A wideband GaN HEMT power amplifier based ... - Wiley Online Library
A WIDEBAND GaN HEMT POWER AMPLIFIER BASED ON THE DUAL-FED DISTRIBUTED STRUCTURE FOR WiMAX APPLICATIONS Yong-Sub Lee, Mun-Woo Lee, and Yoon-Ha Jeong Department of Electronic and Electrical Engineering, Pohang University of Science and Technology, San 31, Hyoja-Dong, Nam-Gu, Pohang, Gyungbuk 790 –784, Republic of Korea; Corresponding author: [email protected] Received 26 June 2008 ABSTRACT: In this article, we propose a wideband GaN HEMT power amplifier (PA) based on the dual-fed distributed structure for 2.6 GHz WiMAX applications. For a continuous wave, the distributed PA shows the wideband performance compared with the conventional balanced PA. When the distributed PA is optimized by controlling the gate bias voltages, the wideband performance over 150 MHz is achieved for a WiMAX signal with a PAR of 9.47 dB and the signal bandwidth of 28 MHz. © 2008 Wiley Periodicals, Inc. Microwave Opt Technol Lett 51: 574 –577, 2009; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.24090 Key words: gallium nitride (GaN); distributed power amplifier (DPA); wideband 1. INTRODUCTION
Modern wireless communication systems should deliver high data rate signals with large signal bandwidth and high peak-to-average ratio (PAR). The mobile world interoperability for microwave access (WiMAX) signal such as IEEE 802.16 –2004 deals with up to 15 Mbps of data rate and 28 MHz of signal bandwidth. Therefore, the power amplifiers (PAs) have to operate large back-off region with broadband performance to amplify linearly the signals with a PAR of about 10 dB [1, 2]. For the wideband performance, the distributed amplifications have been used because of their flat gain, linear phase, and low return losses. However, their low output power and low efficiency are main disadvantages [3– 6]. To overcome severe limitations, the dual-fed distributed PAs, the tapered drain line distributed PAs, and so on have been suggested [5– 8]. Their performances are suited for the wideband gain amplifiers or low power driver amplifiers. On the one side, the distributed structure has been applied to the Doherty amplifier because the distributed PAs do not need the N-way power divider and combiner [9]. For high-power amplifications over 2 GHz, gallium nitride (GaN) high electron mobility transistors (GaN HEMTs) have recently developed with several advantages of high-power densities, high electron saturation velocity, high operating temperature, and high cutoff frequency [10, 11]. In this article, we propose the wideband dual-fed distributed GaN HEMT PA for 2.6 GHz WiMAX applications. From the measured results for a continuous wave (CW) show that the dual-fed distributed PA shows wideband characteristics with flat gain and efficiency compared with the conventional balanced PA. When the distributed PA is optimized by the gate bias voltages, the wideband performance is achieved for a WiMAX signal.
method where power combining is performed directly at the device level without the need for N-way combiners [3– 8]. The conventional distributed PA provides ultra-wide bandwidth, but has low efficiency and some of the transistors are idle [4]. As shown in Figure 1(b), the dual-fed distributed amplifier incorporates a pair of transmission lines (TLs) that are periodically coupled by the transistors, a hybrid to drive both ends of the gate (input) line, and another hybrid to combine waves appearing at the ends of the drain (output) line [5–7]. Previous literature has investigated the gain and efficiency of the dual-fed distributed PA compared with the conventional distributed PA [5–7]. For the distributed PA, the magnitudes of the FET drain voltage and current are equal when the FETs are located by a /2 at the center frequency. Therefore, the electrical spacing between the FETs is /2, and the short-circuit terminations of the gate and drain lines are /4 from the nth FET [7]. The line impedances of the gate and drain lines are both equal to 50 ⍀. 3. IMPLEMENTATION AND EXPERIMENTAL RESULTS
For the measurement, the two-way balanced and distributed PAs have been designed and implemented using a push-pull RFHIC RT440 GaN HEMT with the P1dB of 20 W, which avoids the circuit complexity and large size. The balanced GaN HEMT PA has employed a 3-dB hybrid coupler of Anaren 1 ⫻ 603 as the power divider and combiner. However, the signal separation and combination in the distributed PA was simply implemented with TLs (/2 and /4) without the two-way power divider and combiner as shown in Figure 2. The shunt capacitors of Cd were inserted to optimize the performance and were 3.6 pF [9]. Figure 3 shows the photographs of the fabricated balanced and distributed
PA1 RFout
RFin N-way Power Divider
PAn
(a) RFout
RFin
λ
2
λ
2
λ
λ
2. WIDEBAND DUAL-FED DISTRIBUTED POWER AMPLIFIER
574
PA1
PA2
4
Figure 1(a) shows the schematics of the N-way balanced PA. It is necessary for the N-way balanced PA to combine the input and output powers of the several transistors using the N-way power divider and combiner. However, the distributed amplification is a
N-way Power Combiner
PA2
4
PAn
(b) Figure 1 Schematics of (a) the N-way balanced PA and (b) the N-way dual-fed distributed PA
MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 2, February 2009
DOI 10.1002/mop
60
50
RFin
λ
PA1
2
2
λ
λ
4
4
Cd
PA2
45
Gain [dB], Output Power [dBm]
λ
Cd
Figure 2 Schematic of the two-way dual-fed distributed PA
Solid: Balanced PA Hollow: Distributed PA 50
40
Pout
35
40
PAE
30
30 25 20
20 15
10 10
Gain 5
0 10
PAs. To reduce the memory effects, the drain bias lines in both PAs incorporated a quarter-wavelength bias line and several decoupling capacitors. Each matching circuit was optimized using shunt capacitors. The balanced and distributed PAs had the quiescent current (IDQ) of 240 mA with the drain bias voltage (Vdd) of 28 V. Figure 4 shows the measured output power, gain, and poweradded efficiency (PAE) characteristics of the balanced and distributed PAs according to input power level for a CW of 2.6 GHz. From the results, both amplifier shows similar performance beside of the lower gain of the distributed PA. Figure 5 depicts the measured gain and PAE performances of the balanced and distributed PAs according to the change of operating frequency at an average output power of 33 dBm for a CW. For the balanced PA, the good gain and PAE performances are achieved around the center frequency, but the performance is degraded at the lowermost and uppermost frequencies. However, the distributed PA shows flat gain and PAE performances in wider frequency range than the distributed PA. The gain flatness is less
Vdd2
Vgs2 RFin PA2
Power Divider
Input Matching
Output Matching
PAE [%]
RFout
Power Combiner
PA1
12
14
16
18
20
22
24
26
28
30
32
34
Input Power [dBm]
Figure 4 Measured output power, gain, and PAE characteristics of the balanced and distributed PAs according to input power level for a continuous wave of 2.6 GHz. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
than 0.5 dB up to the bandwidth of 180 MHz. Therefore, the distributed PA shows wideband performance compared with the balanced PA. Figure 6 shows the third- and fifth-order intermodulation distortions (IMD3s and IMD5s) of the balanced and distributed PAs for two-tone signals with various tone spacings. The unbalance between the lower and upper components of less than 4 dB is obtained for the two-tone signal with up to 100-MHz tone spacing, which proves that the distributed PA has a considerably reduced memory effect compared with the balanced PA. For the WiMAX signal, the relative constellation error (RCE) instead of an adjacent channel leakage ratio (ACLR) should be measured. The RCE shows the amount of in-band-error like the error vector magnitude (EVM) [2]. The IEEE 802.16 –2004 with 28 MHz signal bandwidth, the burst type of 64-QAM-3/4, and the PAPR of 9.47 dB was used as the WiMAX signal. The linearity specification of the modulated signal is the RCE of ⫺31 dB. Figure 7 shows the measured RCE and PAE characteristics of the distributed PA according to output power level for a WiMAX signal of 2.6 GHz. The distributed PA was optimized at an average output
RFout 20
20
Average Output Power = 33dBm
Solid: Balanced PA Hollow: Distributed PA
Vdd1
18
PAE
18
(a) 4
Vgs2
Vdd2
4
Cd
12
Cd
λ
2
Input Matching
PA2
Output Matching
PA1
14
16
λ
λ
Gain [dB]
λ
16
14
10
Gain
8 12
PAE [%]
Vgs1
6
2
4
10
2
RFin
Vgs1
Vdd1
RFout
8 2.50
2.55
2.60
2.65
0 2.70
Frequency [GHz]
(b) Figure 3 Photographs of (a) the fabricated balanced PA and (b) the fabricated distributed PA. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
DOI 10.1002/mop
Figure 5 Measured gain and PAE performances of the balanced and distributed PAs according to the change of operating frequency at an average output power of 33 dBm for a CW. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 2, February 2009
575
-25
1.5
Balanced PA
IMD3(-) IMD3(+) IMD5(-) IMD5(+)
-30
Distributed PA
IMD3(-) IMD3(+) IMD5(-) IMD5(+)
1.0
-35
Imaginary
IMD [dBc]
0.5 -40
IMD3 -45
-50
IMD5
-55
-0.5
-1.0
-60 0
0.0
20
40
60
80
-1.5 -1.5
100
-1.0
-0.5
0.0
Tone Spacing [MHz]
0.5
1.0
1.5
Real
Figure 6 Measured IMD3 and IMD5 characteristics of the balanced and distributed PAs according to tone spacing for a two-tone signal. [Color figure can be viewed in the online issue, which is available at www. interscience.wiley.com.]
Figure 8 Measured constellation of the distributed PA at an average output power of 33 dBm for a WiMAX signal. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
power of 33 dBm with the optimization of the gate bias voltages. The gate bias voltages of the PA1 and PA2 in the distributed PA were changed to ⫺1.45 (IDQ of 200 mA) and ⫺1.59 V (IDQ of 20 mA), respectively. From the measured results, the distributed PA can deliver an average output power of 35.6 dBm with a PAE of 27.1% at a RCE of ⫺31 dB. Figure 8 shows the measured constellation of the distributed PA at an average output power of 33 dBm. Figure 9 shows the measured RCE, gain, and PAE characteristics of the distributed PA according to the change of operating frequency at an average output power of 33 dBm for a WiMAX signal. Over 150-MHz bandwidth, the RCE of ⫺31 dB was maintained with the gain flatness of less than 0.5 dB and the PAE variation of less than 3%.
applications. For accurate comparison, the two-way balanced PA as well as the two-way distributed PA has been implemented using a push-pull RFHIC RT440 GaN HEMT. Compared with the balanced PA, the distributed PA was implemented using TLs instead of the 3-dB hybrid couplers as the two-way power divider and combiner. For a CW, the distributed PA shows more wideband gain and PAE performances than the balanced PA. After optimization by controlling the gate bias voltage, the distributed PA shows wideband RCE, gain, and PAE characteristics over 150 MHz for a WiMAX signal with the PAR of 9.47 dB and the signal bandwidth of 28 MHz. From the measured results prove that the wideband dual-fed distributed GaN HEMT PA can be a promising solution for a WiMAX applications with high PAR and large signal bandwidth.
4. CONCLUSIONS
ACKNOWLEDGMENTS
In this article, we have proposed a wideband distributed GaN HEMT PA based on the dual-fed structure for 2.6 GHz WiMAX
This work was partially supported by the BK21 program and the National Center for Nanomaterials Technology (NCNT) in Korea.
40
-15
25
-20
20
-15
Average Output Power = 33dBm
35
-20
30
20 15
-30
Linearity Specification
RCE [dB]
PAE [%]
-25
10
15
-25
Gain RCE
10
-30
RCE [dB]
27.1% 25
Gain [dB], PAE [%]
PAE
Linearity Specification
-35
5
-40
0 2.50
-35
5
35.6 dBm
0 28
30
32
34
36
38
2.55
2.60
2.65
-40 2.70
Output Power [dBm]
Frequency [GHz]
Figure 7 Measured RCE and PAE characteristics of the distributed PA according to output power level for a WiMAX signal of 2.6 GHz. [Color figure can be viewed in the online issue, which is available at www. interscience.wiley.com.]
Figure 9 Measured RCE, gain, and PAE characteristics of the distributed PA according to the change of operating frequency at an average output power of 33 dBm for a WiMAX signal. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
576
MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 2, February 2009
DOI 10.1002/mop
REFERENCES
Key words: reflector antenna; spherical reflector; distortion compensation; reflectarray
1. K.R. Santhi and G.S. Kumaran, Migration to 4 G: Mobile IP based solutions, In Proceedings of AICT/ICIW’06, 2007, pp. 76 – 82. 2. J. Moon, J. Kim, I. Kim, J. Kim, and B. Kim, A wideband envelope tracking Doherty amplifier for WiMAX Systems, IEEE Microwave Wireless Compon Lett 18 (2008), 49 –51. 3. T.T.Y. Wong, Fundamentals of distributed amplification, Artech House, Boston, MA, 1993. 4. J.L.B. Walker, Some observations on the design and performance of distributed amplifiers, IEEE Trans Microwave Theory Tech 40 (1992), 164 –168. 5. C.S. Aitchison, N. Bukhari, C. Law, and N. Nazoa-Ruiz, The dual-fed distributed amplifier, IEEE MTT-S Int Microwave Symp Dig, New York, NY (1988), 911–914. 6. C.S. Aitchison, M.N. Bukhari, and O.S.A. Tang, The enhanced power performance of the dual-fed distributed amplifier, In European Microwave Conference, 1989, pp. 439 – 444. 7. K.W. Eccleston, Compact dual-fed distributed power amplifier, IEEE Trans Microwave Theory Tech 53 (2005), 825– 831. 8. P.N. Shastry, S.N. Prasad, and A.S. Ibrahim, Design guidelines for a novel tapered drain line distributed power amplifier, 36rd European Microwave Conference, Manchester, UK, 2006, pp. 1274 –1278. 9. K.J. Cho, W.J. Kim, J.Y. Kim, J.H. Kim, and S.P. Stapleton, N-way distributed Doherty amplifier with an extended efficiency range, IEEE MTT-S Int Microwave Symp, Honolulu, HI, (2007), 1581–1584. 10. Y.S. Lee, M.W. Lee, and Y.H. Jeong, Experimental analysis of GaN HEMT and Si LDMOS in analog predistortion power amplifier for WCDMA applications, Microwave Opt Technol Lett 50 (2007), 393– 396. 11. Y.S. Lee, M.W. Lee, and Y.H. Jeong, Linearity-optimized GaN HEMT Doherty amplifiers using derivative superposition technique for WCDMA applications, Microwave Opt Technol Lett 50 (2008), 701–705. © 2008 Wiley Periodicals, Inc.
A COMPENSATED SPHERICAL REFLECTOR ANTENNA USING SUBREFLECTARRAYS Shenheng Xu and Yahya Rahmat-Samii Department of Electrical Engineering, University of California, Los Angeles, Los Angeles, CA 90095-1594; Corresponding author: [email protected] Received 10 September 2008 ABSTRACT: Spherical reflectors have been used as wide-angle scanning antennas due to the unique geometric properties of the sphere. The inherent spherical aberration, however, causes considerable performance degradation. This article presents a novel sub-reflectarray compensating technique to correct it. A microstrip reflectarray is a planar, inexpensive, passive device with unique capabilities of controlling reflection phases. By exploiting this feasible property, it is employed as an auxiliary phase corrector, and the phase errors caused by spherical aberration are compensated. Two examples, a 35-m spherical reflector antenna at Ka-band and a 1.5-m breadboard spherical reflector, are presented, and simulation results show great performance improvement for both no scan and large scan cases. Because of the difficulty in measuring large reflector antennas at UCLA, it is part of future plan to collaborate with other institutions for the verification of the concept. However, we believe that the simulation results generated in this article should provide acceptable performance prediction of the proposed antenna design. © 2008 Wiley Periodicals, Inc. Microwave Opt Technol Lett 51: 577–582, 2009; Published online in Wiley InterScience (www. interscience.wiley.com). DOI 10.1002/mop.24088
DOI 10.1002/mop
1. INTRODUCTION
Although a parabolic reflector antenna provides perfect focusing performance, it suffers from severe limitations in large scanning applications due to its poor off-focus performance. By contrast, a spherical reflector antenna is well suited because of its perfect angular symmetry. By rotating an optimally located feed system about the center of curvature of the spherical reflector and illuminating different portion of the reflector aperture, acceptable performances can be achieved across a large angular region [1, 2]. The inherent spherical aberration, however, causes tremendous performance degradation. To eliminate the spherical phase aberration, complex feed systems or auxiliary phase correctors, such as phased line sources [3, 4], planar array feeds [5, 6], and Gregorian reflectors [7], are usually employed in large spherical reflector antennas. With the recent needs of spaceborne remote sensing projects, more stringent demands are imposed on the spherical reflector antenna designs. For instance [6], the second generation of a geostationary orbiting radar requires an antenna system operating at Ka-band (35.6 GHz) with directivity of 77 dB, a half power beamwidth of 0.02°, and an off-axis scan up to ⫾4° (⫾200 beamwidths). Therefore, a 35.5-m spherical reflector antenna with an effective aperture of 28 m was proposed to satisfy the great challenge of producing an unconventionally narrow beam at ultra large beamwidth scan angles, and a 271-element hexagonal planar array feed was used to compensate for the spherical aberration. However, the fabrication of such large array feed at the high operating frequency imposes another serious challenge, considering the fact that both amplitude and phase excitations are necessary. Moreover, dielectric losses are typically considerable in such active devices. A Gregorian corrector [7] can overcome these drawbacks, but it encounters great difficulty in achieving low sidelobe levels because of the resulted high inverse illumination taper across the reflector aperture. The overall size of the antenna system is larger as well, because the Gregorian corrector must be placed outside the caustic region of the spherical reflector. A microstrip reflectarray is a planar, low-profile, inexpensive, easy-to-build, passive device which combines some best features of reflectors and microstrip arrays [8]. An individual reflectarray element is capable of reflecting the incident field with predesigned phase. By controlling the reflection phase of the reflectarray, a specific wavefront can be formed to generate a pencil beam, to electronically steer a pencil beam, or to compensate for reflector surface distortions. Different element designs are readily available in literature. Microstrip patches of variable size [9] and microstrip patches with stubs of variable length [10] are widely used due to the ease of implementation. Reflectarrays have attracted more attention in antenna design communities, especially for spaceborne applications, because of those aforementioned favorable features. In this article, a novel sub-reflectarray compensating technique [11, 12] is proposed to compensate for the spherical aberration. By exploiting the phase controlling capability, a microstrip reflectarray is employed as an auxiliary phase corrector to eliminate the spherical phase errors. The implementation of this novel technique is presented in Section 2. The concept of sub-reflectarray compensation is introduced, and some design guidelines are discussed. In Section 3, the 35.5-m spherical reflector antenna operating at Ka-band is successfully compensated using a sub-reflectarray of 0.3-m in diameter. The compensation of a 1.5-m breadboard spherical reflector antenna is presented as well for future experimental
MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 2, February 2009
577
Modern wireless communication systems should deliver high data rate signals with large signal bandwidth and high peak-to-average ratio (PAR). The mobile world interoperability for microwave access (WiMAX) signal such as IEEE 802.16 –2004 deals with up to 15 Mbps of data rate and 28 MHz of signal bandwidth. Therefore, the power amplifiers (PAs) have to operate large back-off region with broadband performance to amplify linearly the signals with a PAR of about 10 dB [1, 2]. For the wideband performance, the distributed amplifications have been used because of their flat gain, linear phase, and low return losses. However, their low output power and low efficiency are main disadvantages [3– 6]. To overcome severe limitations, the dual-fed distributed PAs, the tapered drain line distributed PAs, and so on have been suggested [5– 8]. Their performances are suited for the wideband gain amplifiers or low power driver amplifiers. On the one side, the distributed structure has been applied to the Doherty amplifier because the distributed PAs do not need the N-way power divider and combiner [9]. For high-power amplifications over 2 GHz, gallium nitride (GaN) high electron mobility transistors (GaN HEMTs) have recently developed with several advantages of high-power densities, high electron saturation velocity, high operating temperature, and high cutoff frequency [10, 11]. In this article, we propose the wideband dual-fed distributed GaN HEMT PA for 2.6 GHz WiMAX applications. From the measured results for a continuous wave (CW) show that the dual-fed distributed PA shows wideband characteristics with flat gain and efficiency compared with the conventional balanced PA. When the distributed PA is optimized by the gate bias voltages, the wideband performance is achieved for a WiMAX signal.
method where power combining is performed directly at the device level without the need for N-way combiners [3– 8]. The conventional distributed PA provides ultra-wide bandwidth, but has low efficiency and some of the transistors are idle [4]. As shown in Figure 1(b), the dual-fed distributed amplifier incorporates a pair of transmission lines (TLs) that are periodically coupled by the transistors, a hybrid to drive both ends of the gate (input) line, and another hybrid to combine waves appearing at the ends of the drain (output) line [5–7]. Previous literature has investigated the gain and efficiency of the dual-fed distributed PA compared with the conventional distributed PA [5–7]. For the distributed PA, the magnitudes of the FET drain voltage and current are equal when the FETs are located by a /2 at the center frequency. Therefore, the electrical spacing between the FETs is /2, and the short-circuit terminations of the gate and drain lines are /4 from the nth FET [7]. The line impedances of the gate and drain lines are both equal to 50 ⍀. 3. IMPLEMENTATION AND EXPERIMENTAL RESULTS
For the measurement, the two-way balanced and distributed PAs have been designed and implemented using a push-pull RFHIC RT440 GaN HEMT with the P1dB of 20 W, which avoids the circuit complexity and large size. The balanced GaN HEMT PA has employed a 3-dB hybrid coupler of Anaren 1 ⫻ 603 as the power divider and combiner. However, the signal separation and combination in the distributed PA was simply implemented with TLs (/2 and /4) without the two-way power divider and combiner as shown in Figure 2. The shunt capacitors of Cd were inserted to optimize the performance and were 3.6 pF [9]. Figure 3 shows the photographs of the fabricated balanced and distributed
PA1 RFout
RFin N-way Power Divider
PAn
(a) RFout
RFin
λ
2
λ
2
λ
λ
2. WIDEBAND DUAL-FED DISTRIBUTED POWER AMPLIFIER
574
PA1
PA2
4
Figure 1(a) shows the schematics of the N-way balanced PA. It is necessary for the N-way balanced PA to combine the input and output powers of the several transistors using the N-way power divider and combiner. However, the distributed amplification is a
N-way Power Combiner
PA2
4
PAn
(b) Figure 1 Schematics of (a) the N-way balanced PA and (b) the N-way dual-fed distributed PA
MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 2, February 2009
DOI 10.1002/mop
60
50
RFin
λ
PA1
2
2
λ
λ
4
4
Cd
PA2
45
Gain [dB], Output Power [dBm]
λ
Cd
Figure 2 Schematic of the two-way dual-fed distributed PA
Solid: Balanced PA Hollow: Distributed PA 50
40
Pout
35
40
PAE
30
30 25 20
20 15
10 10
Gain 5
0 10
PAs. To reduce the memory effects, the drain bias lines in both PAs incorporated a quarter-wavelength bias line and several decoupling capacitors. Each matching circuit was optimized using shunt capacitors. The balanced and distributed PAs had the quiescent current (IDQ) of 240 mA with the drain bias voltage (Vdd) of 28 V. Figure 4 shows the measured output power, gain, and poweradded efficiency (PAE) characteristics of the balanced and distributed PAs according to input power level for a CW of 2.6 GHz. From the results, both amplifier shows similar performance beside of the lower gain of the distributed PA. Figure 5 depicts the measured gain and PAE performances of the balanced and distributed PAs according to the change of operating frequency at an average output power of 33 dBm for a CW. For the balanced PA, the good gain and PAE performances are achieved around the center frequency, but the performance is degraded at the lowermost and uppermost frequencies. However, the distributed PA shows flat gain and PAE performances in wider frequency range than the distributed PA. The gain flatness is less
Vdd2
Vgs2 RFin PA2
Power Divider
Input Matching
Output Matching
PAE [%]
RFout
Power Combiner
PA1
12
14
16
18
20
22
24
26
28
30
32
34
Input Power [dBm]
Figure 4 Measured output power, gain, and PAE characteristics of the balanced and distributed PAs according to input power level for a continuous wave of 2.6 GHz. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
than 0.5 dB up to the bandwidth of 180 MHz. Therefore, the distributed PA shows wideband performance compared with the balanced PA. Figure 6 shows the third- and fifth-order intermodulation distortions (IMD3s and IMD5s) of the balanced and distributed PAs for two-tone signals with various tone spacings. The unbalance between the lower and upper components of less than 4 dB is obtained for the two-tone signal with up to 100-MHz tone spacing, which proves that the distributed PA has a considerably reduced memory effect compared with the balanced PA. For the WiMAX signal, the relative constellation error (RCE) instead of an adjacent channel leakage ratio (ACLR) should be measured. The RCE shows the amount of in-band-error like the error vector magnitude (EVM) [2]. The IEEE 802.16 –2004 with 28 MHz signal bandwidth, the burst type of 64-QAM-3/4, and the PAPR of 9.47 dB was used as the WiMAX signal. The linearity specification of the modulated signal is the RCE of ⫺31 dB. Figure 7 shows the measured RCE and PAE characteristics of the distributed PA according to output power level for a WiMAX signal of 2.6 GHz. The distributed PA was optimized at an average output
RFout 20
20
Average Output Power = 33dBm
Solid: Balanced PA Hollow: Distributed PA
Vdd1
18
PAE
18
(a) 4
Vgs2
Vdd2
4
Cd
12
Cd
λ
2
Input Matching
PA2
Output Matching
PA1
14
16
λ
λ
Gain [dB]
λ
16
14
10
Gain
8 12
PAE [%]
Vgs1
6
2
4
10
2
RFin
Vgs1
Vdd1
RFout
8 2.50
2.55
2.60
2.65
0 2.70
Frequency [GHz]
(b) Figure 3 Photographs of (a) the fabricated balanced PA and (b) the fabricated distributed PA. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
DOI 10.1002/mop
Figure 5 Measured gain and PAE performances of the balanced and distributed PAs according to the change of operating frequency at an average output power of 33 dBm for a CW. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 2, February 2009
575
-25
1.5
Balanced PA
IMD3(-) IMD3(+) IMD5(-) IMD5(+)
-30
Distributed PA
IMD3(-) IMD3(+) IMD5(-) IMD5(+)
1.0
-35
Imaginary
IMD [dBc]
0.5 -40
IMD3 -45
-50
IMD5
-55
-0.5
-1.0
-60 0
0.0
20
40
60
80
-1.5 -1.5
100
-1.0
-0.5
0.0
Tone Spacing [MHz]
0.5
1.0
1.5
Real
Figure 6 Measured IMD3 and IMD5 characteristics of the balanced and distributed PAs according to tone spacing for a two-tone signal. [Color figure can be viewed in the online issue, which is available at www. interscience.wiley.com.]
Figure 8 Measured constellation of the distributed PA at an average output power of 33 dBm for a WiMAX signal. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
power of 33 dBm with the optimization of the gate bias voltages. The gate bias voltages of the PA1 and PA2 in the distributed PA were changed to ⫺1.45 (IDQ of 200 mA) and ⫺1.59 V (IDQ of 20 mA), respectively. From the measured results, the distributed PA can deliver an average output power of 35.6 dBm with a PAE of 27.1% at a RCE of ⫺31 dB. Figure 8 shows the measured constellation of the distributed PA at an average output power of 33 dBm. Figure 9 shows the measured RCE, gain, and PAE characteristics of the distributed PA according to the change of operating frequency at an average output power of 33 dBm for a WiMAX signal. Over 150-MHz bandwidth, the RCE of ⫺31 dB was maintained with the gain flatness of less than 0.5 dB and the PAE variation of less than 3%.
applications. For accurate comparison, the two-way balanced PA as well as the two-way distributed PA has been implemented using a push-pull RFHIC RT440 GaN HEMT. Compared with the balanced PA, the distributed PA was implemented using TLs instead of the 3-dB hybrid couplers as the two-way power divider and combiner. For a CW, the distributed PA shows more wideband gain and PAE performances than the balanced PA. After optimization by controlling the gate bias voltage, the distributed PA shows wideband RCE, gain, and PAE characteristics over 150 MHz for a WiMAX signal with the PAR of 9.47 dB and the signal bandwidth of 28 MHz. From the measured results prove that the wideband dual-fed distributed GaN HEMT PA can be a promising solution for a WiMAX applications with high PAR and large signal bandwidth.
4. CONCLUSIONS
ACKNOWLEDGMENTS
In this article, we have proposed a wideband distributed GaN HEMT PA based on the dual-fed structure for 2.6 GHz WiMAX
This work was partially supported by the BK21 program and the National Center for Nanomaterials Technology (NCNT) in Korea.
40
-15
25
-20
20
-15
Average Output Power = 33dBm
35
-20
30
20 15
-30
Linearity Specification
RCE [dB]
PAE [%]
-25
10
15
-25
Gain RCE
10
-30
RCE [dB]
27.1% 25
Gain [dB], PAE [%]
PAE
Linearity Specification
-35
5
-40
0 2.50
-35
5
35.6 dBm
0 28
30
32
34
36
38
2.55
2.60
2.65
-40 2.70
Output Power [dBm]
Frequency [GHz]
Figure 7 Measured RCE and PAE characteristics of the distributed PA according to output power level for a WiMAX signal of 2.6 GHz. [Color figure can be viewed in the online issue, which is available at www. interscience.wiley.com.]
Figure 9 Measured RCE, gain, and PAE characteristics of the distributed PA according to the change of operating frequency at an average output power of 33 dBm for a WiMAX signal. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
576
MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 2, February 2009
DOI 10.1002/mop
REFERENCES
Key words: reflector antenna; spherical reflector; distortion compensation; reflectarray
1. K.R. Santhi and G.S. Kumaran, Migration to 4 G: Mobile IP based solutions, In Proceedings of AICT/ICIW’06, 2007, pp. 76 – 82. 2. J. Moon, J. Kim, I. Kim, J. Kim, and B. Kim, A wideband envelope tracking Doherty amplifier for WiMAX Systems, IEEE Microwave Wireless Compon Lett 18 (2008), 49 –51. 3. T.T.Y. Wong, Fundamentals of distributed amplification, Artech House, Boston, MA, 1993. 4. J.L.B. Walker, Some observations on the design and performance of distributed amplifiers, IEEE Trans Microwave Theory Tech 40 (1992), 164 –168. 5. C.S. Aitchison, N. Bukhari, C. Law, and N. Nazoa-Ruiz, The dual-fed distributed amplifier, IEEE MTT-S Int Microwave Symp Dig, New York, NY (1988), 911–914. 6. C.S. Aitchison, M.N. Bukhari, and O.S.A. Tang, The enhanced power performance of the dual-fed distributed amplifier, In European Microwave Conference, 1989, pp. 439 – 444. 7. K.W. Eccleston, Compact dual-fed distributed power amplifier, IEEE Trans Microwave Theory Tech 53 (2005), 825– 831. 8. P.N. Shastry, S.N. Prasad, and A.S. Ibrahim, Design guidelines for a novel tapered drain line distributed power amplifier, 36rd European Microwave Conference, Manchester, UK, 2006, pp. 1274 –1278. 9. K.J. Cho, W.J. Kim, J.Y. Kim, J.H. Kim, and S.P. Stapleton, N-way distributed Doherty amplifier with an extended efficiency range, IEEE MTT-S Int Microwave Symp, Honolulu, HI, (2007), 1581–1584. 10. Y.S. Lee, M.W. Lee, and Y.H. Jeong, Experimental analysis of GaN HEMT and Si LDMOS in analog predistortion power amplifier for WCDMA applications, Microwave Opt Technol Lett 50 (2007), 393– 396. 11. Y.S. Lee, M.W. Lee, and Y.H. Jeong, Linearity-optimized GaN HEMT Doherty amplifiers using derivative superposition technique for WCDMA applications, Microwave Opt Technol Lett 50 (2008), 701–705. © 2008 Wiley Periodicals, Inc.
A COMPENSATED SPHERICAL REFLECTOR ANTENNA USING SUBREFLECTARRAYS Shenheng Xu and Yahya Rahmat-Samii Department of Electrical Engineering, University of California, Los Angeles, Los Angeles, CA 90095-1594; Corresponding author: [email protected] Received 10 September 2008 ABSTRACT: Spherical reflectors have been used as wide-angle scanning antennas due to the unique geometric properties of the sphere. The inherent spherical aberration, however, causes considerable performance degradation. This article presents a novel sub-reflectarray compensating technique to correct it. A microstrip reflectarray is a planar, inexpensive, passive device with unique capabilities of controlling reflection phases. By exploiting this feasible property, it is employed as an auxiliary phase corrector, and the phase errors caused by spherical aberration are compensated. Two examples, a 35-m spherical reflector antenna at Ka-band and a 1.5-m breadboard spherical reflector, are presented, and simulation results show great performance improvement for both no scan and large scan cases. Because of the difficulty in measuring large reflector antennas at UCLA, it is part of future plan to collaborate with other institutions for the verification of the concept. However, we believe that the simulation results generated in this article should provide acceptable performance prediction of the proposed antenna design. © 2008 Wiley Periodicals, Inc. Microwave Opt Technol Lett 51: 577–582, 2009; Published online in Wiley InterScience (www. interscience.wiley.com). DOI 10.1002/mop.24088
DOI 10.1002/mop
1. INTRODUCTION
Although a parabolic reflector antenna provides perfect focusing performance, it suffers from severe limitations in large scanning applications due to its poor off-focus performance. By contrast, a spherical reflector antenna is well suited because of its perfect angular symmetry. By rotating an optimally located feed system about the center of curvature of the spherical reflector and illuminating different portion of the reflector aperture, acceptable performances can be achieved across a large angular region [1, 2]. The inherent spherical aberration, however, causes tremendous performance degradation. To eliminate the spherical phase aberration, complex feed systems or auxiliary phase correctors, such as phased line sources [3, 4], planar array feeds [5, 6], and Gregorian reflectors [7], are usually employed in large spherical reflector antennas. With the recent needs of spaceborne remote sensing projects, more stringent demands are imposed on the spherical reflector antenna designs. For instance [6], the second generation of a geostationary orbiting radar requires an antenna system operating at Ka-band (35.6 GHz) with directivity of 77 dB, a half power beamwidth of 0.02°, and an off-axis scan up to ⫾4° (⫾200 beamwidths). Therefore, a 35.5-m spherical reflector antenna with an effective aperture of 28 m was proposed to satisfy the great challenge of producing an unconventionally narrow beam at ultra large beamwidth scan angles, and a 271-element hexagonal planar array feed was used to compensate for the spherical aberration. However, the fabrication of such large array feed at the high operating frequency imposes another serious challenge, considering the fact that both amplitude and phase excitations are necessary. Moreover, dielectric losses are typically considerable in such active devices. A Gregorian corrector [7] can overcome these drawbacks, but it encounters great difficulty in achieving low sidelobe levels because of the resulted high inverse illumination taper across the reflector aperture. The overall size of the antenna system is larger as well, because the Gregorian corrector must be placed outside the caustic region of the spherical reflector. A microstrip reflectarray is a planar, low-profile, inexpensive, easy-to-build, passive device which combines some best features of reflectors and microstrip arrays [8]. An individual reflectarray element is capable of reflecting the incident field with predesigned phase. By controlling the reflection phase of the reflectarray, a specific wavefront can be formed to generate a pencil beam, to electronically steer a pencil beam, or to compensate for reflector surface distortions. Different element designs are readily available in literature. Microstrip patches of variable size [9] and microstrip patches with stubs of variable length [10] are widely used due to the ease of implementation. Reflectarrays have attracted more attention in antenna design communities, especially for spaceborne applications, because of those aforementioned favorable features. In this article, a novel sub-reflectarray compensating technique [11, 12] is proposed to compensate for the spherical aberration. By exploiting the phase controlling capability, a microstrip reflectarray is employed as an auxiliary phase corrector to eliminate the spherical phase errors. The implementation of this novel technique is presented in Section 2. The concept of sub-reflectarray compensation is introduced, and some design guidelines are discussed. In Section 3, the 35.5-m spherical reflector antenna operating at Ka-band is successfully compensated using a sub-reflectarray of 0.3-m in diameter. The compensation of a 1.5-m breadboard spherical reflector antenna is presented as well for future experimental
MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 2, February 2009
577
