Design of High-efficiency Continuous Class F Doherty Power Amplifier

The high efficiency of the saturation state and rollback state is an important part of the front end of modern wireless communication systems. Improving the efficiency of RF power amplifiers is of great significance for energy saving and amplifier performance. Based on the traditional Doherty power amplifier, the harmonic suppression network is introduced before the fundamental frequency matching, so that the second harmonic impedance and third harmonic impedance can achieve continuous class F impedance matching conditions, and then the saturation efficiency and regression efficiency can be improved without affecting the fundamental wave matching. The results of the simulation show that the saturation efficiency of the Doherty amplifier reaches 73.94%~77.22%. The drain efficiency of back-off 6 dB reaches 61.80%~62.92%. The average saturation drain efficiency is 75.58%, and the average back-off 6 dB drain efficiency is 62.36%.


Introduction
With the popularity of network video, it is necessary to increase the data transmission speed to meet the needs of communication services.To meet the development of wireless communication technology, there are higher requirements for the efficiency and transmission rate of amplifiers.Orthogonal frequency division multiplexing (OFDM) [1] high-order modulation technique uses multiple baseband signals with similar frequencies for carrier modulation, which reduces the bandwidth of the signal but produces a high peak-to-mean power ratio, which reduces the back-off efficiency of the RF amplifier.To improve the efficiency of the traditional power amplifier at the regression point, there are many solutions, such as outphasing input [2], envelope following, and Doherty technology [3], among which Doherty technology has been widely concerned by the academic community due to its simple structure, easy implementation, and excellent performance.
A combined network is adopted to integrate drain-source capacitor Cds into output matching, thus achieving high back-off drain efficiency [4].Multi-state integrated output matching is adopted to enhance the back-off efficiency of the power amplifier [5].A CM-DPA harmonic control method is proposed to improve efficiency [6].A new hybrid continuous class EFJ power amplifier is designed [7], and harmonic matching is used at the junction.The harmonic control topology is added to the output matching junction, which effectively increases the back-off efficiency [8].By properly setting the bias of two peak power amplifiers [9], a modulation is proposed to expand the bandwidth and improve the efficiency of backtracking.However, one more branch increases the complexity of the circuit and also increases the losses.
In this paper, the harmonic suppression network and Doherty structure are combined under the condition of continuous class F impedance matching, so as to improve the saturation efficiency and rollback efficiency without affecting the fundamental wave matching.Based on this method and traction technology, a high-efficiency 1.9~2.1 GHz power amplifier is realized.

Harmonic suppression network design
Harmonic control is one of the effective technical means to increase the back-off efficiency of power amplifiers, but the traditional Doherty power amplifiers usually take no account of the influence of the impedance matching at double and triple fundamental frequencies.To solve this problem, a harmonic suppression network is added in front of the fundamental frequency matching in this paper.In this way, the impedance at double and triple fundamental frequencies after the addition of harmonic suppression network reaches the condition of continuous class F [10], thereby improving the regression efficiency and saturation efficiency.

Figure 1 Harmonic suppression network
Figure 1 shows the circuit schematic diagram of the harmonic suppression network.In the figure, Z load represents the equivalent impedance seen from the reference surface; Z 1 , Z 2 , and Z 3 represent the equivalent impedance seen from the dotted line to the right; TL 1 and TL 2 represent the tuned microstrip lines; Z T1 and Z T2 represent the characteristic impedance of TL 1 and TL 2 ; θ T1 and θ T2 represent the electrical length of TL 1 and TL 2 ; TL open1 and TL open2 represent the quarter-wavelength open transmission line under the third harmonic; Z open1 , θ open1 , and Z open2 , θ open1 represent the characteristic impedance and electrical length of TL open1 and TL open2 , respectively; TL open3 represents the quarter-wavelength open transmission line under the second harmonic; Z open3 represents the characteristic impedance and θ open3 represents the electrical length of TL open3 ; TL short represents the quarter-wavelength open transmission line at the fundamental frequency; Z short represents the characteristic impedance and θ short represents the electrical length of TL short .
To achieve the best 3 rd harmonic matching, quarter wavelength open transmission lines TL open1 and TL open2 are used to achieve the short circuit at A, and then TL 1 converts the short circuit at A into the open circuit at the reference plane.And to achieve the best 2 nd harmonic matching, a quarter wavelength open transmission line TL open3 and a half wavelength short transmission line TL short are used to realize the short circuit at B, while the half wavelength short transmission line TL short converts the short circuit into an open circuit at B, and then the 2 nd harmonic impedance is adjusted to the condition of continuous class F by adjusting the length of TL 2 .

Optimization of Doherty power amplifier
The substrate is Roger4350, the thickness is 0.508 mm, the thickness of copper skin is 35 um, and the loss tangent is 0.0037.The capacitors are all Murata GRM18 models.
Figure 2 shows the basic connection block diagram after adding a harmonic control circuit, where the output matching consists of a harmonic suppression network and fundamental frequency matching.This design uses a reverse load modulation network, and the basic diagram includes a Wilkinson power divider, carrier power amplifier, peak power amplifier, and quarter wavelength converter.The phase shift line is utilized to adjust the phase equality of the two signals at the closing point to achieve the maximum power transmission.The peak compensation line is utilized to reach the peak load impedance of infinite before the peak regression point.
When the input signal is very small, the carrier amplifier works, and the efficiency curve is similar to the traditional one.When there is a large signal, the power will fluctuate greatly, and the peak amplifier will be turned on all the way so that there is high efficiency at both high and low input.
The best equivalent impedance looking into the load and the source at 2 GHz is obtained by load pulling and source pulling.The input power at the back 6 dB point is 20 dBm, and the input power corresponding to the saturation point is 28 dBm.The carrier power amplifier is Class AB biased, the grid bias voltage is -2.9 V, the drain bias voltage is 28 V, the drain static current is 115 mA, and impedance points through load traction and source traction are shown in Table 1.
The input matching uses four stepped impedance transmission lines to match 10.33+j*0.75Ω to 50 Ω in the original impedance diagram.The matching method of two-stepped impedance transmission lines and harmonic suppression network is adopted.The junction point loaded at the backoff point is 25 Ω, while the junction point loaded at the saturation point is 50 Ω.The output matching circuit is designed to match the saturation point of 50 Ω to 17.23+j*13.88Ω and 25 Ω to 22.01+j*21.17Ω.The peaking power amplifier is carried out.The carrier power amplifier operates under Class C conditions, the gate bias is 5.6 V, the drain bias is 28 V, and the drain static current is 28 uA.The impedance points after load pulling and source pulling are shown in Table 2.The load impedance of the peak power amplifier before the regression point is infinite, and a section of the peak compensation line can be used to transform the load impedance of the closing point to infinity.The impedance value of the peak power amplifier is 50 Ω, which is adjusted to 14.58+j*9.60Ω after matching the impedance output of the three steps through the peak compensation line.The difference between the improved Doherty the traditional one is that the harmonic suppression network is added before the fundamental frequency matching in the output matching.The impedance converter uses a section of characteristic impedance of 35.3 Ω and an electrical length of λ/4 to convert the load impedance from 50 Ω to 25 Ω.
Figure 3 shows the specific parameters of the circuit schematic after matching.As can be seen in Figure 3, the 2 nd impedance is on the left edge of the original impedance diagram, and the 3 rd harmonic impedance approaches infinity.Thus, by adding a harmonic suppression circuit in front of the traditional fundamental matching, the impedance at double and triple fundamental frequencies is controlled in the impedance design of continuous class F.   As can be seen from Figure 4, continuous drain voltage and current waveform of class F are basically realized.The overlapping area between voltage and current waveform is reduced, so that the power loss between the drain is reduced, the output power is increased, and the efficiency is improved.As can be seen from Figure 5, after adding the harmonic suppression network, both the second and third harmonics are greatly suppressed, and the power spectrum is reduced by 10 dBm~20 dBm, thus improving the fundamental frequency power and drain efficiency.The simulation results in Figure 6, the gain of the conventional Doherty amplifier is 12.3~18.0dB, and 11.2~19.5 dB with the addition of the harmonic suppression network.The conventional Doherty amplifier has a back-off 6 dB drain efficiency of about 55.69% to 56.87%, and the improved Doherty amplifier has a back-off 6 dB drain efficiency of about 61.80% to 62.92%, and the saturated drain efficiency is improved from 68.31% to 71.19% to 73.94% to 77.22%.The back-off 6 dB drain efficiency is increased by approximately 7% and saturation drain efficiency is increased by approximately 6% compared to conventional Doherty amplifiers.

Conclusion
By adding a harmonic suppression network into the output matching, the continuous class F impedance transformation condition is realized without changing the fundamental frequency matching.The 2 nd harmonic and 3 rd harmonic are suppressed, the fundamental frequency power is improved, and the backoff 6 dB drain efficiency and the drain saturation efficiency are increased to a certain extent.The energy loss of the RF transceiver is reduced and the heating problem of the equipment is effectively solved.This design method is very suitable for practical application and has important significance for the highperformance realization of modern wireless communication systems.

Figure 2 Figure 3
Figure 2 Basic block diagram of the improved Doherty amplifier

Figure 4
Figure 4 Drain voltage and current waveform (a) Carrier power amplifier (b) peak power amplifier

Figure 5
Figure 5 Spectrum simulation comparison diagram (a) Spectrum diagram of output power without harmonic suppression network (b) Spectrum diagram of output power with harmonic suppression network

Figure 6
Figure 6 (a) Layout simulation of the network without harmonic suppression (b) Layout simulation of network with harmonic suppression

Table 1
Optimal Load and Source Impedance of Carrier PA

Table 2
Optimal Load and Source Impedance of Peak PA