Ij

Simulation Of Specific Absorption Rate Measurements

Speeds Wireless Design

By Stuart Nicol,

APREL Laboratories

Specific absorption rate (SAR) is a measure of the amount of radio frequency (RF)

energy absorbed by the body when using a wireless device. The Federal Communications

Commission (FCC) and other regulatory bodies have adopted limits for safe exposure to

RF energy that are expressed as SAR in W/kg. Several companies have developed SAR

measurement systems that manufacturers of wireless products can use to physically test

SAR associated with their products. In the process of designing an antenna used to

validate one of these measurement systems -- the ALSAS-10U (APREL Laboratories

SAR Assessment System), designed and developed by the systems team at APREL

Laboratories -- there was a need to develop a simulation model along with rules and

application that accurately predicts SAR measurement results to within a few percent

without the need for significant prototype development and testing. This makes it

possible to address SAR compliance problems in the early phases of the design process

when problems can be fixed at a much lower cost because much less money has been

invested in physical design changes. This article will explain the challenges that were

overcome in the development of this model's rules and application for use.

Figure 1: The ALSAS-10U system

As seen in the November 6, 2007 edition of the RF Globalnet (www.rfglobalnet.com) newsletter.

Case Study

No negative effects on human health have been demonstrated from the low levels of RF

energy produced by cell phones and other wireless devices. But regulatory bodies have

determined that the possibility remains that such effects may exist and so they have

developed standards for exposure to RF energy. The FCC requires wireless phones to

comply with a safety limit of 1.6 watts per kilogram over 1g of tissue (1.6 W/kg) in terms

of SAR. Many countries in Europe and elsewhere use similar exposure guidelines

developed by the International Commission on Non-Ionizing Radiation Protection

(ICNIRP). These guidelines are described in standards EN50361 and IEEE 1528, which

provide methods for measuring SAR that are nearly universally accepted by regulatory

bodies. Several companies have introduced measurement systems that typically focus on

the measurement of the electric field since E-field probes have fast time response and

sufficient sensitivity for SAR measurements.

Original Goal Was Establishing The Rules For Optimizing Dipole Design

Scientists from APREL Laboratories set out on the challenging task of extending the

capability of their ALSAS 10U SAR measurement system with the original goal being to

design a dipole antenna to validate the performance of measurements across a wide

frequency range for technologies operating at 5 to 6 GHz. Up to that point, companies

that have modeled tools for SAR measurement systems had typically created simulation

models whose predictions were only in the ballpark of the actual physical testing results,

then calibrated the models to make them more accurate. Models of this type are effective

when used to address a single design challenge, such as the dipole antenna, but are not

very accurate when they are used to predict the performance of a new product without

recalibrating the model. So APREL's scientists decided to create a series of rules for

constructing models that would generate accurate results that could be used along with

the physical quantity of calibration not only to design antennas but by customers to

define models and predict SAR measurements for their own products.

APREL scientists used the Chebychev polynomial matching method to increase the

operational bandwidth of the dipole design. They used numerical methods to create an

optimum wideband antenna used in the near-field of a phantom shell (APREL

Laboratories Universal Phantom) filled with biological tissue simulation fluid. This setup

matches the method used to determine peak and average SAR by the ALSAS 10

measurement system. The next step was simulating the initial design of the antenna and

the measurement system. The researchers picked XFDTD software from Remcom Inc.,

State College, Pennsylvania, because it provides the flexibility for modeling complex

structures with the high degree of fidelity needed to evaluate antenna performance in

near-field exposure conditions. This ability has been extended in XFDTD v6.2 with the

addition of an advanced meshing algorithm that makes meshing of certain difficult

geometry features possible. Adaptive meshing capabilities reduce solution times while

maintaining high levels of accuracy by automatically adjusting the mesh to provide more

cells in areas with high transients and reducing cells in areas where there is less variation.

As seen in the November 6, 2007 edition of the RF Globalnet (www.rfglobalnet.com) newsletter.

Case Study

Figure 2: A solid model of the broadband dipole and Universal phantom taken from XFDTD.

Developing The Electromagnetic Model

The major challenge faced by APREL scientists was creating the rules and guidelines for

constructing a model with the high accuracy needed to evaluate antenna performance in

near-field exposure conditions while keeping the model simple enough to keep solution

times to reasonable levels. Since SAR can be highly dependent on the surface current

distribution of the device, every effort was made to model the antenna and critical

radiation structures of the circuit with the optimized accuracy of the antenna matching

components. A particular challenge was defining the geometry and fine features of the

complex dipole antenna in order to optimize cell size and object orientation to keep errors

to a minimum. Another critical challenge was modeling antenna matching components

and other elements within the circuit that can potentially change the current distribution.

While the scientists recognized that computational times would be excessive if they

modeled the exact shape and size of all of the RF current contributing components, they

needed to correctly represent and account for the effects on the near-field distribution

produced by the overall device. By defining the rules and guidelines used to build the

model, a high degree of confidence can be achieved in the resultant data.

The radial center of the dipole is 10 mm from the tissue-equivalent liquid of the model.

The phantom shell is made from a low relative permittivity and conductivity material and

the interior of the phantom is filled with a tissue-equivalent liquid to a depth of 100 mm

with frequency-dependent dielectric properties for the frequencies 5.2 GHz and 5.8 GHz.

A general rule of thumb for electromagnetic simulation is that the cell size should be 1/10

of the wavelength or less at the highest frequency of interest. However, in this case the

dipole and electrical geometry was meshed with a cubic cells size of 0.1 and 0.3 mm,

much smaller than the 1/10 cells/wavelength limit, so that the geometric features would

be accurately modeled. Comparisons were made between both simulation results and a

As seen in the November 6, 2007 edition of the RF Globalnet (www.rfglobalnet.com) newsletter.

Case Study

decision was made based on resultant data to utilize a 0.3mm voxel size for the final run

of the simulation models. For the outer boundary of the model, Liao absorbing boundary

conditions were used, with 20 cells of separation from all geometry facets to this outer

boundary.

Figure 3: A solid cross sectional filed representation of the magnetic fields outside of the Universal

Phantom as emitted by the broadband dipole.

Ensuring Simulation Accuracy And Optimizing The Design

APREL scientists ran the simulation and compared the results to physical test results for

the same antenna design. Their goal was to match the impedance, return loss, and

standing wave ratio of the simulation to the physical test results. Despite the fact that the

geometry of the model closely matched the real antenna, the initial simulation results

varied substantially from physical testing. APREL scientists could have used the

traditional approach of simply adjusting the model so it would achieve the correct

electrical results, but this would have made its results valid for only this one antenna

condition. Instead they carefully looked at both the details of the mesh, geometry and the

boundary conditions to understand the reasons for the errors and construct a series of

rules that one would follow to achieve optimum results with a high confidence level.

They finally focused on the feed point of the antenna, which was difficult to model

because limitations on the initial models did not allow for the physical quantitative input

stage in order to keep the computational requirements at a manageable level.

APREL scientists worked several weeks to identify a location of the feed point that

matched the effects of the input stage. The scientists finally determined that the feed

point had to be at the midpoint of the tuning element to achieve simulation results that

matched the physical model. From this point the reference between the physical model

and simulation was fixed and the scientists worked to modify the antenna design to

achieve a goal of at least -20 dB return loss over the entire frequency range of interest.

As seen in the November 6, 2007 edition of the RF Globalnet (www.rfglobalnet.com) newsletter.

Case Study

Figure 4: A cross sectional mesh model of the electric fields produced by the broadband dipole within the

tissue contained inside the Universal Phantom.

They changed key dimensions of the antenna and re-ran the simulation repeatedly in

order to determine the effect on return loss. The simulation provided plots of Standing

Wave Ratio and return loss vs. frequency as well as other information that helped them

better understand the sensitivity of the circuit to various design parameters.

Figure 5: A cross sectional mesh model of the electric fields produced by the broadband dipole within the

tissue contained inside the Universal Phantom.

The scientists finally achieved their goals by reducing return loss as predicted by the

simulation to -26 dB at 5.2 GHz and -31 dB at 5.8 GHz. The next step was to export the

geometry of the antenna design that had achieved these results to a CAD system that

As seen in the November 6, 2007 edition of the RF Globalnet (www.rfglobalnet.com) newsletter.

Case Study

provided an environment where the new design could be built. When the prototype was

built and tested, the results matched the simulation within a few percent. The dipole

antenna designed with the aid of the simulation is now offered to users of the ALSAS 10

and other SAR measurement systems and allows them to validate the performance of the

measurement system over a wide frequency range. Further, APREL has released the

XFDTD-produced CAD model of the dipole and the Universal Phantom and is making it

available through Remcom to companies designing wireless products that would like to

be able to predict SAR measurement results before they proceed to build a prototype.

This model is of great interest to manufacturers designing WiMAX and Wi-Fi products

and has been used for a number of biological studies investigating the effects of wireless

communication systems on human tissue at the higher frequencies.

For more information, contact Remcom: 315 South Allen Street, Suite 222, State College,

PA 16801; Telephone: 1-814-861-1299, 1-888-773-6266 (1-888-7REMCOM) Toll-Free

in U.S. and Canada; Fax: 1-814-861-1308, 1-888-973-6266 (1-888-9REMCOM) Toll-

Free in U.S. and Canada; e-mail: info@remcom.com; Web site: www.remcom.com

As seen in the November 6, 2007 edition of the RF Globalnet (www.rfglobalnet.com) newsletter.