The heating effect from RF devices causes the most concern from an RF safety point of view. The human body counters local heating by thermoregulation (blood flow through the affected organs). The eyes and male testes are particularly susceptible to RF heating because these organs have no direct blood supply and, hence, no way of dissipating heat. The heating effects in biological tissue escalate with the increase in frequency, although the heat's penetration depth decreases.

With the proliferation of cellular phones, most RF safety concerns have focused on RF absorption by the head, particularly from mobile handsets. The dose of RF exposure is linked to exposure time: maximum SAR is normally averaged over a 6-minute period during the 24-hour day.

Some concerns have focused on other effects of RF exposure. Most communications systems are pulse-like in nature, and their effects on brain function have been discussed recently. For example, the global system for mobile communications (GSM) frame rate, at 8.33 Hz, is close to that characteristic of alpha waves in the brain. Although there is no conclusive proof of such effects, considerable research is currently examining the effects of RF. Much of the research in this area was sparked by a report published by the Independent Expert Group on Mobile Phones, chaired by Sir William Stewart. The report, released in April 2000, is also known as the Stewart Report.

Several organizations have set exposure limits for acceptable RF safety via SAR levels. The International Commission on Non-Ionizing Radiation Protection (ICNIRP) was launched as an independent commission in May 1992. This group publishes guidelines and recommendations related to human RF exposure.

For the American National Standards Institute (ANSI), the RF safety sections now operate as part of the Institute of Electrical and Electronic Engineers (IEEE). IEEE recently wrote one of the most important publications for SAR test methods.¹

In the UK, the National Radiological Protection Board (NRPB) sets SAR limits. SAR limits are expressed for two different classes of people: workers (occupational/controlled exposure) and the general population (uncontrolled exposure). Because the general-population exposure is considered to be uncontrolled, the limit for this group is five times more stringent than the limit for the workers, whose environment and exposure can be monitored and controlled.

The limits are defined for exposure of the whole body, partial body (e.g., head and trunk), and hands, feet, wrists, and ankles. SAR limits are based on whole-body exposure levels of 0.4 W/kg^­1 for workers and 0.08 W/kg^­1 for the general population. Limits are less stringent for exposure to hands, wrists, feet, and ankles. There are also considerable problems with the practicalities of measuring SAR in such body areas, because they are not normally modeled. In practice, measurements are made against a flat phantom, providing a conservative result.

SAR testing was originally performed by measuring minute changes in temperature at specific locations in a tissue-simulant material. The tissue simulant had to be extremely viscous to prevent convectional currents from producing errone-ous results. SAR probes can still be calibrated by this method.

Several key developments have been made in SAR test methods. Manufacturers are required to use a new head phantom called the specific anthropomorphic mannequin (SAM) phantom. SAM is based on the 90th percentile of a survey of American male military service personnel and represents a large male head. The SAM phantom, which has human features (ears, nose, etc.), replaces the featureless generic twin phantom. SAM has extremely well-defined dimensions, particularly for parameters such as phantom shell thickness.

Fluid properties for SAR testing are now well defined. The methods are also well defined for making and measuring fluids for the most common frequencies used in testing. The IEEE P1528 specification contains excellent references for fluid properties and methods. It is essential to verify that fluid properties are within the tolerances of the specifications.

Measurement uncertainties are defined in the specifications. Overall measurement uncertainties must be below 30% for a 95% confidence level. An uncertainty in measurements of 30% may seem a bit high, but this percentage is small in decibel terms. EN 50361 lists 21 individual uncertainty contributions.² Depending on the setup, additional contributions may be required.

Most SAR probes now measure E-field in volts per meter (V/m^­1), which allows SAR to be calculated. In addition to the E-field present, SAR is also dependent on the conductivity and permittivity of the tissue simulant. The equation used to calculate temperature-change SAR relates directly to the one used in current measurements.

SAR probes must be physically small. They must also have good spherical isotropy (i.e., measure equal amounts of E-field regardless of the angle or direction that the probe points toward the radiation source). In addition, SAR probes and their associated test setups must be designed so that they have an insignificant effect on the RF field.

For newer test methods, the probe is positioned at various points within either a phantom head or body filled with an appropriate tissue-simulant liquid. Head and body phantoms, in general, can only represent the shape of the human body; they do not, for example, mimic bone structure. Phantom heads have been produced that mimic the tissue structure of a human head with skin, bone, muscle, and brain tissue. However, these tissue phantoms are not practical for SAR testing. The probe cannot be moved within them, hence, the use of homogeneous phantom shells filled with tissue-simulant liquids. The phantoms do not take into account natural body thermoregulation by bloodflow; therefore, the rates of temperature rise within the body deduced from SAR measurements include a safety margin.

Because no known recipes for fluids are representative of body tissue at all frequencies, different tissue simulant fluids are required for different frequencies (e.g., 900 MHz for GSM 900 and 1800 MHz for 1800 products). The brain simulant must be calibrated to ensure that the permittivity and conductivity are correct for the frequency being tested. Fluids are often made from a mixture of distilled water, sugar, and salt. Some frequencies, however, require other chemicals to obtain the required properties.

SAR testing is performed on handset devices by placing them at various positions on both sides of the phantom head. The tip of the SAR probe is moved to exact points in a three-dimensional grid within the tissue simulant. A complex mathematical formula then calculates the volume-averaged SAR using extrapolation and interpolation processes.

Several groups have pushed recently to standardize test methods for SAR testing, including uncertainty calculations. Although new standards for measurement have been issued, the overall SAR limits have not changed. CENELEC and IEEE have produced similar specifications because the majority of people involved in writing them were on both boards. The CENELEC standard, EN 50360, has recently been published in the Official Journal of the European Communities as a harmonized standard. EN 50360 references EN 50361, which contains the test methods. SAR test method specification IEEE P1528 is already in draft format and should be due for release shortly.

In Europe, a key problem with the CENELEC standard is that it is only concerned with devices held next to the human ear, that is, handset testing next to a phantom head. EN 50360 is applicable to all RF devices that are "to be used in close proximity to the human ear."³ The standard does not contain the actual limits. Actual limits can be found in either the ICNIRP Guidelines (April 1998) or Council Recommendation 1999/519/EC Annex II.^4,5 EN 50360 applies to devices transmitting with an average power greater than 20 mW and in the frequency range of 300 MHz to 3 GHz.

Devices that transmit ¾20 mW are "deemed to comply with the basic restrictions without testing." No standards have been harmonized for devices other than those such as mobile phones and cordless phones. However, manufacturers must still comply with the EU SAR limits for devices such as PDAs that have an integral RF module for GSM. Such devices are tested against flat phantoms that simulate body parts.

In the United States, the limits and applicable products are contained in Title 47 of the Code of Federal Regulations 47 CFR Part 2.1093, which covers portable devices with transmitters within 20 cm of a user's body.⁶ It also includes an applicability list that encompasses virtually all radio products, depending on their output power. A full explanation of the relevant parts, SAR limits, and SAR test methods is contained in FCC OET Bulletin 65 Supplement C.⁷

For the UK, the Stewart Report recommends that information on SAR values for mobile phones should be readily accessible to consumers at the point of sale. For example, the report recommends that the information be printed on the product's box. The report also suggests that stores provide leaflets with explanatory and comparative information. Other recommendations include placing the information on the phone's label, and making it available via the phone's display. The report also recommends publishing such data on a national Web site.

New well-defined testing methods incorporate recent key developments.

In the United States, the Cellular Telecommunication Industry Association (CTIA) requires that any mobile phone it certifies be sold with explanatory information. This information must confirm that the phone has passed FCC safety standards. Manufacturers must also include applicable SAR data for that phone and an explanation of how the SAR testing was done.