978-1-4799-0902-5/13/$31.00 ©2013 IEEE 509

Linearization of Broadband Doherty Amplifier Aleksandra Đorić1, Nataša Maleš-Ilić2, Aleksandar Atanasković2,

Bratislav Milovanović2

Abstract – The linearization of broadband two-way Doherty

amplifier for application in the frequency range 0.85-1.15GHz is considered in this paper. The carrier and peaking amplifiers in Doherty configuration comprise Freescale’s transistor MRF281S LDMOSFET characterized by the maximum output power 4W and the broadband lumped element matching circuits. The linearization of the amplifier is carried out by the second harmonics and fourth-order nonlinear signals that are extracted at the output of the peaking cell, adjusted in amplitude and phase and injected at the input and output of the carrier cell in Doherty amplifier. The effects of linearization are considered for two tones separated in frequency by 10MHz, 20MHz, 40MHz and 80MHz at different input power levels of 0dBm, 5dBm, 10dBm and 15dBm, as well as for WCDMA digitally modulated signal.

Keywords – Doherty amplifier, broadband, linearization, second harmonics and fourth-order nonlinear signals, intermodulation products.

I. INTRODUCTION

Modern wireless comunication systems (CDMA-2000, WCDMA, OFDM etc.) are developing toward the augmentation of frequency bandwidth to transmit a large number of carriers, with high velocity. Demanding requirements of new systems in order to meet both linearity and high power efficiency present a serious task for transmitter designers. The Doherty amplifier is capable of achieving a high efficiency of power amplifiers in base station. Different linearization methods have gained significant interest with the aim to reduce the nonlinear distortions while keeping the power amplifier in efficient mode. The various linearization methods of Doherty amplifier have been reported: post-distortion-compensation [1], the feedforward linearization technique [2], the predistortion linearization technique [3] and combination of those two linearization techniques [4].

The linearization effects of the fundamental signals’ second harmonics (IM2) and fourth-order nonlinear signals (IM4) at frequencies that are close to the second harmonics to the standard narrowband (two-way, three-way and three-stage) Doherty amplifiers were investigated through the simulation process, [5], [6] as well as the experiment, [7].

In this paper, a broadband two-way Doherty amplifier with

1Aleksandra Đorić is with the Innovation centre of advanced technology

Niš, Serbia, Vojvode Mišića 58, 18000 Niš, Serbia, E-mail: alexdjoric@yahoo.com.

2Nataša Maleš-Ilić, Aleksandar Atanasković, and Bratislav Milovanović are with the Faculty of Electronic Engineering, University of Niš, Aleksandra Medvedeva 14, 18000 Niš, Serbia, E-mails: [natasa.males.ilic; aleksandar.atanaskovic; bratislav.milovanovic] @elfak.ni.ac.rs

the additional circuit for linearization is designed to operate over the frequency range 0.85-1.15GHz. The linearization technique applied utilizes the second harmonics and fourth-order nonlinear signals at frequencies close to the second harmonics, which are generated at the output of the peaking cell. They are adjusted in amplitude and phase through the linearization branches and run at the carrier transistor input and output over the bandpass filters. The effects of the linearization are considered through the simulation for two sinusoidal signals with different frequency interval between them starting from 10MHz and going up to 80MHz, and for input signal power levels rangeing up to approximatelly 3dB below saturation. Additionally, Doherty amplifier is linearized for WCDMA digitally modulated signal.

II. LINEARIZATION

Theoretical analysis of the linearization approach that uses the second harmonics and fourth-order nonlinear signals (IM2 and IM4) for linearization has been given in [5] and [7]. According to this, it is possible to reduce spectral re-growth caused by the third-order distortion of fundamental signal by choosing the appropriate amplitude and phase of the IM2 signals injected at the input and output of the amplifier transistor. Additionally, the fifth-order intermodulation products can be suppressed by adjusting the amplitude and phase of the IM4 signals that are inserted at the input of amplifier transistor and put at its output.

The IM2 and IM4 signals generated at the output of peaking transistor are extracted through the bandpass filter characterized by the center frequency around the second harmonics to pass the signals for linearization (IM2 and IM4 signals). The IM2 and IM4 signals are tuned in amplitude and phase by the amplifier and phase shifter over two independent linearization paths. They are inserted at the input and output of the carrier amplifier transistor over bandapass filters.

According to the theoretical analysis of the linearization approach, [5], IM2 and IM4 signals can reduce both IM3 and IM5 products. However, the suppression rate depends on the relations between amplitudes as well as phases of the IM2 and IM4 signals generated at the peaking amplifier output. Nevertheless, when the required relations are not fulfilled, only one kind of the intermodulation products can be lowered sufficiently that will be confirmed by the results presented in further text.

510

Fig. 1. Schematic diagram of broadband two-way Doherty amplifier with additional circuit for linearization

III. BROADBAND DOHERTY AMPLIFIER DESIGN

Agilent Advanced Design System-ADS software has been used for the design of two-way Doherty amplifier for broadband application, which schematic diagram is shown in Fig. 1.

Two-way Doherty amplifier was designed in standard configuration [1], [2], [4], [5] with two quarter-wave impedance transformers with the characteristic impedance R0=50Ω and 0 2tR R= Ω in the output combining circuit. In a low-power region, the peaking amplifier should be an open circuit until it is turned on in a higher power region. Phase difference of 90° caused by the 50Ω quarter-wave impedance transformer at the output is compensated at the inputs by 3dB quadrature branch-line coupler. In a low-power region, the output impedance of the peaking transistor is transformed from strongly reactive to the open by the output matching circuit and the proper offset line to prevent considerably power leaks from the carrier amplifier to the peaking amplifier.

The carrier and peaking cells were designed using Freescale’s MRF281S LDMOSFET which non-linear MET model is incorporated in ADS library. The transistor shows a 4-W peak envelope power. The matching impedances of carrier cell for source and load at 1GHz are ( )5.5 15sZ j= + Ω and

( )12.5 27.5lZ j= + Ω , respectively. In the case of peaking cell

they are ( )3.55 15.7sZ j= + Ω and ( )3.95 30.85lZ j= + Ω . These impedances were obtained by using load-pull and source-pull analysis in ADS.

The carrier amplifier is biased in class-AB (VD = 26 V, VG = 5.1 V (13.5% IDSS-transistor saturation current)), whereas the peaking amplifier operates in class-C (VD = 26 V, VG = 3.6 V).

In order to design a broadband amplifier circuit, the input and output matching circuits of the transistors are based on the filter structures with lumped elements. Primarily, the third order lowpass filter prototype was designed. The method of minimum reflection [8], [9] was used for calculating the

normalized admittances of the prototype elements which were then transformed to the reactive elements of the lowpass filter and bandpass filter sequentially by using adequate calculations [8]. Norton transformations [9] were applied in order to reduce values of some inductances and capacitances in the matching circuits, which did not correspond to the commercially available components. These transformations provide the scaling of the terminating resistors of matching circuits upwards or downwards to 50Ω. After applying Norton transformations, approximately the same transmission characteristic and reflection losses were achieved in comparison with the basic matching circuit with lumped elements.

Norton transformations implemented at the input matching circuit increase the terminating impedance from 31.87Ω to 50Ω, while, at the output matching circuit, another type of Norton transformations was exploited to reduce the terminating load from 56.45Ω to 50Ω. The peaking amplifier was designed by transforming the input impedance upward from 37Ω to 50Ω and output impedance downward from 72.185Ω to 50Ω.

The stabilization of the carrier and peaking cell was performed by the resistances connected in parallel with RF chocks in DC power supply circuits of the transistor, as well as by the resistance parallel to the input matching circuit in both amplifying cells, as shown in Fig. 1. The values of these resistances were selected in the range 300-600Ω in order to prevent losses of the fundamental signals.

The proposed linearization technique applied to the narrowband amplifiers [5]-[7] extracts the signals for linearization (IM2 and IM4) at the output of the peaking cell and feeds them to the input and output of the carrier amplifier transistor over the frequency diplexers that separate fundamental signals and their second harmonics. As this paper analyzes the broadband amplifier linearization, the application of diplexers will not give satisfactory results. The IM2 and IM4 are extracted through the bandpass filter and also they are delivered to the input and output of the carrier amplifier transistor throughout the bandpass filters characterized by

511

2GHz center frequency and 1GHz frequency bandwidth.

Fig. 2. Output power, Gain, and

PAE for total input power 18dBm.

Fig. 3.Gain and PAE in terms of single-tone input power at

0.85GHz, 1GHz and 1.15GHz

The linearization circuit comprises two branches consisting of the variable attenuator, voltage variable phase shifters and amplifier to adjust IM2 and IM4 signals in amplitude and phase before they are inserted at the carrier transistor input and output over the bandpass filters.

Fig. 2 plots the output power, gain, and power-added efficiency-PAE of the designed broadband Doherty amplifier in terms of the fundamental signal frequency in the case when total input power is 18dBm (3dB below saturation region). It follows that the output power is higher than 35dBm over the frequency range from 0.85GHz to 1.15GHz, gain is equalized around 18dB, whereas PAE is found between 27% and 47%.

PAE and gain of the Doherty amplifier as a function of the input power are shown in Fig. 3, for 0.85GHz, 1GHz and 1.15GHz CW excitation.

In a low power range, the gain observed at excitation frequencies 1.15GHz differs significantly (around 5dB) in comparison with the gain at 0.85GHz and 1GHz. However, if we consider the 18dBm input signal power, the gain is ranged between 17dB to 19dB in the frequency band. The PAE achieved by the designed Doherty amplifier is higher than 50% at saturation. At 3dB back-off from saturation, PAE is between 27% and 45% in the observed frequency range. When a consumption of the circuit for linearization is included into analysis, PAE of linearized Doherty amplifier is lower for around 7% than before linearization.

IV. LINEARIZATION RESULTS

In order to assess the impact of the proposed linearization technique on the designed broadband Doherty amplifier, two-tone test was performed in ADS. Two sinusoidal signals shifted in frequency by ±5 MHz, ±10 MHz, ±20 MHz, or ±40MHz in reference to the center frequency 1GHz were simultaneously driven at the amplifier input.

The third-order intermodulation products, IM3, and fifth-order intermodulation products before and after the linearization, in terms of the frequency interval between the signals are presented in Fig. 4 to Fig. 7 for different power levels of fundamental signals at the amplifier input, 0dBm, 5dBm, 10dBm, and 15dBm.

a) b)

Fig. 4. Intermodulation products of broadband Doherty amplifier for Pin=0dBm before and after linearization: a) Third-; b) Fifth-order

a) b)

Fig. 5. Intermodulation products of broadband Doherty amplifier for Pin =5dBm before and after linearization: a) Third-; b) Fifth-order

a) b)

Fig. 6. Intermodulation products of broadband Doherty amplifier for Pin =10dBm before and after linearization: a) Third-; b) Fifth-order

a) b)

Fig. 7. Intermodulation products of broadband Doherty amplifier for Pin =15dBm before and after linearization: a) Third-; b) Fifth-order

The parameters of the linearization circuits were optimized

to suppress third-order intermodulation products while the fifth-order intermodulation products are required to retain at as low as possible power level.

512

-6 -4 -2 0 2 4 6-8 8

-70

-60

-50

-40

-30

-20

-10

0

10

-80

20

Frequency (MHz)

Fig. 8. Output spectra before linearization (grey line) and after

linearization (black line) for WCDMA signal at 15dBm input power It can be noted that, after the linearization, a significant

reduction of the IM3 products was attained in a considered power range. The figures clearly indicate that the augmentation of the input signal power lessens the grade of IM3 reduction. In the case of 10MHz interval between the signals, the IM3 are suppressed by 20dB for the input power 0dBm, whereas the rise of the input power to 15dBm leads to 7dB reduction of the IM3 products. When frequencies of two tones are shifted by 40MHz, the IM3 products decrease for approximately 10dB at 0dBm input power, 17dB for 5dBm input power and around 10dB for higher power levels. Also, the linearization grade of IM3 is around 6dB in case of 80MHz frequency interval by ranging the input power.

However, the IM5 products are suppressed by a few decibels in the case of lower input power. When input power increases, the IM5 are not linearized, though they are kept at the power level close to the linearized IM3 products. However, at the higher power levels the IM5 products are deteriorated by applying the linearization technique and surpass the level of the suppressed IM3 products by a few decibels.

Moreover, the broadband Doherty amplifier was tested for WCDMA signal at central frequency 1GHz, the spectrum width of 3.84MHz and input power of 15dBm. The results of the analysis-output spectra before and after linearization are shown in Fig. 8. Parameter - Adjacent channel power ratio, ACPR, is improved by 10dB at ±5MHz offset from the carrier.

V. CONCLUSION

This paper presents the analysis of the impact of the linearization technique, which uses the second harmonics of the fundamental signals and fourth-order nonlinear signals at frequencies close to the second harmonics, on suppression of the third-and fifth order intermodulation products in case of broadband two-way Doherty amplifier. The amplifier is designed to operate over the frequency range 0.85-1.15GHz in configuration with lumped element matching circuits. In the applied linearization method, the second harmonics and fourth-order nonlinear signals, which are extracted at the output of the peaking transistor, are adjusted in amplitude and

phase throughout two independent branches and inserted into the input and output of the carrier cell over the bandpass filters. Very good results are achieved in reduction of the third-order nonlinearity of the amplifier observing the two-tone test for a signal power range up to 3dB back-off from saturation. However, the fifth-order intermodulation products go down in a lower power range and slightly ascend for the levels close to the saturation. Furthermore, the satisfactory results in the IM3 linearization are gained in cases when the frequency interval between signals rises. However, it can be noticed that the decrease of the intermodulation products is smaller when the power levels and interval between signals grow up. In addition, test of the amplifier for the broadband WCDMA digitally modulated signal also shows positive effect of the applied linearization method.

ACKNOWLEDGMENT

This work was supported by the Ministry of Education, Science and Technological development of Republic of Serbia, the project number III-44009.

REFERENCES

[1] K. J. Chao, W. J. Kim, J. H. Kim and S. P. Stapleton, “Linearity optimization of a high power Doherty amplifier based on post-distortion compensation”, IEEE Microwave and Wireless Components Letters, vol.15, no.11, pp.748-750, 2005.

[2] K. J. Cho, J. H. Kim and S. P. Stapleton, “A highly efficient Doherty feedforward linear power amplifier for W-CDMA base-station applications”, IEEE Trans., Microwave Theory Tech., vol. 53, no. 1, pp.292-300, 2005.

[3] B. Shin, J. Cha, J. Kim, Y. Y. Woo, J. Yi, B. Kim , “Linear power amplifier based on 3-way Doherty amplifier with predistorter”, IEEE MTT-S Int. Microw. Symp. Digest, pp.2027-2030, 2004.

[4] T. Ogawa, T, Iwasaki, H. Maruyama, K. Horiguchy, M. Nakayama, Y. Ikeda and H. Kurebayashi, “High efficiency feed-forward amplifier using RF predistortion linearizer and the modified Doherty amplifier”, IEEE MTT-S Int. Microw. Symp. Digest, pp.537-540, 2004.

[5] A. Atanasković, N. Maleš-Ilić, B. Milovanović, “The linearization of Doherty amplifier”, Microwave review, No.1, Vol. 14, pp.25-34, September 2008.

[6] Aleksandar Atanasković, Nataša Maleš-Ilić, Bratislav Milovanović: “The linearization of high-efficiency three-way Doherty amplifier”, TELFOR2008, Conference Proceedings on CD, 3.17, Belgrade, Serbia, 25.-27. November, 2008.

[7] A. Atanasković, N. Maleš-Ilić, B. Milovanović: "Linearization of two-way Doherty amplifier", EuMIC 2011, Conference Proceedings on CD, Manchester, UK, EUMA October 10-11, poster01-17, pp.304-307, 2011.

[8] G. Matthei, L. Zoung, and E. M. T. Jones, Microwave Filters, Impedance-Matching Networks, and Coupling Structures. Norwood, MA: Artech House, 1980.

[9] D. Dawson, “Closed-Form Solution for the Design of Optimum Matching Networks”, IEEE Trans. Microw. Theory Tech., vol 57, no.1, pp. 121-129, 2009.

Out

put s

pect

rum

[dB

m]