Analysis and Simulation of Memory Effects on Microwave Power Amplifier
Jingchang Nan, Jiuchao Li
School of electrics and information engineering, Liaoning Technical University, Huludao, 125105, China
Email: nanjc886@sina.com,lijiuchao@gmail.com
Jingchang Nan, Jiuchao Li
School of electrics and information engineering, Liaoning Technical University, Huludao, 125105, China
Email: nanjc886@sina.com,lijiuchao@gmail.com
Abstract—Analysis and simulation the existence of electrical memory effects and thermal memory effects in different spacing two-tone input signal and modulated signal source through advanced design system (ADS) software simulation, respectively. Built thermal memory effects compensation circuit in the ADS environment to confirm the compensation circuit has a good compensation to the thermal memory effects. Created an adaptive digital baseband predistortion system, take the power amplifier memory model with sparse delay taps as an example, put it into the created system and system-level simulation, simulation results demonstrate the system have a good predistortion result with the memory PA model, which performance can be more close to a real system than memory-less model, and has a significant sense for designing real system.
Index Terms—power amplifier (PA), memory effects, compensation circuit, sparse delay taps
I. INTRODUCTION
Power amplifier (PA) is an indispensable component of communication system. The design of RF power amplifiers for wireless base-station applications is becoming an increasingly complex task. These applications are characterized by a continuously growing bandwidth and envelope variation, and thereby impose stringent constraints on system designers to meet performance requirements in terms of power efficiency and good linearity. Memory effects are defined as distortion phase and amplitude changes over the modulation bandwidth, which obvious character is spectrum asymmetry. Memory effects of PAs can not be ignored, especially in wideband signal processing and base-station application. This can be divided into electrical memory effects and thermal memory effects. Electrical effects are the dominant portion of the memory effects in the wideband environment. The fundamental reason is the frequency-dependency of the bias and matching networks. The matching impedance network cannot guarantee perfect cancellation due to its static nature and the dynamic behavior in the PA is visible as hysteresis in the envelope transfer characteristics. Thermal memory effects are caused by electro-thermal couplings. From the measurement point of view, when we measure the power amplifier intermodulation distortion (IMD) for the characteristics of its nonlinear behavior, we often observe with frequency-dependent envelope on the bottom with the imbalance and IMD rate changes, which means that AM / AM, AM / PM functions are not static, while depend on the signal envelope frequency and amplitude changes. Memory effects based on the existence of behavioral model can be divided into three categories [1] [2] [3]: memory-less behavioral model, linear (short-term) memory effects behavioral model and nonlinear (long-term) memory effects behavioral model.
This paper is organized as fellows. In Section II, using the ADS simulation methods to analyze the different spacing of two-tone input electrical memory effects existence. In Section III, in the ADS environment to build the temperature model and the temperature compensation circuit and simulate, respectively. Adaptive digital baseband predistortion system and simulation results are given in Section IV. We draw our conclusion in Section V.
Index Terms—power amplifier (PA), memory effects, compensation circuit, sparse delay taps
I. INTRODUCTION
Power amplifier (PA) is an indispensable component of communication system. The design of RF power amplifiers for wireless base-station applications is becoming an increasingly complex task. These applications are characterized by a continuously growing bandwidth and envelope variation, and thereby impose stringent constraints on system designers to meet performance requirements in terms of power efficiency and good linearity. Memory effects are defined as distortion phase and amplitude changes over the modulation bandwidth, which obvious character is spectrum asymmetry. Memory effects of PAs can not be ignored, especially in wideband signal processing and base-station application. This can be divided into electrical memory effects and thermal memory effects. Electrical effects are the dominant portion of the memory effects in the wideband environment. The fundamental reason is the frequency-dependency of the bias and matching networks. The matching impedance network cannot guarantee perfect cancellation due to its static nature and the dynamic behavior in the PA is visible as hysteresis in the envelope transfer characteristics. Thermal memory effects are caused by electro-thermal couplings. From the measurement point of view, when we measure the power amplifier intermodulation distortion (IMD) for the characteristics of its nonlinear behavior, we often observe with frequency-dependent envelope on the bottom with the imbalance and IMD rate changes, which means that AM / AM, AM / PM functions are not static, while depend on the signal envelope frequency and amplitude changes. Memory effects based on the existence of behavioral model can be divided into three categories [1] [2] [3]: memory-less behavioral model, linear (short-term) memory effects behavioral model and nonlinear (long-term) memory effects behavioral model.
This paper is organized as fellows. In Section II, using the ADS simulation methods to analyze the different spacing of two-tone input electrical memory effects existence. In Section III, in the ADS environment to build the temperature model and the temperature compensation circuit and simulate, respectively. Adaptive digital baseband predistortion system and simulation results are given in Section IV. We draw our conclusion in Section V.
II. ELECTRICAL MEMORY EFFECTS SIMULATION
To understand the performance and memory effects of the PA, we simulate it in ADS environment. Transistor amplifier model MRF9742 is an LDMOS type transistor, with the amplifier performance is as follows: VDS =5.8V, VGS = 2.0V, Pout = 30W PEP, Freq = 890MHz, Gps = 11dB, = 60% (leakage efficiency). In the ADS environment build the test circuit for harmonic simulation, with scan range: power 10-24dBm, step for 1dBm; tone between 1MHz-10MHz, the maximum order of harmonic balance is 7. ADS simulation module will run and get the simulation results. Fig.1 and Fig.2 show the third-order IMD products (both high and low sidebands) with the output power and phase with the relationship between two-tone frequencies spacing, respectively [4]. It can be seen that when the output power increases, the third-order IMD products amplitude also increases, and there is more obvious change about 19dBc. When the two-tone spacing change, the phase angle will change in the imbalance situation. Output signal amplitude and phase characteristics as well as the power amplifier will change, indicating the existence of the electric memory effects. Electrical memory effects are caused by varying impedances across the modulation bandwidth .The frequency dependence of the source and load impedances cannot be kept constant for all modulation frequencies. The amplitude and phase of the IMD products are dependent on the frequency of the impedances .Careful design of the bias networks can reduce the electrical memory effects.
To understand the performance and memory effects of the PA, we simulate it in ADS environment. Transistor amplifier model MRF9742 is an LDMOS type transistor, with the amplifier performance is as follows: VDS =5.8V, VGS = 2.0V, Pout = 30W PEP, Freq = 890MHz, Gps = 11dB, = 60% (leakage efficiency). In the ADS environment build the test circuit for harmonic simulation, with scan range: power 10-24dBm, step for 1dBm; tone between 1MHz-10MHz, the maximum order of harmonic balance is 7. ADS simulation module will run and get the simulation results. Fig.1 and Fig.2 show the third-order IMD products (both high and low sidebands) with the output power and phase with the relationship between two-tone frequencies spacing, respectively [4]. It can be seen that when the output power increases, the third-order IMD products amplitude also increases, and there is more obvious change about 19dBc. When the two-tone spacing change, the phase angle will change in the imbalance situation. Output signal amplitude and phase characteristics as well as the power amplifier will change, indicating the existence of the electric memory effects. Electrical memory effects are caused by varying impedances across the modulation bandwidth .The frequency dependence of the source and load impedances cannot be kept constant for all modulation frequencies. The amplitude and phase of the IMD products are dependent on the frequency of the impedances .Careful design of the bias networks can reduce the electrical memory effects.
Figure 1. The relationship between 3rdorder IMD products and output power
Figure 2. Therelationship between 3rdorder IMD phase and frequency spacingIII. TEMPERATURE MODEL AND COMPENSATION CIRCUIT
A. Thermal memory effect simulation
Figure 3. Thermal memory effects simulation circuit
A. Thermal memory effect simulation
To verify the thermal dissipation and junction temperature cause thermal memory effects [4] [5] [6], we build a model temperature after the power amplifier. The input signal source is modulated CDMA2000 signal, testing the use of ADS provided amplifier. Fig.3 is the simulation of the test thermal memory effect circuit structure, which consists of a modulation signal source, a power amplifier and the temperature model. Fig.4 is the result of thermal memory effects of the power amplifier simulation. Because of the existence of the junction temperature, the spectrum of the output voltage has a corresponding change. The spectrum of the Vout is deteriorated than Vin shows that the distortion is caused by temperature increase, which leads to a spectrum expansion and intermodulation component regeneration. Therefore, simulation results show that the thermal memory effects of the PA exists.
Figure 3. Thermal memory effects simulation circuit
Figure 4. Thermal memory effects simulation resultB. Temperature compensation circuit
Thermal power feedback causes thermal memory effects at low modulation frequencies. Increased power dissipation causes the power amplifier junction temperature to increase, which in turn alters the amplifier's gain [4]. These memory effects are observed as the envelope varies over time. The output signal in the memory PA is not only related to the recent input signal, but also to the foregoing input signal. A thermal memory compensator is inducted, which is similar to a general pre-distorter which can restrain the spectrum re-growth and linear of the PAs input and output performance [7]. The structure of temperature compensation circuit is as Fig.5. The simulation results are shown in Fig.6. The simulation results show that the expansion of spectrum and intermodulation components significantly reduce regeneration; Vout and Vin are basic matched. It is confirmed that the compensation circuit of the thermal memory effect has a very good compensation.
Figure 5. Thermal memory effects compensation circuit
Figure 6. Compensation simulation resultIV. SYSTEM-LEVEL SIMULATION
A block diagram of an adaptive baseband digital pre- distortion system [8] is shown in Fig.7. This is architecture of full digital domain, including input signal. With inclusion of DP, the digital complex baseband input signal samples are multiplied prior to the digital-to- analog converter by complex coefficients drawn from the look-up table (LUT) [9]. The adaptation algorithm determines the values of the coefficients by comparing the feedback signal and a delayed version of the input signal. The pre-distortion function is implemented using a complex multiplier, a LUT, and an address-generation block that selects he appropriate coefficient from the LUT, given the magnitude of the input signal. The size of the LUT employed determines the number of points at which the pre-distortion function is calculated. In addition, the distribution of the pre-distortion function points need not necessarily be evenly distributed across the range of the input signal magnitude. Instead, it may be desirable to distribute the pre-distortion function points across the range of the input signal magnitude using a squared (power) or logarithmic relationship. This system-level simulation of the signal source used WCDMA forward link baseband signal, the spectrum width is 5MHz, look-up table size is set to 256 ,put into the memory PA behavioral model with sparse delay taps[10]. When the system simulation circuit is built and parameter configuration is completed in ADS, start simulation. Simulation results are as shown in Fig.8.The simulation results show that: the digital baseband pre-distortion structure will get a good pre-distortion effect to meet the standards required by the power spectral. Input
power of the energy concentrated in the vicinity of the center frequency range of 5MHz, the proliferation of low-frequency distortion become smaller, the power spectral density can achieve the improvement over 20 dB. System simulation performance can be more close to real system.
power of the energy concentrated in the vicinity of the center frequency range of 5MHz, the proliferation of low-frequency distortion become smaller, the power spectral density can achieve the improvement over 20 dB. System simulation performance can be more close to real system.
Figure 7. Adaptive baseband digital predistortion system
Figure 8. The results of predistortionV CONCLUSION
In this paper, we have analyzed and simulated the PA memory effects. The simulation confirms the existence of electric memory effects and thermal memory effects, respectively .We build a thermal memory compensation circuit that makes the thermal memory effects significantly weaken. We put the memory model with sparse delay taps into the created adaptive digital baseband pre-distortion system. After system-level simulating, we can get a good predistortion effect. It confirms that memory PA behavioral model can be more close to real system than memory-less model, which has a significant sense for designing real system
In this paper, we have analyzed and simulated the PA memory effects. The simulation confirms the existence of electric memory effects and thermal memory effects, respectively .We build a thermal memory compensation circuit that makes the thermal memory effects significantly weaken. We put the memory model with sparse delay taps into the created adaptive digital baseband pre-distortion system. After system-level simulating, we can get a good predistortion effect. It confirms that memory PA behavioral model can be more close to real system than memory-less model, which has a significant sense for designing real system
REFERENCES
[1] José C. Pedro, Stephen A. Maas. "A Comparative Overview of Microwave and Wireless Power-Amplifier Behavioral Modeling Approaches." IEEE transactions on microwave theory and techniques, 2005, 53(4): 1150-1163.
[2] C´esar S´anchez, Jes´us de Mingo, Paloma Garc´Õa, Pedro Luis Carro, Antonio Valdovinos, "Memory Behavioral Modeling of RF Power Amplifiers," 2008 IEEE,pp1954-1958.
[3] Georgina Stegmayer Omar Chiotti, "Volterra NN-based behavioral model for new wireless communications devices," Neural Comput & Applic, 2009, pp283-291
[4] Agilent's ADS 2008 Update 1 Design guide.
[5] Long term memory effects in wireless QPSK modulated signals[C].IEEE/MTT-S International Microwave Symposium, 2007, 961-964.
[6] Slim Boumaiza, Fadhel M. Ghannouchi. "Thermal memory effects modeling and compensation in RF power amplifiers and predistortion linearizers," IEEE transactions on microwave and techniques. 2003, 51(12):2427-2433.
[7] Lei Ding, G.Tong Zhou, and Denis R.Morgan. "Memory polynomial predistorter based on the indirect learning architecture". IEEE 0-0703-7632-3/3/02, 2002, pp. 967-971.
[8] Wan-Jong Kim, Shawn P. Stapleton, Jong Heon Kim, Cory Edelman. "Digital predistortion linearizes wireless power amplifiers," IEEE microwave magazine. 2005, 6(3), 54-61.
[9] Chih-Hung Lin, Hsin-Hung Chen, Yung-Yi Wang, Jiunn- Tsair Chen. "Dynamically optimum lookup-table spacing for power amplifier perdistortion linearization," IEEE transactions on microwave theory and techniques, 2006, 54(5): 2118-2127.
[10] Hyunchul Ku, J.Stevenson Kenney. "Behavioral modeling of nonlinear RF power amplifiers considering memory effects", IEEE transactions on microwave theory and techniques. 2003, 51(12): 2495-2503.
Asignatura: CRF
Dujeiny J. Sánchez Q
[1] José C. Pedro, Stephen A. Maas. "A Comparative Overview of Microwave and Wireless Power-Amplifier Behavioral Modeling Approaches." IEEE transactions on microwave theory and techniques, 2005, 53(4): 1150-1163.
[2] C´esar S´anchez, Jes´us de Mingo, Paloma Garc´Õa, Pedro Luis Carro, Antonio Valdovinos, "Memory Behavioral Modeling of RF Power Amplifiers," 2008 IEEE,pp1954-1958.
[3] Georgina Stegmayer Omar Chiotti, "Volterra NN-based behavioral model for new wireless communications devices," Neural Comput & Applic, 2009, pp283-291
[4] Agilent's ADS 2008 Update 1 Design guide.
[5] Long term memory effects in wireless QPSK modulated signals[C].IEEE/MTT-S International Microwave Symposium, 2007, 961-964.
[6] Slim Boumaiza, Fadhel M. Ghannouchi. "Thermal memory effects modeling and compensation in RF power amplifiers and predistortion linearizers," IEEE transactions on microwave and techniques. 2003, 51(12):2427-2433.
[7] Lei Ding, G.Tong Zhou, and Denis R.Morgan. "Memory polynomial predistorter based on the indirect learning architecture". IEEE 0-0703-7632-3/3/02, 2002, pp. 967-971.
[8] Wan-Jong Kim, Shawn P. Stapleton, Jong Heon Kim, Cory Edelman. "Digital predistortion linearizes wireless power amplifiers," IEEE microwave magazine. 2005, 6(3), 54-61.
[9] Chih-Hung Lin, Hsin-Hung Chen, Yung-Yi Wang, Jiunn- Tsair Chen. "Dynamically optimum lookup-table spacing for power amplifier perdistortion linearization," IEEE transactions on microwave theory and techniques, 2006, 54(5): 2118-2127.
[10] Hyunchul Ku, J.Stevenson Kenney. "Behavioral modeling of nonlinear RF power amplifiers considering memory effects", IEEE transactions on microwave theory and techniques. 2003, 51(12): 2495-2503.
Asignatura: CRF
Dujeiny J. Sánchez Q
No hay comentarios:
Publicar un comentario