With the advent of immunity testing above 1 GHz, the self-filtering properties of power amplifiers may provide the best solution for addressing in-band harmonics.

Illustration by TAISHA PAYTON

Many EMC test houses are currently upgrading their facilities to accommodate radiated immunity testing above 1 GHz. Doing so is expensive because it can require new wall cladding for the anechoic chamber, an additional antenna, upgraded software, and so on. Such an upgrade likely requires a higher-frequency power amplifier. Deciding on an amplifier is, of course, influenced by cost, so opting for a higher-frequency amplifier with a range beyond the incoming standards can offset future costs. However, a higher-frequency amplifier may have a hidden business risk attached. This article outlines some of the system performance issues and cost trade-offs of microwave power amplifiers.

When buying an amplifier, it is worth considering how to avoid having to purchase another in just a few years. One way is to select an amplifier capable of higher frequencies than current standards require. For instance, even though the new maximum frequency is currently 2.5 GHz, choosing an amplifier that goes to 4 GHz or higher would ensure that the system could handle new standards with higher frequency limits that are bound to come into force in the near to middle term. So, despite a higher initial investment, this decision could make good business sense. However, consideration should be given to the issue of in-band harmonics.

In-Band Harmonics

Harmonics are always present within an amplifier output signal. Although they are small in amplitude when the amplifier is well within its linear region, their amplitude increases drastically as the amplifier is pushed toward saturation. In-band harmonics are those harmonics that fall within an amplifier's frequency band. Conversely, out-of-band harmonics fall outside the amplifier's frequency band. By their nature, in-band harmonics are greater in amplitude than out-of-band harmonics. Intuitively, if an amplifier were able to generate harmonics of significant power outside its frequency band, it would also be able to amplify regular signals outside its frequency band. It can't, so it doesn't.

In-band harmonics were not a great issue when the upper frequency limit was 1 GHz. By a happy coincidence, the technology used in the manufacture of these solid-state amplifiers means that the amplifier just about makes it to 1 GHz, and then the output power rolls off abruptly. Figure 1 shows this characteristic.

Figure 1. Output power plot of an 80–1000-MHz amplifier showing a steep roll-off at 1000 MHz.

The steep roll-off is caused by a combination of two factors: the limitation of the radio-frequency (RF) transistors and the limitation of the internal and external power combining circuits. This self-filtering capability means that any harmonic noise generated by the amplifier above 1 GHz is out of band, small in value, and relatively inconsequential.

Out-of-band harmonics are typically 10–20 dB lower than in-band harmonics. So, if in-band harmonics are at a power level of say, 1 W, this equates to out-of-band harmonics at a power level of somewhere between one-hundredth and one-tenth of a watt. Consequently, out-of-band harmonics have not been an issue when testing to 1 GHz. Figures 2 and 3 show the in-band and out-of-band harmonics of a typical 80 MHz to 1 GHz solid-state power amplifier. Note that in this particular case, the push-pull transistor configuration used in these amplifiers suppresses the second harmonic, thereby leaving the third harmonic as the dominant harmonic. (The second harmonic occurs at twice the frequency of the fundamental or intended signal; the third harmonic occurs at three times the fundamental.)

Figure 2. In-band harmonic of an 80–1000-MHz power amplifier. The fundamental is at 300 MHz; the third harmonic is at 900 MHz.

Figure 3. Out-of-band harmonic of 80–1000-MHz power amplifier. The fundamental is at 600 MHz; the third harmonic is at 1800 MHz.

New Radiated Immunity Standards. An amplifier with an upper frequency that extends beyond that stipulated by an immunity standard could inadvertently radiate equipment under test (EUT) with nontrivial fields of the wrong frequency. If an EUT fails the radiated immunity test, it would be unclear whether the radiation was caused by intentional fields as stipulated by the standard or by unintentional fields caused by in-band harmonics. Clearly, both test houses and their customers need a high degree of confidence that such failures are due to susceptibility to intended radiation. It would be a nightmare scenario for both parties if time and money are lost investigating susceptibility to unintentional fields. It is critical, therefore, to adhere to the maximum harmonics stipulated by the relevant standard. It is equally important to understand the effects of in-band harmonics.

Unwanted Fields

Several contributors generate unwanted fields: the gain characteristic of the antenna, the nonideal characteristic of the anechoic chamber, the requirement for modulation, and, of course, the contribution of the microwave power amplifier itself (microwave frequencies are loosely defined as 1–30 GHz).

Antenna. It is a well-known fact that antennas tend to be more efficient radiators at higher frequencies. The danger, then, is that an antenna could convert high-frequency harmonic signals into fields of significant volts per meter.

Antenna manufacturers supply performance data to show that an antenna meets certain performance parameters across the intended band of use. One or more of the performance parameters probably drops off just outside the intended band, but then the radiation performance can take off again over another octave. An octave is a doubling of the frequency (e.g., 500 to 1000 MHz, 1 to 2 GHz, 2 to 4 GHz, 4 to 8 GHz, etc.) With no data on the radiation performance outside the intended band, the value of the fields radiated by an antenna presented with in-band harmonic noise is uncertain.

Electromagnetic compatibility (EMC) antenna designers strive to extend the usable frequency band, and today the upper limit stretches to several gigahertz. For these antennas, performance data are available for the in-band harmonics frequency range. However, antennas make no distinction between wanted and unwanted signals, converting both signal types into electromagnetic fields. The higher-frequency signals will tend to be favored in terms of radiation efficiency.

Anechoic Chamber. Ideally, an anechoic chamber is indeed anechoic, and all incident electric fields are completely absorbed by absorptive tiles fastened to the walls, ceiling, and floor of the room. This arrangement is an attempt to replicate free-space conditions in which:

This ideal is never fully realized. Because of inconsistent absorption and myriad reactions between the radiating antenna and the room, no two practical chambers seem to perform the same. To understand the problems this can cause, it is necessary to revisit the calibration process. Before subjecting an EUT to the RF immunity test, the test system is calibrated to ensure the radiated field strength is within pre-defined limits at all frequencies. Field-strength measurements are taken at points across an imaginary plane perpendicular to the direction of the radiated field. These points are equi-distant and form a matrix across the plane. Because of the close proximity of the antenna, the field passing through the plane is not a perfect plane wave.

The field strength measured at the plane is the result of a combination of the field radiated directly from the antenna and any field distortion or reflections caused by imperfections in the anechoic chamber. These departures from the ideal indicate that varying levels of power are needed from the amplifier to obtain the required field strength limits across the plane (in fact, the procedure allows some of the measurement points to be discounted in cases of wide variation).

For each measurement frequency, the input power level to achieve the required field strength at the weakest point in the plane is recorded in a searchable table. The test system software uses the table to control the input power to the amplifier during actual EUT testing. Higher power from the amplifier at certain frequencies inevitably means the amplifier is closer to saturation.

Requirement for Modulation. During the test of the EUT, 1 kHz at 80% modulation is applied to the radiated field. The modulated carrier has the typical AM modulated envelope comprising periodic peaks and troughs. The peak envelope power (the power at the power peaks) is 3.2* greater than the power of the unmodulated carrier. This modulated waveform is applied to the amplifier input, resulting in peak input powers that are 3.2* greater than those in the calibration table. Therefore, the requirement for modulation pushes the amplifier further toward saturation.

Microwave Power Amplifier. The microwave amplifier's contribution to unwanted fields is limited mainly to in-band harmonics. Solid-state microwave amplifiers with various frequency bands and power levels are readily available from several manufacturers. Microwave amplifiers display the same self-filtering characteristic as RF amplifiers, with in-band harmonics typically exceeding out-of-band harmonics by 10–20 dB. Figure 4 shows the output power plot for a typical 0.8–2.5-GHz power amplifier. At 2.5 GHz, the power drops abruptly, and this self-filtering capability provides a stong distinction between in-band (see Figure 5) and out-of-band (see Figure 6) harmonics.

Figure 4. Output power plot of 0.8–2.5-GHz power amplifier displaying same steep roll-off at upper band edge.

Figure 5. In-band harmonic of 0.8–4.2-GHz power amplifier. The fundamental is at 1.3 GHz (see Figure 6) and the second harmonic is at 2.6 GHz.

Figure 6. Out-of-band second harmonic of a 0.8–2.5-GHz power amplifier at the same output power as Figure 5. The fundamental is at 1.3 GHz; the second harmonic is at 2.6 GHz. For the same amplifier output power, the out-of-band harmonic is 10 dB lower than the in-band harmonic shown in Figure 5.

Note that the second harmonic dominates in Class A amplifiers, and in both cases, the magnitude of the dominant harmonic is similar for the same amplifier saturation level. Figure 7 shows a 0.8–4.2-GHz power amplifier amplifying a 1.3-GHz signal. The second harmonic at 2.6 GHz falls in band and is, therefore, high relative to the fundamental 1.3-GHz signal. Compare this with Figure 6 where an identical 1.3-GHz signal produces a 2.6-GHz harmonic that is out of band.

Figure 7. In-band harmonic of a 0.8–2.4-GHz power amplifier. Again, the fundamental is at 1.3 GHz (see Figure 6) and the second harmonic is at 2.6 GHz.

Identifying Hot Spots

Harmonic power levels increase with amplifier output power, and this characteristic is useful in identifying situations in which high harmonic levels may be present. Hot spots occur when:

For example, when testing up to 2.5 GHz with a 0.8–4.2-GHz amplifier, a relatively high input power at 2 GHz could mean that harmonic power is present at 4 GHz. This applies to all test frequencies between 1.25 and 2.1 GHz. Examination of the calibration table identifies these hot spots. Higher RF input power levels along with the additional contributors could indicate that significant harmonic fields are present during the test of the EUT.

Quantitative Measurement of the Harmonic Noise. Once the hot spots are identified from the table, applying modulation allows the worst-case scenario to present itself for further investigation. With the worst-case condition in place, it is still impossible to see the actual harmonic field strengths. The field strength probe is unable to discriminate between the fundamental (wanted) field and the field generated by harmonic noise—it simply provides composite field strength. Also, the power indicator on the power amplifier is unable to discriminate between fundamental power and harmonic power. So, neither measurement means allows the harmonic power levels to be monitored directly. The preferred method is by direct visual monitoring and measurement.

Fortunately, EMC test houses have access to receivers or spectrum analyzers used in identifying hot spots during emissions testing in the anechoic chamber. These instruments clearly measure harmonic power levels reaching the input of the antenna. Ideally, a power amplifier would have a sample port to enable the output spectrum to be monitored. Failing this, a suitable directional coupler can be inserted into the RF path at the output of the power amplifier.

By monitoring the harmonic power levels and knowing the attenuation path (coupling factor of the directional coupler, in-line attenuator values, etc.), the power levels of the displayed harmonics can be calculated. Once the harmonic power levels leaving the amplifier are known, the field strengths produced by the harmonics can be calculated using the cable loss to the antenna and the antenna factor data provided by the antenna manufacturer. This assumes that antenna data covering the in-band harmonic frequencies are available. If antenna factor data are unavailable from the manufacturer, practical data can be collected by monitoring the field strength while conducting a constant-power frequency sweep of the antenna over the band of interest.

Solutions to the In-Band Harmonic Problem

Possible solutions to excessive in-band harmonics are to install a filter after the amplifier, use an antenna with a suitable cutoff frequency, use a higher power amplifier, or use the self-filtering capability of a suitable amplifier.

Filter. Installing a filter with a cutoff frequency at 2.5 GHz removes the unwanted fields above 2.5 GHz. The filter needs a steep roll-off at 2.5 GHz and must be able to withstand the full power of the amplifier. The filter's design would, therefore, be nontrivial. Obviously, the filter would need to be switched out or physically removed for a test requiring frequencies above 2.5 GHz.

Antenna. Using the antenna to filter out higher frequencies is probably not a viable option. As stated previously, antennas are primarily designed to perform within certain parameters over a specific band; however, they may radiate efficiently at frequencies outside the band of operation. To apply this solution with confidence, it is necessary to have the antenna's performance data beyond its operation band.

Higher-Power Amplifier. If a 500-W amplifier is backed off to 50 W, the in-band harmonic levels are low. Unfortunately, this solution is expensive (a tenfold increase in power approximates to a tenfold increase in price). Although this is sometimes presented as the best technical solution, it is a bulldozer approach. It is also important to consider that a conflict of interest between buyer and seller may exist with this option. Other solutions should be given full consideration first.

Self-Filtering. The preferred option may be to copy the example of the 1-GHz scenario and use the self-filtering properties of, say, a 2.5-GHz power amplifier. Figures 5 and 6 demonstrate the self-filtering property of a 0.8–2.5-GHz power amplifier. The ideal solution with this approach would be a range of amplifiers switched in and out as required to make full use of each amplifier's self-filtering property. Multiband amplifiers with local and remote band-switching capability have been around for years, so real examples of this solution already exist.

Conclusion

Nontrivial fields caused by in-band harmonics are a real and present hazard for both test houses and manufacturers. As new standards come into force, it may be difficult to identify the source of in-band harmonics if an amplifier's upper frequency is beyond that of the new standard. Neither the customer nor the test house wishes to lose time and money investigating susceptibility to unintentional fields. Due diligence can minimize this danger. Due diligence entails an understanding of the issues combined with necessary steps to minimize the hazard. Observing the maximum harmonic levels stated in the relevant standard is critical. Collection of hard data on harmonic field strengths produced at hot spots can provide an additional level of confidence.

The old EMC design adage of minimizing noise at the source is applicable here, and using the self-filtering properties of power amplifiers could prove to be the best solution.