The conclusion of this two-part article will demonstrate how an improved waveguide-source technique can be used to validate SAR measurement systems from 5 to 6 GHz.
Specific-absorption-rate (SAR) measurements are critical for understanding the effects of nonionizing radiation, such as the emissions from cellular telephones and cellular base stations, on humans and other biological organisms. As noted last month, a modified waveguide-source technique can help to validate the test systems used for SAR measurements when used in the range from 5 to 6 GHz. Part 2 of this article will show how this technique can help minimize measurement inaccuracies due to leakage and probe positional errors.
When compared to the analytically-derived depth profiles, measured SAR profiles in a calibration waveguide are relatively higher when in close proximity to the surface. This effect arises with a flat-ended probe against a flat surface phantom when the absorbing liquid is, essentially, 'squeezed out' of the gap between the surface of the probe and the low-loss regions outside the liquid. It was shown in ref. 10 that a simple scheme is adequate to make corrections for this effect—at least when against a waveguide matching window, which is not the same as the exposure situation with a thin, shell-wall of lower relative permittivity.
The factors determined for the correction scheme used during probe calibration also need to be applied to the measured results. However, the justification for use of the same correction factors when the "wall" is a very different material is open to question and must introduce some uncertainty. In the measurements reported in this study, boundary corrections according to ref. 10 have been determined during waveguide probe calibration and used to correct the measured data.
Outputs from the model for the validation test geometry in Fig. 1 were obtained both in the form of predicted centerline decay profiles (Figs. 4 and 5) and in terms of maximum averaged SAR per watt of input power over a cube-shaped 1-g averaging volume against the phantom surface. The maximum of the averages over appropriate cubes were used.
Using an exponential curve fit for extrapolation, the average of the integral of the centerline scan from 0 to 10 mm was used to provide a check on the maximum possible value for the 1-g SARs. The ratio of this parameter to the calculated 1-g SAR values (Table 5) gives an indication of the lateral dispersion of the SAR field within the liquid and provides a factor that can be used to check the post-processed volume averages applied to measured results.
The scan extents and dimensions used for the measurements of 1-g SAR were of 10-mm range using 10 steps in all dimensions. For the lateral directions the scans were centered on the maximum region. For the depth scans, the starting point was a distance of 1-mm clearance between the probe tip and the phantom inner surface. The waveguide source was fed with a forward power of 0.25 W and the results have been normalized to 1 W in Table 6.
In addition to the three-dimensional (3D) scans (Fig. 6), centerline profiles were recorded starting from probe contact with the surface and extending for 25 mm in steps of 0.5 mm. Worst-case SAR values were deduced from the centerline scans following the procedure outlined above. The ratio of the 3D 1-g SAR values to those from centerline scans are shown in Table 5.
The procedures involved in SAR testing even at GSM frequencies involve significant uncertainties of the order of ±30 percent and such uncertainties have the potential to rise steeply at higher frequencies. Clearly, it is important to manage these uncertainties for testing in the 5-to-6-GHz band and a range of uncertainty-reduction measures need, ideally, to be applied. These would include, reducing the dimensions of the measurement probes, applying corrections for sensor offsets, managing the 3D scan measurement parameters and, most important, employing reliable system validation techniques.
In current practice at these frequencies, difficulties have been reported with the use of balanced dipoles,5 which have small physical dimensions and are hard to position with the required accuracy at specific spacings from a phantom liquid surface. While a waveguide source has been proposed for system validations,5 the use of the waveguide in "open-ended" mode and spaced some 8 to 10 mm from the reflective phantom wall and the absorbing liquid seems to invite additional uncertainty.
In this study, the waveguide approach is still preferred, but a matching window is used to improve the efficiency of forward power injection into the phantom liquid and the end of the waveguide is placed in contact with the phantom shell wall avoiding any potential errors in the spacing of the source from the phantom.
Using these procedures, measurements and computational assessments have been reconciled within 10 to 15 percent and it is expected that this range could be reduced by further procedural optimization—especially to the scan parameters used for the 3D measurements and to the subsequent post-processing used to determine the maximum volume-averaged SAR values.
In summary, a WR137 (WG13) waveguide source with a matching window was presented in contact with a rectangular phantom for SAR testing at 5 to 6 GHz, avoiding many uncertainties due to RF leakage and positional errors. The proposed test configuration has been modeled using FDTD and used to present "reference values" for the 1-g volume averages that should be obtained from validation measurements (Fig. 7). The validation configuration has also been set up for experimental measurement and a good correlation is found between the computational reference values and the values determined experimentally. The matched waveguide source provides improved performance for routine system validation procedures compared to previously proposed configurations utilizing either dipoles or unmatched waveguides spaced away from the absorbing material of the phantom.
Routine validations of practical SAR systems using probes of 5-mm diameter or less should achieve agreement with the reference values within limits of ±15 percent. This acceptance range may be capable of significant reduction when probe calibration procedures and the parameters used for 3D SAR scans are more tightly defined.
REFERENCES
