6.1.1 Definition and Measurement Method of SAR

Based on experimental studies on animals and humans, the external thermal load to create a 1°C temperature rise over about 15 minutes is about 4 W/kg or 4 mW/g. For partial body exposures, the presence of blood circulation can reduce the temperature increase. In particular, the limbs and head, which are the most likely to receive higher exposure, require around 10 W/kg to generate this rise. From a medical point of view, a 1°C temperature rise is not a severe event at all. This can easily be achieved by internal energy expenditure occasioned by gentle jogging.

Based on the definition of the biodosimetry, the SAR is the rate of absorption or dissipation of energy (W) in unit mass (M) as shown in Equation 6.1 [2]:

(6.1)

Here, ρ is the specific density.

In the world of cellular phones, the SAR is a measure of the rate at which energy is absorbed by the body when exposed to an RF EM field. It can be calculated from the E field within the tissue as [3] follows:

(6.2)

Here, σ is the tissue’s electrical conductivity, ρ is the tissue’s density, and E is the root mean square (RMS) electric field. The commonly used unit of SAR is either W/kg or mW/g, which actually are identical. To smooth out the fluctuation in SAR measurement, SAR is usually averaged over a small sample volume (typically 1 or 10 g of tissue).

In practice, it would be too gross to use real tissues in SAR measurements, so various tissue‐simulating liquids are introduced. The dielectric properties of different human tissues can vary over quite a wide range. To simplify the test procedure, only two human parts, head and body, are considered in SAR measurements. Shown in Tables 6.1 and 6.2 are target dielectric properties for head and body [4], respectively. The values listed here are target values according to IEEE 1528 standard. More than one standard is currently used, and the target values in different standards are also slightly different.

It is obvious that the dielectric properties of either head or body are variables of the working frequency. Ideally, people should find one liquid to simulate the head in all frequencies and another one for the body. However, at present, there is no single head‐ or body‐simulating fluid which can cover all frequencies with an acceptable accuracy and consistency. Different fluids must be used when carrying out SAR measurements in different frequency bands. Shown in Table 6.3 is a recipe for simulating liquid at different frequencies.

As the dielectric properties of head or body are close enough at adjacent frequency bands and a ±5% tolerance from the target value is allowed according to most standards, a single simulating liquid might be used when measuring adjacent bands, such as 850 and 900 MHz bands or 1800 and 1900 MHz bands.

As the dielectric properties are variables of temperature, the ambient temperature and also the tissue‐simulating liquid temperature need to be well controlled within a certain range.

Besides tissue‐simulating liquids, a container is required to hold the liquids. Shown in Figure 6.1 is a phantom produced by Schmid and Partner Engineering AG (SPEAG). The shell corresponds to the specifications of the specific anthropomorphic mannequin (SAM) phantom defined in IEEE 1528‐2003, CENELEC 50361, and International Electrotechnical Commission (IEC) 62209. It enables the dosimetric evaluation of left and right hand phone usage as well as body‐mounted usage at the flat phantom region. The whole container is installed on a wooden frame. The phone holder is made of low loss plastic.

A stand‐alone phantom head, such as the one shown in Figure 6.2, can also be used as the container. The head has the same size and curvature as the one shown in Figure 6.1. A head phantom, which meets the specifications defined by the standards, is often referred to as the SAM’s head. The SAR of both left and right hand phone usage can be measured on a SAM’s head. The mechanical parts on both sides of the SAM’s head are phone holders.

According to the SAR’s definition shown in Equation 6.2, only the E field is required when measuring a device. An E‐field probe moved by a robot, as shown in Figure 6.3, is the heart of any SAR measurement setup. The robot is normally a five‐axis one, which means it can move a probe to different locations in a three‐dimensional (3D) space and also point the probe toward different angles. Comparing with the weak signal a probe detects, a robot is a device that drains enormous power. To mitigate the potential EMI problem, testing signals are amplified and converted into a digital signal in a shielded box, then transmitted to the processing computer through an optical fiber. The data acquisition circuit is powered by batteries to eliminate any galvanic connection toward the external environment.

As E field is a vector field, it is composed of three orthogonal components as shown in Equation 6.3. Inside an E‐field probe, there are actually three individual E‐field sensors. Each sensor measures one orthogonal E‐field component.

(6.3)

Shown in Figure 6.4 are two commonly used configurations: Δ and I [5]. Three short dipoles are used to detect those orthogonal components. In both configurations, the dipoles are arranged in a manner which ensures that they are perpendicular to each other.

It might be thought that the short dipoles are connected to the amplifier circuit through some normal transmission lines, such as a coaxial line or a microstrip line. In fact, they could not be arranged so. Any transmission line which can transmit an RF signal at 900 MHz or 1.8 GHz is composed of long metal traces. Those metal traces will disturb the field distribution around the device under test, thus invalidating the result. It seems that we are facing a dilemma. How can we measure something without putting a probe next to it?

Shown in Figure 6.5 is the solution which is adopted by most commercial SAR systems. Of all the parts of a sensor, only the short dipole is made of metal. As the dimension of the short dipole is much less than one wavelength of even the highest working frequency, the disturbance from the dipole can be ignored. The parallel transmission line connected to the dipole is made of high resistive material and is not able to sustain a current. The transmission line is transparent from the RF point of view. The trick is done by the Schottky diode. A diode can function as a rectifier which converts an RF signal into a baseband envelope signal. The voltage value of the baseband envelope signal is related to the power level of the RF signal. However, the relation between the voltage value and the power level is not linear; the relation can be approximately expressed by the square law. Some means of calibration is required to establish the correlation between the two. The high resistive transmission line cannot carry a current, but it can transmit a baseband voltage signal as long as the input impedance of the signal amplifier is much higher than the total resistance of the transmission line. At the frequencies of cellular communication bands, the shunt capacitance required in a diode rectifier’s circuit can be implemented either through the parasitic capacitance of the diode itself or the distributed capacitance of the layout.

Now, it is obvious that the probe coverts RF power into baseband envelope voltage on the spot and transmits the voltage through an RF transparent transmission line. By doing this, we are successfully putting a probe next to a phone without allowing the phone to “notice” it. The conversion method is sufficient for SAR measurements; however, it might not suit other kinds of near‐field scanning, because the phase information is discarded during the conversion.

When doing a SAR test, a phone should be in the normal working condition. There should not be any cable attached to the phone. The phone must only have all the features which will appear on the mass production phones. Similar to an over‐the‐air (OTA) test, a base station simulator is used to initiate and maintain a call with the phone under test. The phone is configured to transmit at its maximum power.

The goal of SAR measurement is to find the maximum value, which is averaged over either 1 or 10 g volume depending on different standards, inside the phantom head. The tissue, either a real one or a simulated one, is a lossy material. The deeper an RF signal penetrates, the more it is attenuated. As a matter of fact, the highest SAR always appears on the inner surface of the phantom. To expedite the test procedure, the whole test is divided into two steps, an area scan, and a zoom scan, as illustrated in Figure 6.6. The first step is the area scan. A two‐dimensional (2D) area is scanned by following the inner surface of a phantom. The result is a SAR distribution on a curvature surface. Using the maximum point obtained from the first step as the center location, the second step scans a localized 3D cubic.

For the purpose of protection and sealing sensors from a tissue‐simulating liquid, a plastic cover is used as a probe’s outer shell. Thus, there is always a gap between the sensor and the inner surface of a phantom. The SAR value on the surface must be extrapolated from the measured 3D data by postprocessing software.

Shown in Figure 6.7 is an illustration of 1 and 10 g average. A body phantom is used in the measurement. The zoom scanning volume is marked by dashed lines. There are two cubes inside the scanning volume. The larger one is the 10 g average cube. The location of the 10 g cube represents where the maximum averaged SAR over 10 g is. The smaller one is the 1 g average cube. Similarly, its location corresponds to where the peak averaged SAR over 1 g is. The density of tissue‐simulating liquid is approximately 1000 kg/m³, thus the volume of 1 g liquid is approximately 1 cm³.

Sometimes, there might be two regional maximums emerging from the area scan. If the difference between two peaks is less than 2 dB, two zoom scans which are centered on the two maximums, respectively, are required.

A thorough SAR test includes measurements at both sides of the head and the body. On each side of the head, there are two positions, the touch position and the 15° tilt position. There are also two positions, liquid crystal display (LCD) up and LCD down, for the body. At each position, the SAR measurement must be carried out at multiple bands. At each band, there are low, middle, and high channels. For some phones, there are multiple configurations by the phone itself, such as whip up and down or slide open and close.

It is obvious that an enormous number of measurements are required if all the combinations must be measured. Based on the current IEC 62209‐1 standard [6], only the middle channel of all combinations is mandatory. For the configuration which generates the highest SAR, a full set of measurements at low, middle, and high channels are required. It is recommended that low, middle, and high channels are also measured for all combinations whose SAR reading is within 3 dB of the device’s highest SAR value.

6.1.2 SAR Limits in the United States and Europe

There are quite a few regulations which are related to SAR limits. For example, in the United States, the FCC Rules Part 22H regulates the 824–849 MHz bands, Rule 24E regulates the 1850–1910 bands, and Rule 15.247 regulates the 2400–2483.5 MHz bands. As a design engineer, it is not necessary to remember all those regulations. All the regulations in the United States and Europe can be boiled down to the following two values:

1. In the United States, less than 1.6 mW/g or 1.6 W/kg, averaged over 1 g
2. In Europe, less than 2.0 mW/g or 2.0 W/kg, averaged over 10 g

If we recall that 10 mW/g is required to generate 1°C temperature increase inside a brain or limbs, both limits are well within the safety zone. Judging by the limit value itself, the US limit is only 20% tighter than the European one. However, the European standard calculates the SAR value by averaging over a volume of 10 g instead the 1 g used in the US standard. The larger the average volume is, the lower the SAR value is. Thus, the US limit is significantly tighter than its European counterpart.

To give a more intuitive feeling about those two limits, SAR values, averaged over both 1 and 10 g, of eight phones are listed in Table 6.4. The data are collected from FCC reports [7]. The SAR values averaged over 10 g are collected from the same 3D scanning which generated the maximum SAR values averaged over 1 g. The fourth column is the comparison between the two limits. Using the Apple iPhone 3Gs as an example, the American limit is 1.88 times stricter than the European one. Overall, the ratio is around 2 between these two limits.

Phone model	SAR value (mW/g)
	Averaged over 1 g	Averaged over 10 g
Apple iPhone 3Gs	1.19	0.79	1.88
HTC diamond	1.40	0.78	2.24
LG CU920 Vu	1.26	0.76	2.07
Motorola Z9	0.94	0.56	2.10
Nokia N95	0.64	0.37	2.16
Nokia N97	1.01	0.74	1.71
Palm Pre	0.94	0.53	2.22
Samsung Omnia i910	1.25	0.89	1.76

A comparison of some phones’ SAR values between 1 and 10 g can be found on the Internet. Those values might be different from values listed in Table 6.4, because they are collected from certification reports filed in different countries. Let’s use a 2G phone as an example. An international 2G phone might support 850, 900, 1800, and 1900 MHz bands. However, different countries support different portions of the overall bands. In the United States, when filing an FCC application, only the averaged SAR over 1 g at 850 and 1900 MHz are required. Similarly, in Europe, only 900 and 1900 MHz are measured for SAR over 10 g. As a phone’s radiating power can be individually adjusted band by band, a comparison between SAR values over bands is not that meaningful.

For devices which have several coworking transmitters, such as Global System for Mobile Communication (GSM) and Bluetooth, or GSM and wireless local area network (WLAN), the SAR value can be obtained through either a mathematical summation of individual SARs or a measurement with both modules transmitting simultaneously. This detailed information is omitted in the book. Interested readers can refer to related articles [8].

When working on SAR‐related issues, one very useful resource is the FCC website. You can find the SAR reports of all FCC‐approved phones under the “Equipment Authorization Search” page [7]. As required by the FCC filing process, detailed photos of a phone’s internal circuit are also included in the report. From those photos, it might be possible to tell whether some special SAR reduction techniques have been used in certain phones.

6.1.3 Controlling SAR

The SAR might be the most dangerous time bomb for any antenna engineer. However, the SAR measurement itself needs to carry some of the blame. As the SAR test equipment is a quite complex system, the expanded uncertainty from the system itself is more than 20%. This means a phone that has a nice margin of 20% when it is measured in an internal R&D facility might fail when it is measured in an external certificated testing house. To mitigate that risk, most companies adopt an internal SAR limit which is lower than the official one. The internal limits are different from company to company. How much lower the internal limit is depends on a company’s preference and judgment on risks. Of course, a lower internal benchmark only makes it more of a challenge to meet the limits as an antenna engineer.

Besides the measurement uncertainty, the SAR is also sensitive to a phone’s mechanical parts, especially the metal ones. Some of the last‐minute changes from other disciplines, such as adding extra grounding tabs for electrostatic discharge protection, changing the plating technique of the front bezel to improve the production yield, and so on, can all have an impact on the SAR. When those emergencies happen, it might be quite late in a phone’s development cycle, which means more hard work is needed to solve the SAR problem in time.

Before discussing how to control SAR, we need to understand how a phone’s near‐field energy is distributed. As the whip antenna is disappearing from the phone market, the characteristic of a whip’s SAR is not covered in the book. Only internal antennas and short external stubby antennas are discussed. As shown in Figure 6.8, the SAR distribution on a phone mostly depends on the working frequency bands. At the low bands, such as 850 and 900 MHz bands, the metal structure of a phone is the main radiator. The maximum E field on a phone normally appears in the middle of the phone. When a phone instead of the antenna on the phone is the main radiator, an antenna’s form factor, whether it is a monopole or a planar inverted‐F antenna (PIFA), has less impact on the SAR value.

At the high bands, such as 1800 MHz, 1900 MHz and up, the peak near‐field energy appears in locations which are closer to the antenna. At those bands, an antenna’s form factor plays an important role. A normal PIFA has a directional radiating pattern, which means the ground underneath a PIFA can divert the energy away from a user’s head, thus decreasing the field strength on the opposite side of the antenna.

When a phone is measured on a phantom body, the distance between its surface and the phantom’s surface is a relatively consistent value, so the SAR distribution normally agrees pretty well with the near‐field distribution of the phone. For antenna engineers, the body position is not a big concern, as there is no specification for how much the gap between a phone and a phantom should be. We can always decrease a phone’s SAR value at the body position by increasing the gap. For phones with low SAR value, the gap normally chosen is 15 mm.

Unlike a phantom body, a phantom head is a totally different story. According to the standard [3], four positions must be measured on a phantom head. Shown in Figure 6.9 are illustrations of the left ear check position. It is also known as the touch position. Shown in Figure 6.10 are illustrations of the left ear tilt position. Repeating the two aforementioned measurements on the right ear, we have a total of four combinations.

As shown in Figure 6.9, when a phone is correctly placed on a touch position, the phone’s vertical center line should fall into the plane defined by RE (right ear), M (mouth), and LE (left ear) points. The center of the phone’s earpiece should touch the LE point while another point on the front side is also in contact with the cheek of the phantom.

As shown in Figure 6.10, while maintaining the contact between the phone and the phantom, then pivoting against the ear, moving it outward away from the mouth by an angle of 15°, what we get is the tilt position. In both touch and tilt positions, the ear piece of a phone always touches the phantom ear.

It is obvious that the distance between a phone and the surface of a phantom head is not a consistent value. The distance is one of the two most important factors to control SAR. The further the distance, the lower the SAR value. The other most important factor is the near‐field energy distribution. The actual SAR inside a phantom head is a combining effect of both factors.

To control the SAR is to play with those two factors. An antenna engineer should actively participate from the concept stage, which is the first stage when designing a phone. At the beginning stage of a project, most efforts should be focused on securing a correct location and a decent volume for antenna.

As per the requirement of SAR tests, the vertical center line of a phone is in the same plane defined by the LE, M, and RE points. We can extract the contours of a phantom and a phone in the LE–M–RE plane, as shown in Figure 6.11, to study their relative distance.

If an antenna is installed in the top position, there is little gap between the phone and the head in both touch and tilt positions. Whip antenna, stubby antenna, and PIFA can all be used in the top position. From the SAR point of view, the main difference between a PIFA and an internal monopole antenna is the existence of the ground beneath the radiating element. The ground in a PIFA can divert the energy away from the head; however, it can also shrink the antenna’s bandwidth. If you want to put an internal antenna on the top, you have to use a PIFA. If you want to have a smaller antenna, the internal monopole antenna can be used and must be located at the bottom.

If you ever look at those phones with a stubby antenna, you might note that stubby antennas are normally off‐center and are closer to the backside of phones, away from the head. This is a measure to increase the distance, thus decreasing the SAR. Another frequently used technique is tilting the stubby antenna, as shown in Figure 6.12. Tilting the stubby antenna has more effect at the high band. At the lower band, the peak of near‐field energy is located in the middle portion of the phone, which is also the contact point between a phone and a phantom head. Thus, the peak SAR is most likely to appear around the touch point on the phantom cheek. Tilting the antenna has little effect on the energy distribution on the board at the lower band, thus has little impact on the specific SAR.

Increasing a phone’s thickness at the LCD display area is another commonly used technique. The solid line and the dashed line contours, as shown in Figure 6.13, illustrate phones with and without localized thickness increasing, respectively. It is obvious that the phone has effectively moved away from the phantom head.

As shown in Figure 6.14, by extruding the navigation key or adding a bumper at the center, a phone can also be effectively shifted away from the phantom head. For a bottom‐installed antenna, the lever effect can magnify the small extrusion at the middle to a large separation at the bottom, thus decreasing the SAR more significantly. An extruded feature at the center is not only a useful SAR reduction measure but also an important feature from the mechanical point of view. Apart from glass screens used in some high‐tier phones, most phones have LCD screens which are made of plastic and can easily be scratched. The extruded feature can keep the LCD screen away from the rough surface when the phone is placed faceside down.

For a clam shell phone, the flip angle is a critical parameter which decides the SAR. As shown in Figure 6.15, the larger the flip angle, the further the lower part of the phone away from the phantom head. If a phone has a bottom‐installed antenna, the relation between the flip angle and the SAR is monotonous. The large angle is always better.

When a clam shell phone has a stubby antenna, the relation between the flip angle and the SAR value is a little more complex. At the low band, the circuit board inside the phone is the main radiator. So when the flip angle is small enough, as shown in Figure 6.16a, the bottom part of the phone is too close to the head and the SAR value starts to increase. On the other hand, if the flip angle is too large, as shown in Figure 6.16b, the antenna itself becomes too close to the head, which also increases the SAR value. It is obvious that there should be an optimal flip angle in between which can balance the SAR contributions from both the phone body and the antenna radiator.

From some perspectives, slide phones are kind of similar to flip phones, as both of them have moving halves. Some slide phones have top‐installed PIFAs, and the SAR design consideration of those slide phones is pretty much the same as a single‐piece candy‐bar phone. For some slide phones, the antenna is at the bottom. When an antenna is placed at the bottom, the design team expects the antenna solution to be a space‐saving monopole antenna. The only issue here is that a slide phone can make a phone call in both close and open positions. When doing a SAR test, both slide open and close positions must be measured. As shown in Figure 6.17, in the slide close position, a bottom‐installed antenna is too close to the head. Unlike a PIFA, the monopole solution does not have a ground to divert the energy away from the head. A frequently used method to decrease SAR in such circumstances is mismatch. When a phone can be used in multiple configurations, most service providers only have a stringent requirement at the primary user configuration, which is the slide open position in this case, but a relative loose specification for other configurations. The antenna can be optimized for the slide open position and purposely mismatched in the slide close position, thus reducing the power radiated toward the head.

So far, all SAR reduction methods we have discussed focus on how to increase the distance between a phone and a phantom head. The other SAR impact factor, which is as important as the distance, is the near‐field energy distribution. To some degree, by choosing an antenna solution, say, stubby, PIFA, or internal monopole, the near‐field energy distribution is also selected. It is obvious that the near‐field distribution of a PIFA is better than an internal monopole, and a bottom‐installed monopole antenna is better than a top‐installed one. However, there are other ways to influence the near‐field distribution than merely selecting antenna type.

Figure 6.18 is a demonstration of a SAR reduction technique. Figure 6.18a shows the original device. The antenna used here is an IFA, and the length of the antenna’s horizontal arm is 35 mm. In Figure 6.18b, an L‐shaped SAR reduction feature is implemented, which is located on the opposite side of the ground plane to the antenna. The total length of the L strip is 35 mm, its width is 2 mm, and the distance between the strip and the ground is 2 mm.

The resonant frequency of the IFA is 1.8 GHz. As the dimensions of the IFA and the SAR reduction L strip are similar, it is safe to say the resonant frequency of the L strip is also around 1.8 GHz. Shown in Figure 6.19 are the reflection coefficients of the IFA with or without the SAR reduction feature. It is clear that the variation introduced by the SAR reduction feature is negligible.

Figure 6.20 shows the simulated SAR distribution plots. A flat body phantom is used in the simulation. The simulation liquid is 1800 MHz body fluid. The gap between the ground plane and the phantom is 10 mm. The output power is 0.15 W. The dashed lines in Figure 6.20a and b are the contour lines in which SAR equals 1.6 W/kg. By introducing the SAR reduction feature, three effects are realized: (1) the peak SAR is decreased, (2) the area of SAR greater than 1.6 W/kg is shrunk, and (3) the location of the peak SAR is moved. The reason for SAR reduction in the first two is quite obvious. The third one, altering SAR distribution, is also an important technique in decreasing SAR. As has been discussed in the earlier part of the section, when a phone is put next to a phantom head, the distance between the printed circuit board (PCB) and the head is not a constant. SAR reduction can be achieved even without suppressing the peak energy level on the PCB. By simply shifting the hot spot toward the less critical area on the PCB, where the distance between the PCB and the phantom head is greater, the SAR value can be decreased.

Some other guidelines related to the SAR reduction are listed as follows:

• As the SAR is a specification only with regard to the highest value which is averaged in a small cubic, the SAR can be decreased by spreading the energy more evenly over the PCB. By doing this, the total radiation toward the phantom head is pretty much the same, however, the SAR value will be lower.
• The location of a feeding point also plays a role on the SAR value. As a qualitative trend, the corner of a PCB is the best location from the point of view of antenna efficiency and bandwidth; however, it also generates a higher SAR value.
• Decreasing the conductive power of the RF transmitter is always the last resort. Nobody likes it, as it deteriorates a phone’s performance and makes the R&D management team looks bad. Most of the time, their superior will only look at the numbers. For them, a 30 dBm power level is always much better than a 29.8 dBm one, although there is only 0.2 dB difference and nobody can tell the difference in the real world. As the SAR values at low, middle, and high channels are normally different, if it is possible, only adjust the power level at the channel which has SAR issues. By doing this, the perception of the phone specification value might look acceptable.

In some phones, there may not be enough space for any extra SAR reduction feature. The SAR reduction can be realized by utilizing existing metal structures. By selecting grounding locations of the metal frame of the LCD panel, the metal phone bezel, and other metal objects on the phone, a certain level of SAR reduction can still be achieved.

The SAR measurement is quite a time‐consuming process. Even a single channel measurement, which includes an area scan and a fine scan, will take nearly 10 minutes. One way to save time is by using single‐point measurement. You should be able to move the test probe to the peak SAR location and measure the instant SAR value at the location. Depending on the test equipment and the software version, the actual way to invoke the single‐point measurement varies. If you have difficulty locating that function, refer to the equipment manufacturer. When you try out different SAR reduction configurations, you actually only need to check out the single‐point instant SAR value. If that value does not decrease, it means the current configuration does not work. A single‐point measurement takes at most 1 minute, so it can give you more time to try out different designs.

When you get some configurations which generate promising value at single‐point measurements, you need to carry out a full SAR measurement. It is possible that only the location of the peak SAR has shifted and the SAR value has not been decreased. Then you must keep working. If a SAR reduction is really observed, you need to recheck the antenna’s performance to make sure it still works well. Sometimes, the SAR reduction is due to the antenna’s performance degradation and that is still not a working solution. After you are sure everything works correctly, you must duplicate the SAR solution on several phone samples and check the consistency of the solution. As an engineer, you should never broadcast the great news until you have done all the tests.

Always bear in mind that SAR reduction techniques can only help to some extent. If the SAR value is 20% above the specification, it is appropriate to seek help from SAR reduction techniques. If the SAR value is 100% higher than the specification, the design is totally flawed and it might have to be redesigned from scratch.

6.1.4 Updates on SAR requirement

Since the first edition of the book, cellular technology has progressed a lot. The 3G network reached its prime and then handed the torch to the 4G network. Now 4G has also matured and even the most sub‐100 USD phones are 4G ready. People are working toward 5G system. As most 4G phones are back‐compatible, the SAR qualification procedure has become a very long laundry list. In the United States, the current standards applied to SAR measurement are FCC 47 CFR § 2.1093, IEEE STD 1528‐2003, TCB workshop notes GPRS testing considerations, and FCC‐published RF exposure KDB procedures listed later. One thing for certain is, with the progress of cellular technology, this list will be longer and longer.

• 447498 D01 General RF Exposure Guidance
• 648474 D04 Handset SAR
• 941225 D01 SAR test for 3G devices
• 941225 D02 HSPA and 1× Advanced
• 941225 D03 SAR Test Reduction GSM GPRS EDGE
• 941225 D04 SAR for GSM E GPRS Dual Xfer Mode
• 941225 D05 SAR for LTE Devices
• 941225 D05A LTE Rel.10 KDB Inquiry Sheet
• 941225 D06 Hotspot Mode SAR
• 248227 D01 SAR Meas for 802 11abg
• 865664 D01 SAR Measurement 100 MHz–6 GHz
• 865664 D02 SAR Reporting

When doing an official FCC qualification test, there are many detailed considerations. The best way to grab those details is from published SAR reports of FCC‐approved phones [7]. The contents in the book are only a brief introduction. For each phone model, required SAR measurements of cellular bands are combinations of the following configurations.

• Multiple 2G, 3G, and 4G standards: For 2G, there are GSM and CDMA. For 3G, there are code division multiple access (CDMA) 2000, WCDMA, and TD‐SCDMA. For 4G, there are TD‐LTE and LTE FDD.
• For each standard, there are also multiple frequency bands.
• For each frequency band, there might be different substandards, such as voice and data.
• For each communication setting, there are several holding positions. Beside the head positions and body‐worn positions discussed in Section 6.1.3, hotspot positions are added and will be discussed later.

Beside cellular bands, SAR of Wi‐Fi bands (2.4 and/or 5 GHz) are also required. For those phones with more than one Wi‐Fi antennas, SAR of each Wi‐Fi antenna needs to be measured separately.

The newly added hotspot positions include six positions and are all measured by using a body phantom. There are six surfaces, which are front, back, top, bottom, left, and right surfaces, on a phone. Each hotspot position is formed by placing one phone surface next to a body phantom.

Shown in Figure 6.21 are four out of six positions. Figure 6.21a are left or right hotspot positions and Figure 6.21b are hotspot top or bottom positions. The separation distance between a phone’s surfaces and scanning grids, marked as G in Figure 6.21, is set by the FCC procedure “941225 D06 Hotspot Mode SAR.” For big devices (>9 cm × 5 cm), the gap is 10 cm. The gap is 5 cm for smaller devices.

The front and back hotspot positions, which are omitted in Figure 6.21, are similar to body‐worn positions. The only difference is the fixed separation distance G, which can be freely chosen when measuring body SAR. In most cases, the G of hotspot positions is considerably smaller than body positions.

When measuring body positions, G can be adjusted to meet SAR requirement. In the hotspot positions, the freedom of choosing G has gone and in return we get the freedom of adjusting the output RF power. Because a phone knows when it is in the hotspot mode, it indeed can dial down its output power based on its preset power table.

Once upon a time, a phone can only do one thing at a time and never turns on more than one transmitter. Nowadays, phones are just like human and capable multitasking. Today’s phone might simultaneously turn on up to four transmitters. For example, someone is sharing 4G data through Wi‐Fi in the hotspot mode, a 2G voice call comes in, and he/she picks the call through a Bluetooth headset.

For simultaneously transmitters, one SAR evaluation method is adding SAR values of all simultaneously working transmitters. If the sum value can meet the SAR requirement, no more SAR evaluation is necessary. In fact, the simultaneous transmission of SAR is not really a challenge. Because a phone always knows how many transmitters are simultaneously working, it can legally dial down its output power accordingly. The hotspot mode is one kind of simultaneous transmission cases. It needs to turn on at least a Wi‐Fi transmitter for routing and a cellular band transmitter for data connection. For some phones, which have the Wi‐Fi multiple input and multiple output (MIMO) capability, Wi‐Fi mode itself can also be a simultaneous transmission scenario.

6.2.1 HAC Measurement

When measuring SAR, it is actually gauging the heating effect due to the conductive loss. The conductive losses are directly related to the E field, so only the electric field needs to be measured. When talking about HAC, as both the E field and the H field can interfere with a hearing aid, both fields must be measured.

The HAC measurement capability is generally provided as an extension of a SAR system. For antenna engineers, an HAC measurement relates to two probes: the E‐field probe and the H‐field probe. It should be no surprise that the E‐field probe used in HAC is identical to the one used in SAR. The H‐field probe is used for HAC measurement only. Similar to an E field, an H field is also a vector field and it is composed of three orthogonal components as shown in Equation 6.4:

(6.4)

Inside an H‐field probe, there are actually three individual H‐field sensors. Each sensor measures one orthogonal H‐field component. As shown in Figure 6.22, there are three metal wire rings inside an H‐field probe. The axes of rings are perpendicular to one another. Each ring functions as a magnetic dipole and can detect the H‐field along its axis. Of course, Schottky diodes are needed as rectifiers to convert the RF signal to the voltage signal; high impedance lines are also needed to transmit the voltage signal back. Both of them are omitted in Figure 6.22. The E‐field probe and the H‐field probe provided by the same company normally use the same type of connectors. Thus, an H‐field probe can be connected to the data acquisition unit used in SAR measurements.

Shown in Figure 6.23 are E‐field and H‐field probes. Although an H‐field probe will never be inserted into tissue liquid during an HAC measurement, it still uses the same sealing package used by the E‐field probe. It is quite easy to distinguish an H‐field probe from an E‐field probe if they are made by the same manufacturer. The dimensions of an H‐field probe are normally larger than an E‐field probe. One might think it is because a structure made of metal rings is more difficult to miniaturize. If it were the real reason, why couldn’t we make the E‐field probe bigger? This way the packages of both probes can be standardized and the logistical burden can be eased a little bit.

The correct answer can be found in most EM textbooks [10]. The radiation resistance R of a Hertzian (electric) dipole is given by Equation 6.5. The radiation resistance is a quantitative property which tells how well a device is capable of radiating or receiving signals. The higher the value, the better the device.

(6.5)

Here, dl is the length of the electric dipole and λ is the wavelength. The radiation resistance R of a magnetic dipole is given by Equation 6.6:

(6.6)

Here, S is the area of the magnetic dipole and D is the diameter of the ring.

Both Equations 6.5 and 6.6 are valid approximations only when dipoles are electrically small, which means or . An electric dipole’s radiation 2 resistance is proportional to and a magnetic dipole’s is proportional to . The radiation resistance of an electric dipole is always higher than a magnetic dipole with the same external dimensions. For example, if the total length of an electric dipole is , its radiation resistance is 1.97 Ω. The radiation resistance of a magnetic dipole with a diameter is 0.12 Ω. As both the E‐field probe and the H‐field probe share the same data acquired unit, it is preferable to minimize the difference between their output voltages. Therefore, the H‐field probe is always bigger.

Shown in Figure 6.24 is the required HAC test grid [9, 11]. The grid is 50.0 mm by 50.0 mm area that is divided into nine evenly sized blocks or subgrids. The grid is centered on the audio frequency output speaker of the phone. The measurement plane is 10.0 mm above the phone’s top surface. Unlike a SAR test, which requires a 3D surface scanning, an HAC test only needs a simple 2D scanning. In a real HAC system, measurements are first taken on much finer grids, say, at 2 or 5 mm increments. The final 3 × 3 grid is then obtained through the average of the raw measurement data.

6.2.2 HAC Specification in the United States

The current HAC specification in the United States is according to ANSI C63.19‐2007. The limit value of ANSI 63.19‐2007 is the same as ANSI 63.19‐2006. The version used before the 2006 one was ANSI C63.19‐2001. Shown in Table 6.5 are the specifications of ANSI C63.19‐2007. The top half of the table is the specification for devices operating in bands below 960 MHz, and the bottom half is for devices operating above 960 MHz. In each frequency range, there are four categories. An M4 grade device has the lowest RF emission, so it is the best from the RF emission point of view. The articulation weighting factors (AWFs) used for the standard transmission protocols is shown in Table 6.6. The AWF is set as 0 dB for almost all protocols except the GSM, whose AWF is 5 dB. Actually, the GSM protocol is also a TDMA system, just like EIA‐136 and iDEN; however, its repeating frequency is 217 Hz and its fundamental and harmonical components fall within the audible range of human beings (20 Hz–20 kHz). To take this into consideration, a 5 dB penalty was enforced on GSM. The 5 dB is calculated by 20·log₁₀ (V/m or A/m), so when calculated in V/m or A/m, the GSM’s limit equals 56% of other standards.

For antenna engineers, the M3 category is what they are aiming for. When designing a GSM phone, the limits for E‐field emission are 266.1 and 84.1 V/m for bands below and above 960 MHz, respectively. The limits for H‐field emission are 0.8 and 0.25 A/m. A phone must pass all four limits to get an M3 rating, then it can be labeled as HAC approved. In reality, nobody really works toward an M4 grade. If a phone happens to pass the M4 limit, engineers will be happy with the result.

For CDMA or UMTS phones, the E‐field limit is 354.8 and 112.2 V/m, respectively. The H‐field limit is 1.07 and 0.34 A/m, respectively. The discrimination between bands below and above 960 MHz is a relatively new thing. Before the implementation of ANSI 63.19‐2006, the limits for all frequency bands were the same, which is identical to what the high band value is today. During that period, hardly any dual‐band candy‐bar phone, except a few bulky models, could get M3 approval. Phone manufacturers depended on flip phones, slide phones, and so on, to comply with the FCC regulations. For most phones which could not get the HAC approval at that time, the low band is the problem. After many antenna engineers had struggled for thousands of hours, one fine day the HAC Committee made the famous decision: as the technology of the hearing aid has evolved significantly and almost nobody can pass the low band limit, let’s increase the limit by 10 dB. The 10 dB increase equals to a 316% increase in V/m or A/m. This is the unofficial birth story of ANSI 63.19‐2006. The official version of the story is that some studies showed that most contemporary hearing aids have more immunity in bands below 960 MHz than above. Then combined with the consideration of different transmitting power levels at different bands, the standard was revised to have the band‐dependent limits. I still wonder why, if the lower band limit can be raised by 10 dB, why don’t they lift the high band limit also by 10 dB? Then all antenna engineers can stop worrying about HAC altogether and live happily ever after.

As shown in Figure 6.25, the measurement grid defined in C63.19 consists of nine evenly sized blocks, which are used to define permissible exclusion areas. For both E‐field and H‐field measurements, three contiguous blocks may be excluded from the measurements except for the center block that may never be excluded. Both exclusion examples shown in Figure 6.25 are permissible. There must be four blocks left that are common to both E‐field and H‐field measurements, so a maximum of five different blocks can be excluded (e.g., three blocks excluded from the E‐field and two blocks from the H‐field).

Shown in Figure 6.26 is the measurement result of a sample phone. In the E‐field plot, three highest continuous blocks at the bottom‐left corner are excluded. The exclusion does not help too much in this case, as the center block, which can never be excluded, has the strongest field distribution among all blocks. In the H‐field plot, three blocks along the bottom edge are excluded. As the bottom‐left block has the highest value, the exclusion of that block is actually helping the overall rating.

Most techniques used in SAR reduction can also be used in HAC reduction. However, the HAC is even trickier to work with. As the distribution of the E field and the H field are not correlated, if one field is suppressed, that might lead the other field to rise. Based on the rule of exclusion, shifting the peak away from the center block always helps.

So far, we have focused on the RF emission part of ANSI C63.19‐2007. This is the part which is directly related to antenna engineers. However, the scope of ANSI C63.19‐2007 is much wider than that. As an engineer, it is always good to know the whole picture. The standard was developed in cooperation with three main parties: representatives of organizations representing people with hearing loss, hearing aid manufacturers, and the digital wireless telephone industry. It regulates both hearing aids and cell phones. The overall frame of the standard can be simplified as follows:

• Wireless device
  • RF emission/category test (directly antenna related)
  • T‐coil mode/category test.
• Hearing aid
  • RF immunity/category test
  • T‐coil immunity/category test.

Corresponding to the RF emission test for cellular phones, there is an RF immunity test for hearing aids. The RF immunity test evaluates how well a hearing aid can sustain RF emission. The ANSI C63.19‐2007 limits for hearing aid are shown in Table 6.7. The hearing aid immunity is measured by using continuous wave (CW), and it is not frequency dependent. Similar to RF emission categories, the M4 grade is also the best one, which means a hearing aid can sustain the strongest RF emission.

To determine the compatibility of a phone and a particular hearing aid, simply add the numerical part of the hearing aid category with the numerical part of the phone emission rating to arrive at the system classification for this particular combination of phone and hearing aid. A total of four would indicate that the combination is usable; a total of five would indicate that the combination would provide normal use; and a total of six or greater would indicate that the combination would provide excellent performance. A category total of less than four would likely result in a performance that is judged unacceptable by the hearing aid user. In theory, the user experience of a combination of an M2 phone and an M3 hearing aid is similar to an M3 phone and an M2 hearing aid.

In ANSI C63.19‐2007, there are contents about the T‐coil in both the wireless device and hearing aid parts. For a phone, which must obtain HAC approval, the T‐coil portion of the standard is not mandatory. The microphone inside any hearing aid can pick up the audio wave transmitted by a phone’s speaker and then amplify it to a level which the user can hear it clearly. This coupling path is called “acoustic passage.” For phones and hearing aids equipped with T‐coils, there exists another passage. T‐coil, also known as “Tele‐Coil,” was originally developed to support hearing aid use with landline telephone handsets that employ a magnetic earpiece transducer (such as those made in the mid‐1980s). Later landline phones began to use piezoelectric transducers that did not generate an H field. Consequently, some landline phones and some cellular phones include a special inductor, which is also called a T‐coil, specifically intended to generate a strong audio‐band H field for T‐coils in hearing aids. Hearing aids equipped with a T‐coil can be configured by the user to disable the microphone and instead reproduce audio that is magnetically coupled to the T‐coil.

If a phone is equipped with a T‐coil, then the RF emission test might have to be repeated. For the T‐Coil mode M‐rating assessment, determine if the T‐coil is contained in an included subgrid of the first scan, for both E fields and H fields. If so, then a second scan is not necessary. The first scan and resultant category rating may be used for the T‐coil mode M rating. Otherwise, the test must be repeated with the 50 mm × 50 mm grid shifted so that it is centered on the axial measurement point of the T‐coil. The lowest category obtained in the first or the repeated tests for either E‐field or H‐field determines the M category assessment.

Shown in Figure 6.27 is the probe for T‐coil measurements. The probe is used to measure the audio‐band magnetic field. Unlike the RF E‐field or H‐field probes, a T‐coil probe can only measure one component of the vector H field at a time. Therefore the test must be repeated three times along three orthogonal axes. Detailed procedures of T‐coil measurements are omitted in the book; more information can be found in ANSI C63.19‐2006 [9].

Unlike SAR standards, a phone can still be sold in the United States even without an HAC certificate. The HAC regulation requests a certain percentage of all phone models sold in the United States by any device manufacturer or any wireless service provider are HAC certified [12]. The keyword here is “phone models.” In theory, a company can comply with the HAC regulation if only 1% of their phones have been HAC certified, as long as their model counts are more than the requirement.

6.2.3 Updates on HAC Requirement

The current HAC standard has been updated to ANSI C63.19‐2011 [13]. The book only addresses some significant revisions. To get more information, please refer to the ANSI C63.19‐2011, which can be purchased from either IEEE Xplore or webstore of ANSI.

The separation distance between test probes and a phone’s front surface used to be 10 mm for both E probe and H probe. It is now 15 mm for E probe and 10 mm for H probe. The distance is measured from the center of an E‐probe’s dipole or an H‐probe’s coil to the front surface of a phone.

AWFs have been replaced by modulation interference factor (MIF). The physics behind the MIF is the same as the AWF. If the MIF value of one standard is 3 dB, its actual E‐field or H‐field limit is 3 dB more stringent than the limit shown in Table 6.7. In the preceding standard, the AWF values are preset in the standard. However, the MIF values have to be measured for each phone models.

Shown in Table 6.8 are sample MIF values given in the ANSI C63.19‐2011. The real values given in each HAC report are a little different, because they are measured on the spot.

Before starting any E‐probe or H‐probe measurement, MIF must be measured first. If one protocol’s conducted power plus its MIF is less than +17 dBm for all its operating modes, the protocol can be exempt from further testing.

The required protocol lists are quite long, which can include GSM850, GSM1900, CDMA‐Full, CDMA 1/8th, UMTS‐RMC, UMTS‐AMR, LTE‐FDD, 2.4 GHz WLAN, 5 GHz WLAN, and so on. For each protocol, MIF needs to be measured on different channels, different data rates, different modulations, and any other combinations if they are applicable. However for most protocols, because either their transmit power is quite low or its MIF is significant small, their sum is less than +17 dBm and can be exempted. In most cases, the protocols, which have to go through E‐probe or H‐probe measurements, are still those old friends, GMS850, GSM1900, and CDMA 1/8th.