Effects of Envelope Tracking Technique on an L-band Power Amplifier

Transcription

1 Effects of Envelope Tracking Technique on an L-band Power Amplifier Elisa Cipriani, Paolo Colantonio, Franco Giannini, Rocco Giofrè, Luca Piazzon Electronic Engineering Department, University of Roma Tor Vergata Via del Politecnico 1, 133 Roma Italy Abstract This contribution presents the application of Envelope Tracking technique on a LDMOS nd harmonic tuned PA, designed to operate at.14 GHz. Accounting for the input dynamic signal, optimum drain and gate bias voltages are derived starting from a simplified model of the active device. Experimental results give an increase up to 4% in average efficiency, when bias tracking is applied. The application of Envelope Tracking does not impact on output power and gain performance, in the selected back-off region, implying also a reduction in the spectral regrowth when complex signals are involved. Index Terms Envelope Tracking, LDMOS, High efficiency. I. INTRODUCTION The wide variety of services provided by new wireless communication systems, requires non constant envelope modulation schemes (WCDMA, OFDM, ) in order to increase channel efficiency and reach a high data rate transmission [1]. Obviously, these types of signals require linear amplification to avoid nonlinear distortion and to preserve the information contained in the envelope modulated input signal. On the other side, wireless communication systems are required to be efficient, in order to minimize size, cost, heating and saving battery life for portable devices. During past years, linearity and efficiency have been object of a suitable trade-off by the designer, since often a high efficiency is only attainable at the expense of linear response of the power amplifier. Moreover, even for high efficiency PAs (like Class F, Class E, etc.), the efficiency is maximized only at the compression point or in saturation regime. As a consequence, if considering an input signal with a typical peak to average power ratio (PAPR) between 6 and 1 db, the resulting value of the average efficiency dramatically decreases. Recently, a large effort has being devoted in the investigation and development of high efficient-high linear transmitter architectures. Doherty [], Chireix [3], Envelope Elimination and Restoration (EER) [4]-[6], Envelope Tracking (ET) [7]-[11] etc. have been proposed and are still studied [13]. In this paper we focused on the application of Envelope Tracking technique on a nd harmonic tuned LDMOS PA at.14 GHz [14]. The operating principle of this architecture, whose scheme is illustrated in Fig. 1, is to dynamically change the polarization of the power amplifier according to the input signal, in order to force the amplifier to operate always in the high efficiency region. Starting from a simplified model of a general active device, drain and gate bias laws, derived in [] [1], are summarized. Then, the validity of the theoretical derivation is experimentally demonstrated through nonlinear measurements. Finally, system simulations are performed on the amplifier, in order to evaluate the impact of Envelope Tracking on the spectral regrowth, when complex signals are applied at the input of the PA. Input Signal Splitter Fig. 1. Envelope tracking architecture. II. THEORETICAL APPROACH In Fig., the simplified device model adopted, is reported []. The active device is modelled as a voltage controlled current source (g m V gs ) having an input resistance R in and an output resistance R ds. Knee voltage V k is considered to be constant; input and output capacitances are also added (C gs and C ds ), but they do not have influence on the DC analysis. Output load is represented by the parallel connection of L and R L. Fig.. Considered device model. Delay τ Envelope Signal DC/DC Converter Output Signal Output characteristics of such idealized device are reported in Fig. 3, where the operating limits of the active device are PA //$6. IEEE 348

2 highlighted: maximum current I MAX, built-in voltage V bi, and knee voltage V k. The contribution of resistance R ds in the slope of the characteristics is also stressed. In this figure, a generic load line is showed, crossing the two points (, ) and (I MAX, V k ). Hence, its angular coefficient (μ) is given by: I μ = V DC DD I V MAX k and it is coincident to the quantity -1/R L only for a purely Class A bias condition. (1) in which I ac is the dynamic drain current component. Small characters subscripts herein indicate a quantity dependent on the dynamic bias point considered. DC and AC current components generated by the active device for a generic Class AB bias at the EDC are expressed by: VDD Vk IDC = gm ( VGG Vp ) + Rds VDD Vk IAC = gm VGS Rds (4) i d i d I MAX V bi I MAX1 V bi I MAX A μ V gs= V GG1 V p 1 B V p V GG1 V GG Fig. 3. Ouput IV characteristic of the active device. The transconductance g m is assumed to be constant and it is expressed by: g m IMAX = V V bi V k p Maximum output power condition, which is achieved for the maximum value of the magnitude of envelope input signal, is considered as the starting point to determine bias laws. In order to preserve linearity, it is useful to chose this operating point, herein called End Dynamic Condition (EDC) as the 1 db compression point of the amplifier. In order to properly apply the Envelope Tracking approach, two conditions have to be fulfilled: 1) the conduction angle of the amplifier must be kept constant; ) for each value of input power, the maximum operating current is reached in correspondence of the knee voltage V k, that is, the angular coefficient μ has to be kept constant. These conditions, illustrated in Fig. 4, are analytically expressed by: Idc = ξ = const Imax I = μ V V ( ) ac k dd v ds () (3) V k Fig. 4. Load lines of the active device. Bias laws are obtained after some manipulation of expressions (1), (), (3) and (4), and can be expressed as functions of a dynamic variable x and EDC bias condition []: ( ) ( ) ξ ( 1 μ Rds ) 1 ( ξ ) ( 1 ξ) ( 1 μ R ) 1 Vdd x = x VDD Vk + Vk Vgg, x = Vgs, x ds ( ξ ) III. CW SIMULATIONS AND MEASUREMENTS The derived ET approach is applied on the nd harmonic tuned PA discussed in [14],[1], whose internal parameters, extracted from the DC-IV curves, are summarized in Table I. TABLE I DEVICE PARAMETERS Parameter Symbol Value Maximum current I max.6 A Pinch-off voltage V p 7 V Built-in voltage V bi.6 V Knee voltage V k 7 V Drain-Source resistance R ds 8 Ω Then, equations given in () are implemented in the Advanced Design System CAD environment, so that the gate and drain bias voltages were varied according to the input power, which is related to the variable x by a simple expression: v ds () P ( x) = log( x ) + P (6) in in,1 dbg. c. 349

3 The 1 db gain compression point of the PA at the nominal fixed bias ( =3V and =7mA, corresponding to V GG =3.V) corresponds to 8dBm input power, 4.8dBm output power, 1.8dB power gain and 4%efficiecy (PAE=1 %). This point has been chosen as EDC (x=1). In Fig. simulated dynamic load lines at 1 db gain compression for different bias conditions are depicted. It is evident that each load line has almost the same slope and does not overstep the knee voltage: conditions expressed by (3) are satisfied and the ET technique is properly applied. supply. At nominal supply voltage, measured 1 db gain compression point occurred at 8dBm of input power, in correspondence of 1.4dB of power gain, slightly less than the simulated one (1.8dB). 3, I ds (A) 3,,, 1, Pin= 8 dbm Pin= dbm Pin= dbm Pin= 19 dbm 1,,, 3 V ds Fig.. Dynamic load line of the PA, applying ET, for several values of the input power. The comparison of PA s simulated output performance with and without ET are shown in Fig. 6. Output power and gain are maintained closed to the ones of the simple PA until 13 db of input back-off (IBO): this identifies an operating region in which the value of g m can be assumed constant and thus the approximations made in section are verified. The application of ET technique arises the average efficiency from 33.6% to 49.% throughout the considered IBO (1 db). Output power (dbm) & Gain (db) Gain with ET Pout with ET Gain without ET Pout without ET Efficiency with ET Efficiency without ET Fig. 6. Comparison of simulated output power, gain and efficiency with and without ET applied to the nd harmonic tuned PA. Fig. 7 shows the setup used for power measurements on the amplifier. Bias laws are implemented via software by a personal computer which controls RF source and DC power Efficiency (%) Fig. 7. Implemented setup to perform CW measurements. Fig. 8 is a plot of simulated and experimentally set, V GG bias laws versus input power. As can be noted, the behaviours are overlapped while the little shift between the V GG values is due to the different pinch-off voltage value shown by the actual device. Moreover, as evident from such figure, the difference between measured and theoretical values of is lower than % until 1dB of IBO, thus confirming the validity of the simplified model used to derive voltage bias laws. V GG (ma) ,1 3,,9,8,7, set theoretical V GG set theoretical V GG measured theoretical Fig. 8. Comparison between theoretical and implemented bias laws versus input power. 3

4 Measured performance with and without Envelope Tracking are depicted in Fig. 9. As evident, output power and gain are not affected with respect to operation in normal conditions until 1dB of IBO, while the average efficiency, calculated over the measured IBO, arises from 3.6%, for the PA with fixed bias, to 43.6%, when ET is applied, resulting in a 4% improvement. Output power (dbm) & Gain (db) 6 6 Pout without ET Pout with ET Gain without ET 4 Gain with ET Efficiency with ET Efficiency without ET Fig. 9. Comparison between measured output power, gain and efficiency with and without ET applied on the nd harmonic tuned PA. IV. SYSTEMS SIMULATION In order to evaluate the effects of the ET technique when complex signals are managed by the PA, system simulations have been performed. For this purpose, a source generating a signal based on a 18QAM constellation with the carrier frequency at.14ghz and a rate of Msimple/sec has been selected. The behavioural model of the PA is based on the AM/AM and AM/PM distortions, reported in Fig. (a) and (b), respectively. For both cases, with and without ET, the behaviour of the distortions has been extracted through CW simulations. In the considered 1dB of IBO, the AM/AM and AM/PM distortions are reduced when ET is applied to the PA. The former represents the input power range in which the constant transconductance, assumed for the theoretical model of the active device, can be considered as a good estimation of the actual one. On the other hand, for higher IBO, the amount of distortion introduced by the PA biased with ET increases, in particular considering the AM/AM distortion. This effect is the expected consequence of the transconductance reduction, which happens when the bias condition of the device becomes very close to the ohmic region. In order to evaluate how ET technique could affect the spectral regrowth of the output signal, system simulations have been performed, selecting an average input power of 18dBm, dbm and dbm, respectively. The related Probability Density Function (PDF) of the three cases is reported in Fig. (c). From this figure is also evident that the End Dynamic Condition of the PA (peak input power of 8dBm) is reached only in the case of dbm of average Efficiency (%) power. For the other cases, in fact, the PDF of the signal reaches a null value before 8dBm of input power. AM/AM (db) AM/PM ( ) PDF (%) (a) 1 (b) ,6 1,4 (c) 1, 18dBm average power 1, dbm average power,8 dbm average power,6,4,, Fig.. AM/AM (a) and AM/PM (b) simulated distortions of the PA with and without ET applied. (c) Probability density function of the input signals for different values of the average power. Fig. 11, Fig. 1 and Fig. 13 report the simulated output spectrums of the PA with and without ET for an average input power of 18dBm, dbm and dbm respectively Average input power = 18dBm Fig. 11. Simulated PA output spectrum for an input average power of 18dBm. When 18dBm is selected as average input power (see Fig. 11), the spectral regrowth in the adjacent channels is lower, if the ET is not applied. The PA, in fact, is managed in a deep back off condition. Consequently, the AM/AM distortion when ET is applied is greater than the case of constant bias point. Increasing the average input power up to dbm, the gap in the spectral regrowth between the cases with and without ET is reduced (see Fig. 1). Moreover, reaching dbm, the spectral regrowth for constant bias condition becomes greater 31

5 than the other case, i.e. when the ET is applied (see Fig. 13). In fact, for this level of average power, the EDC of the PA is reached, thus maximising the linearity for the AM/AM and AM/PM behaviour in the case with ET applied. Consequently, it seems that the ET technique allows to improve the performance also in terms of linearity, if the level of the average power is properly selected Average input power = dbm Fig. 1. Simulated PA output spectrum for an input average power of dbm Average input power = dbm Fig. 13. Simulated PA output spectrum for an input average power of dbm. V. CONCLUSIONS In this paper, bias laws for Envelope Tracking technique have been derived and also applied to a nd harmonic tuned PA based on LDMOS device. The benefits of such approach have been demonstrated, in terms of average efficiency and linearity performance. In particular, a 4% of increase in average efficiency has been achieved throughout a 1 db of Input back-off. Moreover, system simulation results shown a reduction in the output spectral regrowth of about db with respect to the case of constant bias condition, maintaining the same saturated output power and gain level. ACKNOWLEDGMENT Authors wish to gratefully acknowledge the support of Freescale Semiconductors for providing the active device and its non linear model. REFERENCES [1] J. Govil, 4G Mobile Communication Systems: Turns, Trends and Transition, in Proc. ICCIT 7, 7, pp [] W. H. Doherty, A New High Efficiency Power Amplifier for Modulates Waves, in Proc. IRE 36, 1936, pp [3] H. Chireix, High Power Outphasing Modulation, in Proc. IRE 3, 193, pp [4] L.R. Kahn, Single-Sideband Transmission by Envelope Elimination and Restoration, in Proc. IRE, 19, pp [] Narisi Wang, Xinli Peng. V. Yousefzadeh, D. Maksimovic, S. Pajic, Z. Popovic, Linearity of X-band class-e power amplifiers in EER operation, in IEEE Transactions on Microwave Theory and Techniques, vol. 3, no. 3, part, pp. 96-1, March. [6] W. Feipeng, D.F. Kimball, J.D. Popp, A.H. Yang, D.Y. Lie, P.M. Asbeck, L.E. Larson, An Improved Power-Added Efficiency 19-dBm Hybrid Envelope Elimination and Restoration Power Amplifier for 8.11g WLAN Applications, in IEEE Transactions on Microwave Theory and Techniques, vol. 4, pp , Dec. 6. [7] C. Buoli, A. Abbiati, D. Riccardi, Microwave power amplifier with Envelope Controlled Drain Power Supply, in Proc. EuMC 9, 199, pp [8] Feipeng Wang, A.H. Yang, D.F. Kimball, L.E. Larson, P.M. Asbeck, Design of wide-bandwidth envelope-tracking power amplifiers for OFDM applications, IEEE Transactions on Microwave Theory and Techniques, vol 3, no. 4, part 1, pp April. [9] F. Wang, D. F. Kimball, J. D. Popp, A. H. Yang,; D. Y. Lie, P. M. Asbeck, L. E. Larson, An Improved Power-Added Efficiency 19-dBm Hybrid Envelope Elimination and Restoration Power Amplifier for 8.11g WLAN Applications, IEEE Trans. on MTT, vol. 4, n. 1, December 6, pp [] P. Colantonio, F. Giannini, R. Giofrè, L. Piazzon, Bias Relationships for Envelope Tracking Technique, Proceedings of the Integrated Nonlinear Microwave and Millimetre-wave Circuits (INMMiC 8), Malaga, Spain, 4- November 8, pp , DOI.19/INMMIC [11] K. Takahashi, S. Yamanouchi, T. Hirayama, K. Kunihiro, An Envelope Tracking Power Amplifier Using an Adaptive Biased Envelope Amplifier for WCDMA Handsets, in Proc. RFIC 8, 8, pp [1] T. Rautio, S. Hietakangas, T. Rahkonen, Development Environment for EER and Envelope Tracking RF transmitters, IEEE Norchip Conference, November 6, pp [13] F. H. Raab, P. Asbeck, S. Cripps, P. B. Kenington, Z. B. Popovic, N. Pothecary, J. F. Sevic, and N. O. Sokal, Power Amplifiers and Transmitters for RF and Microwave, IEEE Trans. on MTT, vol., n. 3, March, pp [14] E. Cipriani, P. Colantonio, F. Giannini, R. Giofrè, A second harmonic tuned power amplifier in LDMOS technology, XVII International Conference on Microwaves, Radar and Wireless Communications MIKON 8, Poland, Wroclaw, 19-1 May, Conference Proceedings Volume pp ISBN [1] E. Cipriani, P. Colantonio, F. Giannini, R. Giofrè, L. Piazzon, Envelope Tracking Technique Applied on a W nd Harmonic Tuned Power Amplifier at.14 GHz, Proceedings of the 39th European Microwave Conference, EUMC 9, Rome, Italy, Sept. 9, pp