【正文】 y of the amplifier can be increased even more, which can be seen from . Drain efficiency increases steadily when supply is lowered. At a supply voltage of V and frequency of 1625 MHz, the amplifier has a drain efficiency of 65%. This implies that the amplifier can maintain an efficient operation also when used in an envelope elimination and restoration (EER) system. The high peaks in at lowest supply voltages in Fig. 14are caused by drive signal feedthrough that sums into the output signal. Load pull measurements Measurements were performed with load pull system GHz spot frequency to several modified amplifiers. The differences between the amplifiers are shown in the next chapters we will mainly concentrate on amplifiers C and D for reasons that will be apparent later on. Let us now discuss the amplifier C which is very similar to amplifier B measured earlier. Tuners of the load pull system were connected to the outputs and inputs of the amplifier C. The load tuning of fundamental, second harmonic and third harmonic resulted in about W of output power ( dBm) while maintaining about % drainefficiency at this peak power spot. The fundamental load impedance in terms of power was at slightly higher impedance than the optimum drain efficiency point as shown in , where the 1 dB output power points (triangles) and5% unit efficiency points(circles) are shown. The peak efficiency point in the figure is (a) (%) and peak power is (b) ( dBm). The optimum efficiency area is rather large. Both of the load harmonics were even more relaxed, and differences for example in output power had to be measured in tenths of decibels rather than in decibels. Further,the efficiency differences were measured in one or two percentage units instead of five to ten. As an example, the third harmonic optimum output points within dB from maximum (triangles) and efficiency points within 2% units from maximum (circles) are shown in Fig. 16. As it can beseen, the third harmonic impedance is not as critical as the fundamental tone. The optimum efficiency is marked with(a) and maximum output power is marked with (b). It should be noted that the adjustment of the third harmonic did not increase output power on the absolute scale nor the drain efficiency. The peak output power value remained within dB of the peak value of the fundamental load pull and the drain efficiency rose from only to %shown at the Smith chart point (a) in . The insensitivity of the amplifier to harmonic tuning is caused by the long drain bias line that is low impedance at the second harmonic and the low pass matching network at the output that attenuates the third harmonic. Source pull measurementsThe source pull of the fundamental impedance did increase the output power and efficiency of amplifier C slightly. The optimum drain efficiency (circles) and output power points(triangles) (a) and (b), respectively, are shown Fig. 17. The power points are within the limits of dB and efficiency within a difference of 2% units. The amplifier efficiency rose with the fundamental source tuning to % (point a)and the output to about W ( dBm, point b). Harmonic source pull measurements showed that the harmonic impedances were not as critical as the fundamental. This is due to the low pass input matching and the wideband RCsink circuit. The RCsink circuit lowers the calculated magnitude of the impedance especially at high frequencies,as shown in (b). This has an effect to both second and third harmonic impedances. The magnitude of the impedance without the RCsink is shown as a reference in (a). In both cases the input matching circuits were not included in the calculations. Stability of the amplifier At first the amplifier A did show some unstable behaviour due to supply voltage modulation caused by insufficient bias decoupling at the drain. The instability appeared at low input power levels as noise sidebands that lied on both sides of fundamental frequency. When input power was lowered further on, the amplifier did break into full scale oscillation. As a cure, the supply impedance was lowered by a large number of decoupling capacitors (4 470 pF)added to the drain. In an EER application, the supply modulator will provide low enough impedance at the drain. When the load pull was done to the amplifiers A, C and D, a spurious oscillation detection was applied at a level of50 dBc. With this setup we were able to pare the sensitivity of different amplifiers to oscillations. We found out that the RCsink circuit used in amplifiers A and C indeed improved stability, especially in the low input power levels. Data used for parison was measured from amplifier D, where the RCsink was cut using an ultraviolet laser. The oscillation points of the fundamental impedance load pull with a low 15 dBm input power are shown in . The oscillation sidebands detected are1701 and 1489 MHz. If we pare this result to amplifier A, the amount of found oscillation points is considerably smaller and the location of them is rather tightly spaced in the low impedance area, as shown in Fig. 20. The oscillation frequency in this case is 1568 MHz. The amplifier A and amplifier D had a different frequency for the modulating spurious ponents: Without the damping circuit the modulating spurious was ca.177。Schwarz SMU 200A with a buffer amplifier. The input power levels were from 15 to 25 dBm. The RF input and output powers were measured with Anritsu ML2438Apower meter with dual input. The harmonic content of the spectrum and oscillation spikes were measured with Rohdeamp。 Fusco, V. (2005). Seriesl/paralleltuned classepower amplifier analysis. InProc. European microwave conference(Vol. 1, p. 4). doi:5. Tayrani, R. (2007). A spectrally pure w, high pae, (6–12 ghz)gan monolithic class e power amplifier for advanced t/r Proc. IEEE radio frequency integrated circuits (RFIC) symposium(pp. 581–584). doi:6. Hietakangas, S., Rautio, T., amp