A protocol using the wrong-sized probe can lead to erroneous radiation-exposure measurements and inaccurate test results.

Chris Zombolas

 

Chris Zombolas

A recent study reported that its measurement protocol supported a finding that the use of hands-free kits with mobile phones leads to a threefold increase in radiation exposure to the brain. The report, which was published by the UK consumer association Which?, also claimed that the specific absorption rate (SAR) measurements currently used for evaluating the radiation exposure levels in the brain are somehow flawed.

This editorial assesses the protocol used in the UK study. For comparison, an overview of SAR measurement procedures is provided that explains the SAR measurement method for the evaluation of the radiation-exposure levels from both mobile phones and hands-free kits (see the sidebar "SAR Measurements"). Test data derived from tests recently completed for Australia's consumer association, Choice, are used to demonstrate that this method is technically justified and that exposure at the ear is greatly reduced by the use of hands-free kits.

The UK study states:

 

Figure 1. Test layout used in the UK study for measurements inside the phantom head.

"Experimental EMC investigations showed that under particular conditions that could realistically be achieved, in practical use, one can obtain higher electric fields from the earpiece of the hands free kit than from the earphone of the mobile phone alone. Care is required to set up the hands free kit to observe this effect and this probably explains why no one else has measured it before. When the hands free kit is connected to the mobile phones there are electrical currents conducted and induced along the wire. This produces radiated electric fields along the length of the hands free kit wire. A unique condition exists whereby higher electric fields are radiated from the earpiece of the kit when the wire is run closely parallel to the length of the phone and the wire is looped in a uniform configuration. The critical parameter is the distance d between the earpiece and the tip of the mobile phone's aerial. At certain distances of d, there is a tuning effect that increases the electric field at the kit's earpiece [see Figure 1]. These more recent tests have confirmed the findings of previous work, but have added a much greater insight into the physics of the coupling mechanism."

For the UK study, relative measurements were performed first in air and then inside a phantom head filled with tissue-simulating liquid. An Emco 904 ball electric (E)­field probe was used to perform the relative E-field measurements at the ear of the phantom head. E-field measurements were performed with and without the hands-free kits attached (i.e., the difference between the two states) so that only relative measurements were necessary. The UK association apparently did not want to determine the SAR value, but rather simply wanted to determine whether the kits increased or decreased the field levels in the head. Its report said, "This approach was justified because SAR was directly related to the electric field."

Examining the Protocol

The study's methodology is fundamentally flawed primarily because of the use of an inappropriate E-field probe. The ball E-field probe is a simple and inexpensive sniffer probe used for detecting the presence of fields rather than for measuring absolute field levels.

Although these probes are suitable for relative measurements in certain situations, the measurements used in this study were, in effect, not relative. Relative measurements are only possible when all things are equal except for one variable, which in this case would be the radiating E-field source.

The relative measurements performed for this study first measured the reference E-field with the phone at the ear. This was measured at 10 and 30 mm from the ear inside the liquid-filled phantom head. The phone was then placed in a position representing the waist, and the earpiece was placed at the ear. The E-field was again measured at both 10 and 30 mm, and the measured values were compared to the reference measurement. The difference between the two voltages at the output of the probe was claimed to be the increase (or decrease) in the E-field at the ear position.

Parasitic EMC Effects of Probe. The Emco 904 probe is a simple conductive ball (3.6 cm diam) connected to the inner conductor of a coaxial cable. It has an internal 50-W impedance between the ball's surface and its metal stem. The large conductive elements cause loading of high-impedance E-fields present in the extreme near field of a radiating source. This probe is also sensitive, to some extent, to magnetic (H) fields, so its output voltage is not necessarily directly proportional to the E-field level alone. The stem of the probe is 11 cm long and almost parallel with the vertically hanging hands-free kit wires.

There are no absorber ferrites along the probe's stem; therefore, the stem is a very efficient receiving antenna at 900 MHz and can pick up signals from the nearby hands-free kit wires. The study's test to check the probe for spurious pickup was flawed because the 50-W termination was placed at the end of the cable rather than at the end of the probe's sensing element. It should have been placed instead at the same position as the E-field ball sensor.

The probe's conductive stem also confuses the output readings of the probe because the stem receives stray fields from all sources, including from the hands-free kit wires and the phone antenna, as well as reflections from the equipment and the ground plane. In the case of the kit's earpiece measurements, the major contributors to the voltage measured at the probe output would have been from the coupling of radiation from both the kit wires and the phone antenna to the probe stem. Such coupling is a very predictable and commonly known EMC effect. The effect of this flaw on the protocol would be to elevate the voltage (field strength) reading when the radiation from the earpiece was being measured. The output voltage would have been greatest when the hands-free kit wires were tuned to give a maximum field near the end of the wires and closest to the probe.

The report assumed incorrectly that the (elevated) output of the E-field probe was due to the radiation from the earpiece alone. The secondary effects would have distorted the hands-free kit earpiece measurements, thereby invalidating the comparison with the probe voltage obtained when the phone was at the head; i.e., the measurements were not truly relative. A properly designed SAR probe mitigates this problem.

Probe Sensor Problems. Although the report claimed that the measurements were relative E-field measurements, they were, in effect, relative SAR measurements (albeit flawed), because they involved E-field measurements in a lossy liquid. The Emco 904 probe is unsuitable for measurements in liquid, very close to the source, or close to different dielectric materials and surfaces (see the sidebar "SAR Probe Requirements").

The field distortion effects on the phone would be far greater than the effects on the hands-free kit earpiece. Because the size and radiating properties of the phone and the hands-free kit earpiece differ greatly, the locations of the spatial peaks and nulls are unpredictable. Without scanning the Emco probe inside the phantom head, it would be impossible to determine whether the probe was in a spatial peak or a spatial null. Relative measurements, therefore, would not have been possible under the conditions described.

For the study, the isotropic response of the Emco probe was checked in the far field with the stem orthogonal to the field. A ±1 dB isotropic response is unjustified for extreme near-field measurements with the stem nearly parallel with the radiating hands-free kit wires.

The Emco probe's relatively large dimensions result in very poor spatial resolution, and the probe greatly disturbs the steep field gradients in which it is immersed. The boundary effects of different dielectric media cause highly anisotropic fields near the (different) glass surface, making it nearly impossible to measure. Boundary effects are directly dependent on the probe dimensions and diminish as distance from the boundary increases. These factors probably explain why the reference levels in the UK report were lower at 10 mm from the internal ear boundary than those at 30 mm. The 10-mm result is inconsistent with known effects and is indicative of an artifact.

Further indications of measurement flaws were not explained. For example, some of the report's reference measurements at 30 mm from the ear's internal boundary with the phone at the head show a reading higher than the reference measurements in air.

Comparison of Protocols

Australia's Choice report provides a detailed description of the study's protocol and results, including comprehensive SAR plots and photographs of each setup. Some of the assertions in the UK report are critical of the suitability of SAR measurements, and the report criticized some aspects of the SAR test and DTI tests. However, it fails to explain why its findings disagreed with other SAR-based studies.

 

Figure 2. SAR levels, phone at touch position, from Australia's Choice study.

The Choice study, for example, tested two positions of the hands-free kits, one with the phone at the waist and another with the phone in an outstretched-hand position (80 cm above the floor, taped to a wooden plank). The Choice protocol did not investigate the length of the hands-free kit wires because this factor was considered less important than others. The phone was mounted at the waist position of the phantom with the hands-free kit wires taped along the phone and the phone antenna. The wires leading to the earpiece and microphone were taped along the body of the phantom to simulate the body's RF loading effects.

The second test in the Choice study involved placing the phone in the extended-hand position with the hands-free kit wires free in the air above the ground plane. When each of the three phones and six combinations of hands-free kit wiring were tested, a different wire length was used inadvertently. Results in the UK study show a virtual notch or very sharp dip for wire lengths that gave low hands-free kit earpiece readings. This single sharp dip represents a small range of wire lengths that were apparently favorable to reducing radiation at the earpiece. In most of the test cases reported, the dip corresponded to small (sometimes <5 cm) wire-length changes. Most other wire lengths indicated higher readings.

On the basis of these results, it would, therefore, appear unlikely that for all six of the Choice tests, the hands-free kit wires would have been fortuitously set up to the exact length necessary to produce lower readings for the hands-free kit earpiece measurements. Although tuning wires can increase the fields from the earpiece, the Choice tests confirmed that the earpiece is not a significant radiator compared to the phone itself.

Post-Test Analysis of Choice Study

The Choice SAR tests reported the highest SAR both with and without the hands-free kit. In most cases, the highest SAR for the phone was located at the cheek or jaw; nevertheless, the SAR levels at the ear (due to the phone) were still measured and plotted. The plots in the Choice report show that the SAR (or E-field) at the ear due to the phone was still much higher than with the hands-free kit earpiece. This conclusion is supported by data listed in Tables I­III.

 

Phone SAR Measured Inside Skull		Hands Free Kit SAR	Phone Position	SAR Reduction at Ear (%)
Earpiece
Phone Antenna Position SAR (A)	SAR at EarPosition (B)	SAR at EarPosition (C, D)
0.1	0.3	0.02	Waist (C)   Hand (D)	93
0.1	0.3	0.01		96
0.1	0.2	0.01		95
0.2	0.2	0.03		85
0.4	0.3	0.03		90
0.4	0.3	0.03		90
A. With the phone at the ear, the SAR plot was used to determine the SAR value at the head, directly at the location of the antenna of the phone. B. With the phone at the ear, the SAR was determined at the ear. This corresponded to the location of the phone's earphone (speaker). C. With the phone placed at the waist position of the phantom, and the hands-free kit wires taped along the "body" of the phantom and the head, the SAR was determined inside the phantom in the region of the ear. D. With the phone placed in air, about 80 cm above the ground plane, and the hands-free kit wires hanging away from the body, the SAR was determined in the region of the ear.

Table I. Ericsson A1018s at 900 MHz.

 

Phone SAR Measured Inside Skull		Hands Free Kit SAR	Phone Position	SAR Reduction at Ear (%)
Earpiece
Phone Antenna Position SAR (A)	SAR at EarPosition (B)	SAR at EarPosition (C, D)
0.15	0.45	0.004	Waist (C)   Hand (D)	99
0.20	0.41	0.005		98
0.18	0.35	0.004		99
0.11	0.45	0.08		82
0.10	0.37	0.08		78
0.8	0.34	0.04		88
A. With the phone at the ear, the SAR plot was used to determine the SAR value at the head, directly at the location of the antenna of the phone. B. With the phone at the ear, the SAR was determined at the ear. This corresponded to the location of the phone's earphone (speaker). C. With the phone placed at the waist position of the phantom, and the hands-free kit wires taped along the "body" of the phantom and the head, the SAR was determined inside the phantom in the region of the ear. D. With the phone placed in air, about 80 cm above the ground plane, and the hands-free kit wires hanging away from the body, the SAR was determined in the region of the ear.

Table II. Nokia 5110 at 900 MHz.

 

Phone at Head		Hands Free Kit SAR at Ear		Phone Position	SAR Reduction at Ear (%)
SAR at Antenna (A)	SAR at Ear (B)	Antenna Retracted (C)	Antenna Extended (C)
0.40	0.53	0.005	0.005	Waist (C)   Hand (D)	99
0.55	0.61	0.002	0.003		99
0.70	0.84	0.001	0.002		99
0.43	0.63	0.22	0.21		65
0.41	0.83	0.11	0.11		86
0.50	0.83	0.09	0.10		88
A. With the phone at the ear, the SAR plot was used to determine the SAR value at the head, directly at the location of the antenna of the phone. B. With the phone at the ear, the SAR was determined at the ear. This corresponded to the location of the phone's earphone (speaker). C. With the phone placed at the waist position of the phantom, and the hands-free kit wires taped along the "body" of the phantom and the head, the SAR was determined inside the phantom in the region of the ear. D. With the phone placed in air, about 80 cm above the ground plane, and the hands-free kit wires hanging away from the body, the SAR was determined in the region of the ear.

Table III. Nokia 252 AMPS at 900 MHz.

Figure 3. SAR levels in the Z plane, HFK earpiece phone at waist, from Australia's Choice study.

The SAR values were derived from the Choice report, and they clearly indicate that radiation to all parts of the head, including the ear, is greatly reduced with the hands-free kit. The Choice protocol located the worst-case spatial SAR peak by scanning the inside volume of the phantom head. Hundreds of SAR measurements were performed in each predetermined cube of brain tissue. The SAR plots in Figures 2 and 3 clearly show the location of the SAR contours inside the phantom relative to the phone. The SAR (or E-field) level is plotted against the distance from the earpiece as the field penetrates the skull into the brain. It is possible to plot the SAR distributions in the x, y, and z axes. The SAR plots confirm that the fields decrease rapidly as the distance from the radiator increases (see Figure 3).

The assertion that SAR measurements have 100% (3 dB) uncertainty is misleading. State-of-the-art systems allow a 25% uncertainty (<1 dB). SAR measurements are now well refined and will continue to improve once the new standards (EN 50360, EN 50361, and proposed draft IEEE Std 1528-200X) are published. It is pertinent to note that 25% SAR absolute uncertainly is <1 dB.

Conclusion

The UK study's measurements, which used a large E-field ball probe (3.6 cm diam) with a long metal stem for performing SAR measurements, is fundamentally flawed. A state-of-the-art SAR measurement system is easily configured to perform the same tests performed for this study. Simply replacing the Emco ball probe with a SAR probe should disprove the report's findings. The study's assertions about the problems with or limitations of SAR measurements are incorrect. Using an appropriate SAR probe to repeat the protocol described should provide a result that is intuitively obvious: much lower fields with hands-free kits.

 

SAR Probe Requirements Because the field polarization is unknown in the reactive near-field of radiators, only probes with adequate isotropic response in the tissue dielectric should be used for SAR testing.1 This can be accomplished only by measuring the field components with three orthogonal dipole sensors with sufficiently small dimensions. The wavelength in the dielectric will be short, and the induced field will have large spatial gradients. It is absolutely necessary that the probe be as small as possible; it should be no larger than 4 mm. The probe's interaction with the field and the radiating source should be negligible. Conductive scatters in close proximity to the source can significantly modify the radiating source (the phone or earpiece).2 This, in turn, will greatly affect the spatial peak SAR value (or the field distribution). The probe and its stem must be radio-frequency (RF) transparent to avoid secondary modes of reception. The main requirements for a SAR probe in lossy liquid are: Fields measured in close proximity to dielectric discontinuities. Isotropic measurements made in a dielectric medium in the extreme near field. Better than 0.25 dB spherical isotropy. Good linearity. Minimal interaction with the field being measured. Field distortion around the probe causes measurement errors, and boundary effects cause multiple reflections between the sensor probe and nearby objects. Spatial resolution better than the smallest spatial dimension of any local field minima or maxima. In large volumes, the field distribution is highly nonuniform, and gradients vary greatly with different sources. E-fields in a liquid environment are highly anisotropic and unpredictable. Calibrations made in the same medium as that used to make measurements. High-resistance lines. Because secondary reception modes affect measurements, high-resistance lines are necessary so that the probe and its stem are RF transparent. Problems occur due to the direct coupling of the fields from the phone or the hands-free-kit wires, onto the probe's metal stem.

 

SAR Measurements In general, the peak SAR for a mobile phone is usually found near the jaw position because this is where the phone is in contact with the head. This test position is called up by current SAR standards and is referred to as the touch position. The localized exposure at the jaw is generally greater than the localized exposure to the brain or ear region. The draft SAR standard (prEN 50361) calls for an additional test in which the phone is placed in a tilting position whereby the antenna is moved toward the head and the phone is moved away from the jaw. This new tilt position results in the shifting of the maximum SAR value from the jaw closer to the earphone. The study conducted for Australia's consumer association, Choice, was based on measurements performed by the DASY3 system, which performs highly accurate and calibrated SAR measurements in liquid-filled phantoms. The system scans the entire volume of the phantom. The phone is operated at full power while the precision SAR probe (E-field) performs thousands of measurements over preprogrammed grid sizes with better than 1-mm spatial resolution. Measurements cannot be performed near the boundary of the phantom, so the results from inside the phantom are extrapolated using complex calculations that predict the SAR levels at the surface of the phantom. The relative positioning accuracy of the six-axis robot is better than 0.02 mm at any point inside the phantom. The E-field SAR probes are fully calibrated to meet SAR probe requirements. The dielectric properties of the brain-simulating liquid are measured prior to each test, and a complete system validation is performed by means of a small transmitting dipole placed at a known distance from the phantom so that a known SAR level is generated. An anechoic chamber is not strictly necessary for mobile phone SAR measurements because the measurements are made in the extreme near field where the E-field levels are much greater than the surrounding environment.

References

1.Katja Pokovic, "Advanced Electromagnetic Probes for Near Field Evaluations" (dissertation submitted to the Swiss Federal Institute of Technology for a degree of doctor of technical sciences, DISS.ETH No. 13334, 1999).

2.Niels Kuster, Quirino Balzano, and James C Lin, eds., Mobile Communications Safety (London: Chapman & Hall, 1997).

Chris Zombolas is founder and technical director of EMC Technologies Pty Ltd. (Tullamarine, Victoria, Australia). He is also a member of CE's editorial advisory board and can be reached at chris@emctech.com.au.

Back to July/August Table of Contents