settings uses very high currents, of the order of some to tens of kA, and the operator is standing rather close, with their body and their hands very close to the electrodes.Below are some examples with oscilloscope pictures from some of our recent measurements on welding machines in use in the car industry.
The spot weld lasts from some to tens of Hz periods, up to 50 Hz.Often, operators are not using a zero-crossing relay for start of the weld, which can lead to quite high starting currents and transients.The waveform is often shaped by thyristors to obtain just the right amount of energy to produce the spot weld.As can be seen below in Figures 1 and 2, the same machine can have different settings depending on the material to be welded.This means that the measurements need to be taken for all applications of the machine.The harmonic content of the wave needs to be considered, since the use of the highest current setting is not the one giving the highest reading.This was the case below where the readings were lower when maximum current of 34 kA was used versus the other setting with 11 kA.In this case an amplitude-modulated current would have been a better choice than the thyristor-shaped reduction in current.This, since the latter-introduced harmonics which made the total exceed the action level, we see further in Figure 3.
Figure 4 shows the shape of the curve when a newer welding machine is used.This gives a lower exposure of the operator.In this case an amplitude-modulated current would have been a better choice than the thyristor-shaped reduction in current.This, since the latter-introduced harmonics which made the total exceed the action level, we see further in Figure 3. Figure 4 shows the shape of the curve when a newer welding machine is used.This gives a lower exposure of the operator.
Taking measurements for showing compliance with the directive for a spot-welding machine, it is necessary to have a peak hold instrument with a filter according to EU directives.It is quite likely that the readings will exceed the actions levels at some distance from the machine, and if the operator cannot move further away from the electrodes it will be necessary to evaluate if the exposure limit values (ELVs) are exceeded.To do this, numerical calculations are needed, also for the input to the waveform of the curves, plus a more complete measurement of the field distribution around the electrodes is needed.In most cases this is not an option, and the action levels become the de facto limits.
Nadeem et al. [9] calculated induced body currents in a full 3D human model of an operator of a spot-welding machine.When the operator was less than 0.3 m from the electrode arms of the machine, the maximum induced current density exceeded the basic  Taking measurements for showing compliance with the directive for a spot-welding machine, it is necessary to have a peak hold instrument with a filter according to EU directives.It is quite likely that the readings will exceed the actions levels at some distance from the machine, and if the operator cannot move further away from the electrodes it will be necessary to evaluate if the exposure limit values (ELVs) are exceeded.To do this, numerical calculations are needed, also for the input to the waveform of the curves, plus a more complete measurement of the field distribution around the electrodes is needed.In most cases this is not an option, and the action levels become the de facto limits.
Nadeem et al. [9] calculated induced body currents in a full 3D human model of an operator of a spot-welding machine.When the operator was less than 0.3 m from the electrode arms of the machine, the maximum induced current density exceeded the basic restriction of ICNIRP (1998) [10], although the spatially averaged MF over the whole body was well below the reference level.
The time scale was 50 ms per division.Note the different scales on the y-axis.

Frequency Converters
In 2016, the Swedish National Electrical Safety Board published a report authored by Olsson [11] where it was reported that 69% out of 39 installations of frequency converters had EMC-related deficiencies.From the report: "Remarkably often, the right type of equipment has been used in the installation, but in an incorrect way.Several mistakes are constantly repeated, where the absence of zoning is the most important, caused by inappropriate construction of control cabinets, incorrect wiring and inappropriate placement of constituent components.The shortcomings indicate that installers lack a basic understanding of the concept of electromagnetic compatibility (EMC)".
We have often at our site visits to check compliance with the EMF EU directive noticed high electric fields from the installation of frequency converters, often near the cable connecting the converter to the motor.As with the EMC problem, this is due to the installation not being completed in a correct way.Examples are that the motor cable is not shielded 360 • or not correctly bonded for high frequencies-i.e., only using ordinary earthing.Sometimes a switch is installed between the converter and the motor, and if the box is not EMC-safe, the fields will be leaking.
Not only is this an EMC problem but it also conflicts with the action levels for electric fields from the converter.Since the operation of the converter is based on chopping the current into pulses with short rise and fall times, the voltage will not only contain the ELF frequencies, but the spectrum will cover both the kHz and MHz range.Because of this, when the electric field is measured with the weighted peak method, including frequencies up to 400 kHz, the action level is exceeded.
In view of this, it is of importance to make sure the frequency converter in use at the factory is correctly installed both from an EMC point of view as well as regarding occupational exposure to electric fields.

Handheld Electric Tools
When using handheld electric tools exposure to a low-frequency magnetic field is to be expected.The B-field can reach a few parts of mT at close distance, and as such they do not exceed international guidelines.However, when it comes to time-averaging, sometimes a problem occurs.Hansson Mild et al. [1] brought up an example of a handheld drill and showed that the start of the current was 10 times higher during the first few periods, as can be seen in Figure 5, than the steady state current.Following Standard No. EN 62233:2008 [12], the first 200 ms should be neglected, and the measurement should be taken at a certain distance from the machine.This exception then becomes very questionable.The limits for ELF exposure are set to protect against nerve excitation, and according to Reilly [13], this can occur within a half-period of the power frequency alternating current (AC), i.e., during exposure of <10 ms.The time average of the exposure from a spot-welding machine needs further discussion.The basic limits are set in root mean square (rms) values for field strengths, but should the averaging be calculated over one second or a shorter time period?Different answers are given in various standards; however, most commercially available instruments use one second as an averaging time.In Directive 2004/40/EC [2], the specific averaging time for frequencies < 100 kHz is given.Standard No. C.95.6:2002 [14] gives the rms averaging time as the longer of 0.2 s or the time for 5 cycles (up to 10 s).But using this standard might be problematic.In spot-welding the total welding time is typically shorter than one second, only a few periods of 50 Hz (i.e., an order of tens or hundreds of ms) (See further Hansson Mild et al. [1]).The whole weld is over before the averaging time is up.
We are not aware of any new publications dealing with the exposure from handheld tools, but there is a need to clarify these questions with more systematic measurements of different tools.

Radiofrequency EMF
The most common source for high exposure to radiofrequency electromagnetic field (RF EMF) occurs near dielectric heaters such as plastic welding machines or glue dryers.Before taking any measurement, one has to select the worst-case scenario.The leakage fields from the machine depend on things like type of plastic to be welded, number of layers, electrode length, etc.; see further information in [15].Therefore, a good knowledge of the process is needed and to see how the operator works near the electrodes.
Figure 6 shows the recordings of two different settings on a machine producing the same weld.The leakage field is not constant during the welding time and no instrument available integrates over the pulse, and therefore a peak hold approach is usually the answer.Note that measurements are needed both for the magnetic and the electric field component.
tools, but there is a need to clarify these questions with more systematic measurements of different tools.

Radiofrequency EMF
The most common source for high exposure to radiofrequency electromagnetic field (RF EMF) occurs near dielectric heaters such as plastic welding machines or glue dryers.Before taking any measurement, one has to select the worst-case scenario.The leakage fields from the machine depend on things like type of plastic to be welded, number of layers, electrode length, etc.; see further information in [15].Therefore, a good knowledge of the process is needed and to see how the operator works near the electrodes.
Figure 6 shows the recordings of two different settings on a machine producing the same weld.The leakage field is not constant during the welding time and no instrument available integrates over the pulse, and therefore a peak hold approach is usually the answer.Note that measurements are needed both for the magnetic and the electric field component.
Figure 6.The electric field strength squared (E 2 ) from a plastic sealer recorded for two combinations, A and B, of welding time and tuning, giving welding seams of the same quality.The area under the curves is approximately equal, which means that the same amount of energy is needed to produce the welding seams.The settings in the two recordings are the same.From Eriksson and Hansson Mild [15].

Time Average
The total welding time must be recorded since the time average in the directive is over the 6 min time interval (as a note, we may indicate that in the original version a long time ago, the time interval was given as 0.1 h since the body temperature response time is not more precise that this).The total number of welds per 6 min period needs to be counted and the total welding time added.

Time Average
The total welding time must be recorded since the time average in the directive is over the 6 min time interval (as a note, we may indicate that in the original version a long time ago, the time interval was given as 0.1 h since the body temperature response time is not more precise that this).The total number of welds per 6 min period needs to be counted and the total welding time added.
The next step is to measure both the contact currents and induced current.Note that the induced current strongly depends on the type of floor the worker is standing on.If the floor is made of reinforced concrete, from an RF point of view this can be considered as electric ground, leading to a high current in the legs of the operator.It is wise to isolate the worker from the electric ground by, for instance, using a small wooden platform.This would reduce the induced current considerably.
The field measurements must be taken without the operator disturbing the field distribution, and this can be hard to do sometimes.
Zubrazak et al. [16] published some mitigation measures for reducing the exposure of operators of RF sealers, ranging from simple and costless and ending in dedicated EMF-shielding systems.
The directive is based on the guidelines given be ICNIRP, 1998 [10] and presumably when the directive is updated it might follow the new update ICNIRP guidelines from 2020.The new guidelines have introduced a frequency dependence up to 30 MHz, whereas in the earlier version and the directive the flat frequency response started at 10 MHz.Since the dielectric heater often works at frequencies around 13 or 27 MHz, this must be considered.
Most commercial E and H field instruments have a frequency response filter according to the old version.The conservative approach would therefore be to use the old instruments!If the new ICNIRP 2020 [17] is implemented in a new updated EU directive one must measure frequency, which changes during the weld!

Uncertainty in Measurements
In the directive it is said: "If compliance with the ELVs cannot be reliably determined based on readily accessible information, the assessment of the exposure shall be carried out on the basis of measurements or calculations.In such a case, the assessment shall consider uncertainties concerning the measurements or calculations, such as numerical errors, source modelling, phantom geometry and the electrical properties of tissues and materials, determined in accordance with relevant good practice." In Figure 7, the problem of taking into account the uncertainty in measurements is illustrated.The limit value in the directive is represented by the line; it might be the electric or magnetic field value.The mean value of the measured value is represented by the dot.The uncertainties are added to this value, including both instrumental errors (non-isotropicity in the probe, nonlinearity, etc.) and the calibration error, and the standard deviation of the measured values.If we chose to go with 95 or 99% probability, it will lead to different sizes of the error bars.Next, we must be aware of if the measurement is taken for the employer to show compliance or as part of helping the work inspectorate.In the first case, the conservative approach says that the measured value including the 95% limit should stay below the limit value to show compliance.But, if the measurements are calculated for the work inspectorate, then it works the other way; the value plus the 95% values should stay above the limit to be sure that the limit is exceeded.In the first case, one must be certain that the values are below the limit, whereas in the second case, one has to be certain that the values are above the limit.Thus, they display different outcomes of the measurement.
Using a shared uncertainty budget, it is sufficient that the mean values are below or above the line.This approach is often used when the error bars are small, but in the case of electromagnetic fields, this is certainly not the case.
For some of the most common instruments used for measuring RF EMF, the manufacturer specifies, as seen in Figure 8, the following: Instrument error: ±1 dB frequency In the first case, one must be certain that the values are below the limit, whereas in the second case, one has to be certain that the values are above the limit.Thus, they display different outcomes of the measurement.
Using a shared uncertainty budget, it is sufficient that the mean values are below or above the line.This approach is often used when the error bars are small, but in the case of electromagnetic fields, this is certainly not the case.
For some of the most common instruments used for measuring RF EMF, the manufacturer specifies, as seen in Figure 8, the following: Instrument error: ±1 dB frequency response, ±1 dB for lack of isotropicity, ±1 dB from calibration.Add to that uncertainty in individual measurements.

Figure 7.
Examples of how to handle the uncertainty in measurement.The line represents a fictive limit value.The dots are the mean values from fictive measurements.Then, the different scenarios show the problems when the measurements are performed for the employer or the work inspectorate.In the first case, one must be certain that the values are below the limit, whereas in the second case, one has to be certain that the values are above the limit.Thus, they display different outcomes of the measurement.
Using a shared uncertainty budget, it is sufficient that the mean values are below or above the line.This approach is often used when the error bars are small, but in the case of electromagnetic fields, this is certainly not the case.
For some of the most common instruments used for measuring RF EMF, the manufacturer specifies, as seen in Figure 8, the following: Instrument error: ±1 dB frequency response, ±1 dB for lack of isotropicity, ±1 dB from calibration.Add to that uncertainty in individual measurements.After taking all of this into account it is hard to see how RF measurements in industrial settings could be completed without including at least ±3 dB to the measured value when comparing with the action levels in the directive.

Workers at Particular Risk
The EU directive addresses specifically the assessment of risk for workers wearing active or passive implanted medical devices, such as cardiac pacemakers, and workers with medical devices worn on the body, such as insulin pumps, and pregnant workers.In the "Non-binding guide to good practice for implementation of the EU directive" several examples are given [3].The issue has also been discussed by [18].
In case of active implants, for instance a cardiac pacemaker or defibrillator, the risk comes from interference with the EMF and risk for malfunction of the device.When conducting a risk assessment in these cases, it is necessary to look at the limits provided by the manufacturer of the pacemaker.If these are not at hand, then one should use the limits After taking all of this into account it is hard to see how RF measurements in industrial settings could be completed without including at least ±3 dB to the measured value when comparing with the action levels in the directive.

Workers at Particular Risk
The EU directive addresses specifically the assessment of risk for workers wearing active or passive implanted medical devices, such as cardiac pacemakers, and workers with medical devices worn on the body, such as insulin pumps, and pregnant workers.In the "Non-binding guide to good practice for implementation of the EU directive" several examples are given [3].The issue has also been discussed by [18].
In case of active implants, for instance a cardiac pacemaker or defibrillator, the risk comes from interference with the EMF and risk for malfunction of the device.When conducting a risk assessment in these cases, it is necessary to look at the limits provided by the manufacturer of the pacemaker.If these are not at hand, then one should use the limits set up for the general public in the directive from 1998 [19].As an example, the limit for the general public for a 50 Hz magnetic field is set at 100 µT vs. 1000 µT for occupational exposure AL low .Similarly, for 27 MHz, the electric field for the general public is set at 28 V/m vs. 61 V/m for the occupationally exposed.
When surveying a workplace for EMF exposure, it is thus necessary to keep in mind both sets of limits, even if none of the workers are currently wearing an implant or are pregnant, since in the future one might have a new situation.Then, a new risk assessment must be made, in which case it is good to have both sets of measurements at hand for the new situation.

Need to Inform
To handle a situation involving workers at particular risk, information is a necessary tool in this, both from the employer and the employee.The employer needs to inform the worker at the site about the level of exposure they might encounter here.If the levels are above the occupational levels, obviously direct actions are needed, but in case the levels are above those for the general public but below the occupational action levels, it is necessary to inform the workers that if any change occurs in their medical status-for example, receiving an implant or becoming pregnant-a new risk assessment has to be made and it is therefore of outermost importance to inform about the new situation.In vol. 2 of the non-binding guide [3], several examples are given of how adaptations of the workplace can be completed.

Magnetic Resonance Imaging (MRI)
The MRI machine uses three different types of fields: a static magnetic field, a switched low-frequency magnetic field and a radiofrequency field.All these types of fields are at such high levels that they are at a level close to what the human can tolerate, and therefore clear guidelines for staff and patients have been in place even before the directive, as can be seen further in [20].There are also safety courses for the staff on how to work safely with the equipment, both from the aspect of patient and staff safety.The American College of Radiology [21] has issued guidance documents on MR safe practices; if these are followed, there is no need for direct measurements around MRI machines for testing compliance with the directive.
Keevil and Lomas [22] discuss the derogations for MRI and point out the need to provide staff with relevant information and training regarding EMF exposure.

Conclusions
As can be seen from the issues brought up in this paper, taking measurements of electromagnetic fields to show compliance with the regulations based on, for instance, the EU directive [2] or other similar documents demands good knowledge of the document as well the use the text in the document as the instrumentation used for the measurement.In the latter case, knowledge of the frequency response and the calibration of the instrument is of the utmost importance.It is also necessary to state the magnitude of the errors in the measurement before comparing them with the limits.
In the EU directive, the competence and equipment used for showing compliance is only shortly mentioned as: ". . . the assessment, measurement and/or calculations . . .shall be planned and carried out by competent services or persons at suitable intervals, t. .." (EU dir article 4, point 4).
It is clear from the problems discussed herein that the demands on competence and equipment should be more strictly regulated.

Figure 1 .
Figure 1.Example of the waveform from a spot-welding machine operating at 34 kA current.Measurements were taken with a Narda ELT 400 with a PICO oscilloscope attached.Time scale: 50 ms per division.

Figure 2 .
Figure 2. Same machine as in Figure 1 but now with another setting giving 11 kA current.

Figure 2 .
Figure 2. Same machine as in Figure 1 but now with another setting giving 11 kA current.

Figure 3 .
Figure 3. Frequency spectra of the weld in Figure 2. The waveshape with sharp edges leads to a substantial influence of harmonics with 150, 250 Hz, and so on.

Figure 3 .
Figure 3. Frequency spectra of the weld in Figure 2. The waveshape with sharp edges leads to a substantial influence of harmonics with 150, 250 Hz, and so on.

Figure 4 .
Figure 4. Newer spot-welding machine which uses a technique with kHz in the pulse.This often means that the exposure to the operator is less than or within the action levels.Time scale: 5 ms/division.

Figure 5 .
Figure 5. Measurement of the current in a handheld drill.During start-up the current is ten times higher than in the steady state mode.

Figure 6 .
Figure 6.The electric field strength squared (E 2 ) from a plastic sealer recorded for two combinations, A and B, of welding time and tuning, giving welding seams of the same quality.The area under the curves is approximately equal, which means that the same amount of energy is needed to produce the welding seams.The settings in the two recordings are the same.From Eriksson and Hansson Mild [15].

Figure 7 .
Figure 7.Examples of how to handle the uncertainty in measurement.The line represents a fictive limit value.The dots are the mean values from fictive measurements.Then, the different scenarios show the problems when the measurements are performed for the employer or the work inspectorate.In the first case, one must be certain that the values are below the limit, whereas in the second case, one has to be certain that the values are above the limit.Thus, they display different outcomes of the measurement.

Figure 7 .
Figure 7. Examples of how to handle the uncertainty in measurement.The line represents a fictive limit value.The dots are the mean values from fictive measurements.Then, the different scenarios show the problems when the measurements are performed for the employer or the work inspectorate.In the first case, one must be certain that the values are below the limit, whereas in the second case, one has to be certain that the values are above the limit.Thus, they display different outcomes of the measurement.

Figure 8 .
Figure 8. Example of uncertainties as given by Narda for one of their electric field probes, 100 kHz to 3 GHz.

Figure 8 .
Figure 8. Example of uncertainties as given by Narda for one of their electric field probes, 100 kHz to 3 GHz.