Discussion: Impact of Test Signal on the Measured AM/AM and AM/PM Characteristics

The AM/AM and AM/PM characteristics of power amplifiers and transmitters are quite sensitive to the type of drive signal and to its characteristics. In [14], the AM/AM and AM/PM characteristics of a power amplifier prototype were derived for three different excitations signals: a continuous wave, an eight-tone, and a WCDMA (wideband code division multiple access) signal. As illustrated in Figure 8.4, there is significant discrepancy between the characteristics measured with the modulated signal and those measured with the continuous wave signal. However, the multi-tone test signal leads to closer prediction of the DUT nonlinear characteristics. This corroborates the conclusions mentioned in the previous

Measured AM/AM and AM/PM characteristics of a power amplifier prototype for various excitation signals

Figure 8.4 Measured AM/AM and AM/PM characteristics of a power amplifier prototype for various excitation signals. (a) AM/AM characteristics and (b) AM/PM characteristics [14]. ©2009 IEEE. Reprinted, with permission, from S. Boumaiza et al., “Systematic and adaptive characterization approach for behavior modeling and correction of dynamic nonlinear transmitters,” IEEE Transactions on Instrumentation and Measurement, Dec. 2007

sub-sections according to which the use of the modulated signal is more appropriate for accurate characterization of the DUT’s nonlinear behavior.

A closer look at the sensitivity of power amplifiers’ and transmitters’ dynamic nonlinear behavior to the excitation signal reveals that even for a given type of modulated signals, such as CDMA or OFDM, for example, the signal characteristics might noticeably impact the response of the DUT. The characteristics of modulated signals mainly include their average power, bandwidth, and statistics. To describe the statistics of modern communication signals, the complementary cumulative distribution function (CCDF) is widely adopted. The CCDF provides a thorough portrayal of the signal statistics by depicting for each power level above the average power of the signal and the percentage of time the signal power is at or above that value. For a given communication standard, the CCDF curves of all signals are quite similar and variations only occur toward the tail of the curve as the peak to average power ratios of the signals vary. However, these changes are imperceptible by the power amplifier since the CCDF disparity occurs for signal samples with very low probability. The measured memoryless AM/AM and AM/PM characteristics of a power amplifier prototype driven by single carrier WCDMA signals having various peak to average power ratios are reported in Figure 8.5. In these measurements, the bandwidth as well as the average power of the test signals were kept unchanged and only the signal’s PAPR (peak-to-average power ratio) was varied. These results confirm the unnoticeable effects, on the nonlinear behavior of power amplifiers and transmitters, of peak to average power ratio variations in signals of the same

Measured AM/AM and AM/PM characteristics of a power amplifier prototype for WCDMA signals with various PAPR

Figure 8.5 Measured AM/AM and AM/PM characteristics of a power amplifier prototype for WCDMA signals with various PAPR. (a) AM/AM characteristics. (b) AM/PM characteristics standard [19]. However, it is important to mention that any inconsistency in the CCDF characteristics of excitation signals will lead to different nonlinear behaviors of the DUT even if the PAPRs of these signals are comparable. This is the main reason that calls for the careful engineering of multi-tone test signals with particular attention to the distribution of their phases.

Another parameter of the modulated test signal is its average power. The behavior of power amplifiers is sensitive to variations in the average power of the input signal as investigated in [19]. Figure 8.6 depicts the memoryless AM/AM and AM/PM characteristics of a Doherty power amplifier prototype measured for single carrier WCDMA waveforms having different peak to average power ratios and average powers. The average power of each signal was set such that the DUT is operated over its entire power range up to but not beyond its saturation. Accordingly, for each signal, the DUT was driven at an output power back-off that is equal to the signal’s peak to average power ratio. The substantial changes in the measured characteristics, especially the AM/AM ones, reveal the dependency of power amplifiers’ and transmitters’ nonlinear behavior to the average power of the drive signal. This dependency can be solely attributed to the average power variation since no changes in the DUT nonlinearity was observed when the same signals were applied at constant average power (Figure 8.5). Similarly, the nonlinear behavior of power amplifiers and transmitters depend on the signal’s bandwidth, which has a considerable impact on the memory effects generated by the DUT. As the drive signal bandwidth increases, the memory effects exhibited by the DUT become more pronounced. This translates into more significant dispersion in the measured AM/AM and AM/PM [20]. Figure 8.7 illustrates an example of the influence of the signal bandwidth on the measured AM/AM characteristics of a PA using LTE (Long-Term Evolution) signals having similar average power but different bandwidths. Similar effect is observed in the AM/PM characteristics.

 
Source
< Prev   CONTENTS   Source   Next >