Você está na página 1de 6

Int. J. Electron. Commun.

(AE) 68 (2014) 10161021

Contents lists available at ScienceDirect

International Journal of Electronics and


Communications (AE)
journal homepage: www.elsevier.com/locate/aeue

A novel miniaturized power amplier with nth harmonic suppression


M. Hayati a,b, , S. Roshani a
a
Electrical Engineering Department, Faculty of Engineering, Razi University, Tagh-E-Bostan, Kermanshah 67149, Iran
b
Computational Intelligence Research Centre, Razi University, Tagh-E-Bostan, Kermanshah 67149, Iran

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, a novel miniaturized power amplier (PA) matched by two proposed low pass lters (LPFs)
Received 16 November 2013 with nth harmonics suppression is presented. In the proposed PA, the LPFs are employed as an output
Accepted 17 May 2014 and input impedance transformer networks, which transform 50  to the desired impedances. In the
proposed PA the conventional output and input matching networks are eliminated, which results in
Keywords: 52% size reduction and 6% power added efciency (PAE) improvement compared with the conventional
Power amplier (PA)
PA. Moreover, using the LPFs at the output and input impressively suppress the unwanted harmonics
Harmonic suppression
(2nd6th) with high level of attenuation. The proposed PA works at the 2.6 GHz, which is suitable for
Low pass lter (LPF)
Matching networks
long term evolution (LTE) applications. The measured and simulated results are in the good agreement,
which conrm the validity of the proposed method.
2014 Elsevier GmbH. All rights reserved.

1. Introduction a large size, and they have several discontinuities between the
narrow and wide lines, which excite high order modes [1].
With the demand of high-speed data communication, the 4th In [8,9], to enhance PAE and suppress unwanted harmonics, l-
generation (4G), LTE system has attracted a lot of attentions. ters have been used in the PA structure, but in these works, there is
For LTE applications, power ampliers (PAs) should be low cost, not any size reduction and OMN is still an indispensable part. In [1]
high integrated and light weight [1]. Nowadays, the growing BPF is used as the OMN, which results in 21% size reduction com-
demand for higher data rates and the increasing number of wire- pared to the conventional PA and suppresses the 2nd harmonic, but
less communications users have resulted into rapidly rising power the miniaturization and harmonic suppression in this work are not
consumption. The energy efciency of base stations should be con- so prominent.
stantly improved in order to reduce the power loss. Signicant In the proposed structure, a miniaturized power amplier inte-
energy saving can be achieved by improving the efciency of the grated by two low pass lters (LPFs) at input and output is proposed.
power amplier (PA) of RF transmitters used in the base station The applied LPF in this structure is based on a proposed lter in
[2]. Different techniques have been proposed to increase the ef- [12]. It works as an IMN and OMN, therefore, the conventional out-
ciency of power ampliers. Envelope elimination and restoration put and input matching networks are eliminated, which results in
(EER) [3], envelope tracking (ET) [4], Doherty ampliers [5,6] and ultra size reduction and high harmonics suppression. Compared
varactor-based dynamic load modulation [7] are the most common with reported works, the proposed PA demonstrates better size
techniques being proposed to enhance the efciency of PAs. reduction (52%) and suppresses the 2nd6th harmonics with high
Power amplier performances signicantly depend on level of attenuation.
their output matching networks (OMNs). Nowadays, exten-
sive researches have been performed on the OMNs to improve
2. Design process
harmonic suppression [8,9], PAs efciency improvement [10],
bandwidth enhancement [11], multiband capability, etc. The OMN
The schematic structures of the conventional PA, is shown in
and input matching network (IMN) in the conventional PA occupy
Fig. 1. In the conventional PA, the output and input matching
networks occupied a large size, and they have several discontinu-
ities between the narrow and wide lines. The OMN and IMN are
Corresponding author at: Electrical Engineering Department, Faculty of Engi-
used to transform 50  to the desired impedances, which shown
neering, Razi University, Tagh-E-Bostan, Kermanshah 67149, Iran.
as ZL and ZS in Fig. 1.
Tel.: +98 9188312041. The proposed PA is designed using a MW6S004N LDMOS tran-
E-mail addresses: mohsen hayati@yahoo.com, hayati@razi.ac.ir (M. Hayati). sistor. The applied transistor is designed for class A or class AB base

http://dx.doi.org/10.1016/j.aeue.2014.05.003
1434-8411/ 2014 Elsevier GmbH. All rights reserved.
M. Hayati, S. Roshani / Int. J. Electron. Commun. (AE) 68 (2014) 10161021 1017

Fig. 1. Schematic diagram of the conventional power amplier, with conventional IMN, OMN.

Fig. 4. Simulated optimum source impedances (ZS ) of the main device at


2.52.7 GHz frequency range.

Fig. 2. Transfer characteristics of the applied device.

station applications with frequency up to 2 GHz. Typical output


power is 4 W, IDQ = 50 mA and Vds = 28 V.
The transfer characteristics of the applied LDMOS transistor
MW6S004N, which performed with the Advanced Design System
(ADS) software (Agilent Technologies, Santa Clara, CA) is shown
in Fig. 2. The plot represents the range of gate voltages and the
corresponding mode of operation with a drain voltage of 28 V. The
amplier is biased in class AB (quiescent drain current of 60 mA and
gate voltage of 2.8 V). Increasing of the quiescent current resulted
in better performances, but caused extra heating, so in the pro-
Fig. 5. Simulated optimum load impedances (ZL ) of the main device at 2.52.7 GHz
posed design the current did not increase very much and there is a frequency range.
good trade of between heating and performances. Fig. 3 shows the
output characteristics of the applied device.
software (Agilent Technologies, Santa Clara, CA) and resulting opti-
mum load/source impedances at 2.52.7 GHz frequency range are
3. Load/source pull
shown in Figs. 4 and 5, while these values of the optimum load
and source impedances are listed in Table 1. The optimum load and
The rst step to design the PA is to perform load/source pull
source impedances at the 2.6 GHz (desired frequency) are 1214j 
simulation to nd the optimum load/source impedances at the
(ZL ) and 7.512j  (ZS ), respectively.
desired frequency (2.6 GHz). The load/source-pull simulations for
The PA parameters have been inferred from the dc IV curves
the PA are performed with the Advanced Design System (ADS)
of the device (Figs. 2 and 3), and load/source pull analysis
(Figs. 4 and 5). The values of the design parameters and their sym-
bols are listed in Table 2.

4. Proposed PA design and implementation

The schematic diagram of the proposed PA is shown in Fig. 6. As


mentioned above the LPFs have a signicant role in the proposed

Table 1
The optimum impedance values from load/source pull simulation in desired
frequency.

Frequency 2.5 GHz 2.6 GHz 2.7 GHz

Optimum load impedance (ZL ) 1112j  1214j  1116j 


Optimum source impedance (ZS ) 813j  7.512j  711j 
Fig. 3. Output characteristics of the applied device.
1018 M. Hayati, S. Roshani / Int. J. Electron. Commun. (AE) 68 (2014) 10161021

Table 2
Power amplier DC parameters.

Parameter Symbol Value

DC bias current IDC 60 mA


DC supply voltage VS 28 V
DC gate voltage VGG 2.8 V
Optimum load impedance ZL 1214j 
Optimum source impedance ZS 7.512j 

structure, which used as an impedance transformer. Therefore,


based on the applied transistor it should transfer the 50  to the
desired impedance.

Fig. 8. The simulated and measured S-parameters of the proposed LPF.


4.1. Low pass lter

The applied LPF is based on a stepped impedance resonator, The applied LPF passes the fundamental frequency (2.6 GHz)
which is used in [12]. The LPF is fabricated on the RT/Duroid 5880 and rejects the unwanted harmonics at the higher frequency. The
substrate (relative permittivity of 2.2, thickness of 0.508 mm and frequency response of the proposed lter illustrates three transmis-
loss tangent of 0.0009), as shown in Fig. 7. The size of the lter is sion zeros, at 3.7, 4.3, and 6.2 GHz with attenuation levels of 42.38,
only 59.78 mm2 (9.8 mm 6.1 mm = 0.116g 0.0723g). 69.78, and 55.5 dB, respectively. These transmission zeros can
The design process of the lter was discussed in [12], and per- provide the high rejection levels for eliminating the desired har-
formance of applied LPF is briey discussed below. monics in the proposed power divider.
The measurements were carried out on a HP8757A network ana-
lyzer. As shown in Fig. 8, the measured and simulated results are in 4.2. Impedance transformer network
good agreement. The transition band is very sharp, approximately
0.07 GHz from 3.6 to 3.67 GHz with corresponding attenuation lev- Hence, the applied LPF used as the main block of the impedance
els of 3 and 20 dB, respectively. transfer networks in the proposed structure, therefore, the input
LPF as the input impedance transformer network should transfer
50  to the 7.512j  and the output LPF as the output impedance
transformer network should transfer 50  to the 1214j . There-
fore, the low impedance stubs are added to the proposed LPF, to
realize the impedance transferring of the LPF from 50  to desired
impedances.

4.2.1. Analysis of two sections impedance transformer


Two sections matching line is one of the most useful
transmission-line that is used for impedance matching in a lot of
applications like the power ampliers. The two sections line is capa-
ble of achieving ideal impedance matching. Fig. 9 demonstrates the
schematic diagram of the input/output matching network of the
proposed PA.
Fig. 6. Schematic diagram of the proposed power amplier. The input impedance Zin of the two-section line shown in Fig. 10
is obtained as follows:
Rs + jZi2 tan(i2 )
Zx = Zi2 (1)
Zi2 + jRs tan(i2 )
where Rs is the optimum source impedances at the 2.6 GHz. There-
fore, the input impedance Zin is

Zx + jZi1 tan(i1 )
Zin = Zi1 (2)
Zi1 + jZx tan(i1 )
The input impedance is equal to Z0 . Equating Zin to Z0 and solving
(2) for Zx , leads to:
Z0 + jZi1 tan(i1 )
Zx = Zi1 (3)
Zi1 + jZ0 tan(i1 )
from (1) and (3):
Z0 jZi1 tan(i1 ) Rs + jZi2 tan(i2 )
Zi1 = Zi2 (4)
Zi1 jZ0 tan(i1 ) Zi2 + jRs tan(i2 )
from (4):

Zi1 (Z0 jZi1 tan(i1 ))(Zi2 + jRs tan(i2 ))

Fig. 7. The photograph of the fabricated LPF.


= Zi2 (Rs + jZi2 tan(i2 ))(Zi1 jZ0 tan(i1 )) (5)
M. Hayati, S. Roshani / Int. J. Electron. Commun. (AE) 68 (2014) 10161021 1019

Fig. 9. Schematic of the power amplier, with two sections IMN, OMN.

2
Z0 Zo1 RL tan(o2 ) + Z0 Zo2 RL tan(o1 ) = Zo1 Zo2 tan(o1 )
2
+ Zo1 Zo2 tan(o2 ) (11)

To nd the values of the input matching stubs in the conven-


tional PA, assuming normalized value Z0 = 1 , Rs = 0.28  in (6) and
(7), also for further simplication assuming  i1 =  i2 =  i . From (6):
Fig. 10. Two sections IMN. 2 2 2
0.72Zi2 + 0.28Zi1 (tan i ) Zi2 (tan i ) Zi2 (tan i ) = 0 (12)

and from (7):

The real part of Eq. (5) yields: (Zi1 + Zi2 )(Zi1 Zi2 0.28) = 0 (13)
2 from (12) and (13):
Z0 Zi1 Zi2 + Zi1 Rs tan(i1 ) tan(i2 ) = Zi1 Zi2 Rs
3 2 2 2
2
+ Z0 Zi1 Zi2 tan(i1 ) tan(i2 ) (6) f1 (zi2 ) = Zi2 (tan i ) 0.72Zi2 0.078(tan i ) = 0 (14)

The value of Zi2 in (14) is obtained from Fig. 12, which in this
gure, Eq. (14) is plotted for different values of  i .
While the imaginary part of (5) yields:
From Fig. 12, assuming  i = 130 , therefore,
2
Z0 Zi1 Rs tan(i2 ) + Z0 Zi2 Rs tan(i1 ) = Zi1 2
Zi2 tan(i1 ) + Zi1 Zi2 tan(i2 ) Zi2 = 0.682 50 = 34.1  and then Zi1 = 0.41 50 = 20.5 .
Similarly to nd the values of the output matching stubs in the
(7) conventional PA, assuming normalized value Z0 = 1 , RL = 0.37  in
(10) and (11), also for further simplication assuming  o1 =  o2 =  o .
From (10):
On the other hand, the output impedance Zout of the two-section
2 2
line shown in Fig. 11 is obtained as follows: 0.63Zo2 + 0.37Zo1 (tan o ) Zo2 (tan o ) = 0 (15)
RL + jZo2 tan(o2 ) and from (11):
Zy = Zo2 (8)
Zo2 + jRL tan(o2 )
(Zo1 + Zo2 )(Zo1 Zo2 0.37) = 0 (16)
where RL is the optimum load impedances at the 2.6 GHz. Therefore,
From (15) and (16):
the output impedance Zout is
3 2 2 2
f2 (Zo2 ) = Zo2 (tan o ) 0.63Zo2 0.137(tan o ) = 0 (17)
Zy + jZo1 tan(i1 )
Zout = Zo1 (9)
Zo1 + jZy tan(i1 ) The value of Zo2 in (14) is obtained from Fig. 13, which in this
gure, Eq. (17) is plotted for different values of  o .
With the similar process followed for input part, the following
equations are obtained:
2 2
Z0 Zo1 Zo2 + Zo1 RL tan(o1 ) tan(o2 ) = Zo1 Zo2 RL + Z0 Zo1 Zo2

tan(o1 ) tan(o2 ) (10)

Fig. 11. Two sections OMN. Fig. 12. f1 (Zi2 ) for different value of  i .
1020 M. Hayati, S. Roshani / Int. J. Electron. Commun. (AE) 68 (2014) 10161021

Fig. 15. Simulated results of the proposed IMN.


Fig. 13. f2 (Zo2 ) for different value of  o .

From Fig. 13, assuming  o = 120 , therefore, Zo2 = 0.6 50 = 30 ,


so Zo1 = 0.61 50 = 30.5 .
In the input matching network of the proposed PA by replacing
the LPF instead of one stubs (Zi1 ,  i1 ), the values of the another stub
Zi2 ,  i2 and Zi1 ,  i1 are obtained from (6) and (7) as follows:

Zi1 = 32.4 ,  i1 = 87 , Zi2 = 8.3 ,  i2 = 23 , which these are the


dimensions of the proposed input matching network, that is
shown in Fig. 15.

On the other hand, in the output matching network of the pro-


posed PA, replacing the LPF instead of one stub (Zo2 ,  o2 ), the values Fig. 16. Schematic of the (a) conventional power amplier, with IMN, OMN and (b)
of the another stub (Zo2 ,  o2 and Zo1 ,  o1 ) are obtained from (10) proposed PA integrated with two LPFs.
and (11) as follow:

4.3. PA implementation and measurements


Zo1 = 33.84 ,  o1 = 88 , Zo2 = 8.4 ,  o2 = 22 , which these are the
dimensions of the proposed output matching network, that is
The structure of the conventional PA and the proposed PA inte-
shown in Fig. 14.
grated with two LPFs are shown in Fig. 16.
In the conventional PA, IMN and OMN occupy large size and
As seen in Fig. 14, at 2.6 GHz the output impedance trans- in the proposed PA applied LPFs lead to ultra size reduction and
former network transfer the 50  to the 1214j  (ZL ), which this efciency improvement of the proposed PA.
impedance is the desired load impedance that is obtained previ- The fabricated power amplier on a RT/Duroid 5880 substrate
ously by load pull analysis. (relative permittivity of 2.2, thickness of 0.508 mm, and loss tangent
As seen in Fig. 15, at 2.6 GHz the input impedance trans- of 0.0009) is shown Fig. 17. The measurements were carried out on a
former network transfer the 50  to the 7.512j  (Zin ), which HP8757A network analyzer, and simulations were performed with
this impedance is the desired source impedance that is obtained the Advanced Design System (ADS) software (Agilent Technologies,
previously by source pull analysis (ZS ). Santa Clara, CA).

Fig. 14. Simulated results of the proposed OMN. Fig. 17. The photograph of the fabricated PA.
M. Hayati, S. Roshani / Int. J. Electron. Commun. (AE) 68 (2014) 10161021 1021

Table 3
Performance summary of the proposed power divider and previous works.

Ref. PAE Size reduction Harmonic suppression (dB)

2nd 3rd 4th 5th 6th

[1] 46.3% 21% 35


[8] 45% 45 45
[9] 61.4% 33 15 20
This work 47% 52% 85 99 95 70 96

used in the large signal measurement. The measured and simulated


results of the fabricated circuits using LPFs and the conventional
matching method are demonstrated in Fig. 19. As shown in Fig. 19,
Fig. 18. Measured and simulated S-parameters of the proposed PA.
the maximum PAE at 2.6 GHz is 47%, which shows 6% improvement,
compared to the conventional PA and the gain of the proposed PA
is about 15 dB, which conrms the small signal gain.
The comparison of the proposed PA with previous approaches
matched by lters in output is summarized in Table 3. The results
show excellent dimension reduction with superior harmonics sup-
pressions compared to the reported works. It should be mentioned
that the PAs in [9] is class F thus the PAE is higher.

5. Conclusion

A power amplier (PA) with a novel technique for size reduction


and nth harmonics suppression has been proposed and imple-
mented. The proposed PA is integrated by two low pass lters
(LPFs). The applied LPFs are used to suppress the harmonics while
work as input and output matching networks. The proposed PA
suppresses 2nd6th harmonics with high level of attenuation. In
the proposed method, the conventional input and output match-
ing networks are eliminated, which result in 52% size reduction
as compared with the conventional PA. For verication, the pro-
Fig. 19. Output power, gain and PAE of the proposed and the conventional PA at posed power amplier is fabricated and tested. A good agreement
2.6 GHz. between the measured and simulated results is achieved.

As seen in Fig. 16, the size of the conventional and proposed References
PA is about 2070 mm2 (115 mm 18 mm = 1.136g 0.21g) and
990 mm2 (55 mm 18 mm = 0.65g 0.21g), respectively, which [1] Li YCh, Wu KCh, Xue Q. Power amplier integrated with bandpass lter for
long term evolution application. IEEE Microw Wireless Compon Lett 2013;23:
presents 52% size reduction in the PA design. 4246.
[2] Ramzi D, Ghannouchi FM. Digital Doherty amplier with enhanced efciency
4.4. Measurement results and extended range. IEEE Trans Microw Theory Tech 2011;59:2898909.
[3] Raab F. Intermodulation distortion in Kahn-technique transmitters. IEEE Trans
Microw Theory Tech 1996;44:22738.
The proposed power amplier is characterized by small-signal [4] Raab F, Asbeck P, Cripps S, Kenington P, Popovic Z, Pothecary N, et al. Power
and large-signal measurements to evaluate its performance. ampliers and transmitters for RF and microwave. IEEE Trans Microw Theory
Tech 2002;50:81426.
[5] Doherty W. A new high efciency power amplier for modulated waves. Proc
4.4.1. Small-signal measurements Inst Radio Engrs 1936;24:116382.
The proposed PA is characterized in small signal conditions [6] Hayati M, Roshani S. A broadband Doherty power amplier with harmonic
suppression. AEU-Int J Electron Commun 2014;68:40612.
(S-parameters) to verify its frequency behavior. A drain bias of
[7] Nemati H, Fager C, Gustavsson U, Jos R, Zirath H. Design of varactor-based tun-
VDD = 28 V is used. The amplier is biased in class AB (quiescent able matching networks for dynamic load modulation of high power ampliers.
drain current of 60 mA and gate voltage of 2.8 V). The simulated IEEE Trans Microw Theory Tech 2009;57:11108.
and measured S-parameters are depicted in Fig. 18, which show the [8] Wang Z, Park C. Novel substrate integrated waveguide (SIW)-based power
amplier using SIW-based lter to suppress up to the fourth harmonic. In:
good agreements between simulated and measured results. As seen Asia-Pacic Microw Conf. 2012. p. 8302.
in Fig. 18, the small signal gain is 15 dB at the working frequency [9] Wang Z, Gao S, Nasri S, Park C. High power added efciency power amplier
(2.6 GHz). The proposed PA suppresses the 2nd6th harmonics with with harmonic controlled by UWB lter with notched band at 6.42 GHz. In:
Wireless Microw Tech Conf. 2011. p. 14.
high level of attenuation. The measured result shows that, the inser- [10] Sigg A, Heck S, Brackle A, Berroth M. High efciency GaN current-mode class-D
tion loss of the 2nd, 3rd, 4th, 5th and 6th harmonics is better than amplier at 2.6 GHz using pure differential transmission line lters. Electron
85, 99, 95, 70 and 96 dB, respectively. Lett 2013;49:479.
[11] Liao HY, Pan MW, Chiou HK. Fully-integrated CMOS class-E power amplier
using broadband and low-loss 1:4 transmission-line transformer. Electron Lett
4.4.2. Large-signal measurements 2010;46:14901.
Large-signal continuous wave (CW) measurements are per- [12] Hayati M, Roshani S, Roshani S, Shama F. A novel miniaturized Wilkinson
power divider with nth harmonic suppression. J Electromagn Waves Appl
formed to evaluate the power amplier under steady-state
2013;27:72635.
conditions. The biasing, as same as the small signal measurement is