Basic concepts of RF power amplifier
The RF power amplifier (RF PA) is the main part of the transmission system, and its importance is self-evident. In the front-stage circuit of the transmitter, the RF signal power generated by the modulated oscillation circuit is very small, and it needs to go through a series of amplifications (buffer stage, intermediate amplifier stage, final power amplifier stage) to obtain sufficient RF power before it can be fed to the antenna for radiation. In order to obtain a sufficiently large RF output power, an RF power amplifier must be used. After the modulator generates the RF signal, the RF modulated signal is amplified by the RF PA to sufficient power, and then transmitted by the antenna through the matching network.
The function of an amplifier is to amplify the input content and output it. The input and output content are called "signals", which are often expressed as voltage or power. For a "system" such as an amplifier, its "contribution" is to increase the level of what it "absorbs" and "output" it to the outside world. If the amplifier has good performance, it can contribute more, which reflects its own "value". If there are certain problems with the amplifier, then after starting to work or working for a period of time, it will not only fail to provide any "contribution", but may also have some unexpected "oscillations", which are disastrous to both the outside world and the amplifier itself.
The main technical indicators of RF power amplifiers are output power and efficiency. How to improve output power and efficiency is the core of RF power amplifier design goals. Usually in RF power amplifiers, LC resonant circuits can be used to select the fundamental frequency or a certain harmonic to achieve distortion-free amplification. In addition, the harmonic components in the output should be as small as possible to avoid interference with other channels.
Classification
According to different working states, power amplifiers are classified as follows:
The operating frequency of traditional linear power amplifiers is very high, but the relative frequency band is relatively narrow. RF power amplifiers generally use frequency selection networks as load circuits. RF power amplifiers can be divided into three working states according to the different current conduction angles: A, B, and C. The conduction angle of the current of the Class A amplifier is 360°, which is suitable for small signal low-power amplification. The conduction angle of the current of the Class B amplifier is equal to 180°, and the conduction angle of the current of the Class C amplifier is less than 180°. Both Class B and Class C are suitable for high-power working states, and the output power and efficiency of the Class C working state are the highest among the three working states. Most RF power amplifiers work in Class C, but the current waveform of the Class C amplifier is too distorted, so it can only be used for resonant power amplification using a tuned circuit as a load. Because the tuned circuit has filtering capabilities, the loop current and voltage are still close to the sine waveform, and the distortion is very small.
Switching Mode PA (SMPA) makes electronic devices work in a switching state. Common types include Class D amplifiers and Class E amplifiers. Class D amplifiers are more efficient than Class C amplifiers. SMPA drives active transistors into a switching mode. The working state of the transistor is either on or off. There is no overlap in the time domain waveforms of its voltage and current, so the DC power consumption is zero, and the ideal efficiency can reach 100%.
Traditional linear power amplifiers have high gain and linearity but low efficiency, while switch-mode power amplifiers have high efficiency and high output power but poor linearity. See the table below for details:
Circuit composition
There are different types of amplifiers. To simplify, the amplifier circuit can be composed of the following parts: transistors, bias and stabilization circuits, and input and output matching circuits.
1-1. Transistor
There are many kinds of transistors, including transistors with many structures that have been invented. In essence, the work of a transistor is manifested as a controlled current source or voltage source, and its working mechanism is to convert the energy of direct current without content into "useful" output. The direct current energy is obtained from the outside world, consumed by the transistor, and converted into useful components. Different transistors have different "capabilities", such as their ability to withstand power, which is also due to their different abilities to obtain direct current energy; for example, their reaction speed is different, which determines how wide and high the frequency band it can work on; for example, its impedance facing the input and output ends is different, and its external reaction ability is different, which determines the difficulty of matching it.
1-2. Bias circuit and stabilization circuit
Biasing and stabilization circuits are two different circuits, but because they are often difficult to distinguish and their design goals are similar, they can be discussed together.
The operation of transistors needs to be under certain bias conditions, which we call the static operating point. This is the foundation of the transistor and its own "positioning". Each transistor has a certain positioning for itself, and its different positioning will determine its own working mode, and there are different performances in different positioning. Some positioning points have small fluctuations, which are suitable for small signal operation; some positioning points have large fluctuations, which are suitable for high power output; some positioning points have less demand and pure release, which are suitable for low noise operation; some positioning points, the transistor always hovers between saturation and cutoff, and is in a switching state. An appropriate bias point is the basis for normal operation. When designing a broadband power amplifier, or when the operating frequency is high, the bias circuit has a greater impact on the circuit performance. At this time, the bias circuit should be considered as part of the matching circuit.
There are two major types of bias networks, passive networks and active networks. Passive networks (i.e. self-biasing networks) are usually composed of resistor networks to provide the appropriate operating voltage and current for the transistor. Its main disadvantages are that it is very sensitive to changes in transistor parameters and has poor temperature stability. Active bias networks can improve the stability of the static operating point and provide good temperature stability, but it also has some problems, such as increased circuit size, increased difficulty in circuit layout, and increased power consumption.
The stabilization circuit must be placed before the matching circuit, because the transistor needs to have the stabilization circuit as part of itself before it can come into contact with the outside world. From the outside, the transistor with the stabilization circuit is a "new" transistor. It has made certain "sacrifices" to gain stability. The mechanism of the stabilization circuit can ensure the smooth and stable operation of the transistor.
1-3. Input and output matching circuit
The purpose of the matching circuit is to choose a way of acceptance. For those transistors that want to provide greater gain, the way is to accept and output it completely. This means that through the interface of the matching circuit, different transistors can communicate more smoothly. For different types of amplifiers, the matching circuit is not just a design method of "accepting everything". Some small tubes with small DC and shallow foundation are more willing to do some blocking when accepting to obtain better noise performance, but they cannot block too much, otherwise it will affect their contribution. For some giant power tubes, it is necessary to be cautious when outputting because they are more unstable. At the same time, a certain amount of reservation helps them to exert more "undistorted" energy.
Typical impedance matching networks include L matching, π matching and T matching. Among them, L matching is characterized by its simple structure and only two degrees of freedom, L and C. Once the impedance transformation ratio and resonant frequency are determined, the Q value (bandwidth) of the network is also determined. One advantage of the π matching network is that no matter what kind of parasitic capacitance is connected to it, it can be absorbed into the network, which also leads to the widespread application of the π matching network, because in many practical situations, the dominant parasitic element is capacitance. T matching, when the parasitic parameters of the power supply and load ends are mainly inductive, T matching can be used to absorb these parasitic parameters into the network.
How to ensure RF PA stability
Every transistor is potentially unstable. A good stabilization circuit can be integrated with the transistor to form a "sustainable operation" mode. The implementation of the stabilization circuit can be divided into two types: narrowband and broadband.
The narrowband stabilization circuit is to perform a certain gain consumption. This stabilization circuit is achieved by adding a certain consumption circuit and a selective circuit. This circuit allows the transistor to contribute only within a very small frequency range. Another broadband stabilization is to introduce negative feedback. This circuit can work within a very wide range.
The root cause of instability is positive feedback. The idea of narrowband stabilization is to curb some of the positive feedback, which of course also suppresses the contribution. If negative feedback is done well, there are many additional and gratifying advantages. For example, negative feedback may exempt the transistor from matching, and it can be well connected with the outside world without matching. In addition, the introduction of negative feedback will improve the linear performance of the transistor.
RF PA Efficiency Improvement Technology
The efficiency of transistors has a theoretical limit. This limit varies with the choice of bias point (static operating point). In addition, poor peripheral circuit design will greatly reduce its efficiency. Currently, engineers have few ways to improve efficiency. Here we will only talk about two: envelope tracking technology and Doherty technology.
The essence of envelope tracking technology is to separate the input into two types: phase and envelope, and then use different amplifier circuits to amplify them respectively. In this way, the two amplifiers can focus on their respective parts, and the cooperation of the two can achieve the goal of higher efficiency utilization.
The essence of Doherty technology is: using two transistors of the same type, only one of them works at a low input and works in a high efficiency state. If the input increases, both transistors work at the same time. The basis for the realization of this method is that the two transistors must work in harmony. The working state of one transistor will directly determine the working efficiency of the other.
Testing Challenges Facing RF PAs
Power amplifiers are very important components in wireless communication systems, but they are inherently nonlinear, which can cause spectral regrowth and interfere with adjacent channels, and may violate out-of-band emission standards mandated by law. This characteristic can even cause in-band distortion, increasing the bit error rate (BER) of the communication system and reducing the data transmission rate.
Under the peak-to-average power ratio (PAPR), the new OFDM transmission format will have more occasional peak power, making it difficult for the PA to be split. This will reduce the spectrum mask compliance, expand the EVM of the entire waveform and increase the BER. To solve this problem, design engineers usually deliberately reduce the operating power of the PA. Unfortunately, this is a very inefficient method because if the PA reduces the operating power by 10%, it will lose 90% of the DC power.
Most RF PAs today support multiple modes, frequency ranges, and modulation modes, which increases the number of test items. Thousands of test items are no longer uncommon. The use of new technologies such as crest factor reduction (CFR), digital pre-distortion (DPD), and envelope tracking (ET) can help optimize PA performance and power efficiency, but these technologies will only make testing more complex and significantly extend design and test time. Increasing the bandwidth of RF PAs will increase the bandwidth required for DPD measurements by 5 times (possibly more than 1 GHz), further increasing test complexity.
According to the trend, in order to increase efficiency, RF PA components and front-end modules (FEM) will be more closely integrated, and a single FEM will support a wider range of frequency bands and modulation modes. Integrating envelope tracking power supplies or modulators into FEMs can effectively reduce the overall space requirements inside mobile devices. In order to support a larger operating frequency range, a large number of filter/duplexer slots will increase the complexity of mobile devices and the number of test items.
Changes in semiconductor materials:
Ge (germanium), Si (silicon) →→→GaAs (gallium arsenide), InP (indium phosphide) →→→SiC (silicon carbide), GaN (gallium nitride), SiGe (silicon germanium), SOI (silicon on insulator) →→→carbon nanotube (CNT) →→→graphene.
At present, the mainstream process of power amplifiers is still GaAs process. In addition, GaAs HBT, gallium arsenide heterojunction bipolar transistor. HBT (heterojunction bipolar transistor) is a bipolar transistor composed of gallium arsenide (GaAs) layer and aluminum gallium arsenide (AlGaAs) layer.
Although the CMOS process is relatively mature, the application of Si CMOS power amplifiers is not widespread. In terms of cost, although the silicon wafers of the CMOS process are relatively cheap, the layout area of the CMOS power amplifier is relatively large. In addition, the R&D cost invested in the complex design of the CMOS PA is relatively high, making the overall cost advantage of the CMOS power amplifier not so obvious. In terms of performance, the performance of CMOS power amplifiers in terms of linearity, output power, efficiency, etc. is poor, and the inherent disadvantages of the CMOS process are: high knee voltage, low breakdown voltage, and low resistivity of the CMOS process substrate.
Carbon nanotubes (CNTs) are considered to be ideal materials for nanoelectronic devices due to their small physical size, high electron mobility, high current density and low intrinsic capacitance.
Graphene, a zero-bandgap semiconductor material, is bound to become a popular material for the next generation of radio frequency chips because of its high electron migration rate, nanometer-level physical size, and excellent electrical and mechanical properties.
RF PA Linearization Technology
The nonlinear distortion of the RF power amplifier will generate new frequency components, such as second-order distortion will generate second harmonics and two-tone beat frequency, and third-order distortion will generate third harmonics and multi-tone beat frequency. If these new frequency components fall within the passband, they will directly interfere with the transmitted signal. If they fall outside the passband, they will interfere with the signals of other channels. For this reason, the RF power amplifier should be linearized, which can better solve the problem of signal spectrum regeneration.
The principle and method of basic linearization technology of RF power amplifier is nothing more than taking the amplitude and phase of the input RF signal envelope as a reference, comparing it with the output signal, and then generating appropriate correction. At present, the power amplifier linearization technologies that have been proposed and widely used include power back-off, negative feedback, feedforward, predistortion, envelope elimination and restoration (EER), and linear amplification using nonlinear elements (LINC). More complex linearization technologies, such as feedforward, predistortion, envelope elimination and restoration, and linear amplification using nonlinear elements, have better effects on improving the linearity of amplifiers. However, linearization technologies that are easier to implement, such as power back-off and negative feedback, have limited improvement in linearity.
2-1. Power fallback
This is the most commonly used method, which is to use a larger power tube as a low-power tube. In fact, it improves the linearity of the power amplifier at the expense of DC power consumption.
The power back-off method is to back off the input power of the power amplifier from the 1dB compression point (the amplifier has a linear dynamic range, within which the output power of the amplifier increases linearly with the input power. As the input power continues to increase, the amplifier gradually enters the saturation region, and the power gain begins to decrease. Usually, the output power value when the gain drops to 1dB lower than the linear gain is defined as the 1dB compression point of the output power, expressed as P1dB.) by 6-10 decibels, and work at a level far less than the 1dB compression point, so that the power amplifier is away from the saturation region and enters the linear working area, thereby improving the third-order intermodulation coefficient of the power amplifier. In general, when the fundamental power is reduced by 1dB, the third-order intermodulation distortion is improved by 2dB.
The power back-off method is simple and easy to implement. It does not require any additional equipment and is an effective method to improve the linearity of the amplifier. However, its disadvantage is that its efficiency is greatly reduced. In addition, when the power back-off reaches a certain level, when the third-order intermodulation reaches below -50dBc, further back-off will no longer improve the linearity of the amplifier. Therefore, in situations where linearity requirements are very high, it is not enough to rely solely on power back-off.
2-2. Predistortion
Predistortion is to add a nonlinear circuit in front of the power amplifier to compensate for the nonlinear distortion of the power amplifier.
The advantages of pre-distortion linearization technology are that there is no stability problem, a wider signal bandwidth, and the ability to process signals containing multiple carriers. Pre-distortion technology is low-cost and consists of several carefully selected components packaged into a single module, which is connected between the signal source and the power amplifier to form a pre-distortion linear power amplifier. The power amplifier in the handheld mobile station has adopted pre-distortion technology, which uses only a small number of components to reduce the intermodulation product by a few dB, but it is a very critical few dB.
Pre-distortion technology is divided into two basic types: RF pre-distortion and digital baseband pre-distortion. RF pre-distortion is generally implemented using analog circuits, which have the advantages of simple circuit structure, low cost, easy high-frequency and broadband applications, but the disadvantages are that the spectrum regeneration component is less improved and the high-order spectrum component is difficult to offset.
Digital baseband predistortion is a promising method because it can be implemented with digital circuits due to its low operating frequency. It is highly adaptable and can offset high-order intermodulation distortion by increasing the sampling frequency and the quantization order. This predistortion consists of a vector gain regulator that controls the amplitude and phase of the input signal according to the content of the lookup table (LUT). The magnitude of the predistortion is controlled by the input of the lookup table. Once optimized, the vector gain regulator will provide a nonlinear characteristic opposite to that of the power amplifier. Ideally, the output intermodulation product should be equal to the output amplitude of the dual-tone signal through the power amplifier and opposite in phase, that is, the adaptive adjustment module is to adjust the input of the lookup table to minimize the difference between the input signal and the output signal of the power amplifier. Note that the envelope of the input signal is also an input to the lookup table. The feedback path samples the distorted output of the power amplifier, which is then sent to the adaptive adjustment DSP through A/D conversion to update the lookup table.
2-3. Feedforward
Feedforward technology originated from "feedback", which is not a new technology. It was proposed by Bell Labs in the United States as early as the 1920s and 1930s. Except that the calibration (feedback) is added to the output, the concept is completely "feedback".
The feedforward linear amplifier is composed of two loops through couplers, attenuators, synthesizers, delay lines, power dividers, etc. After the RF signal is input, it is divided into two paths by the power divider. One path enters the main power amplifier. Due to its nonlinear distortion, the output end has third-order intermodulation interference in addition to the main frequency signal that needs to be amplified. A part of the signal is coupled from the output of the main power amplifier, and the main carrier frequency signal of the amplifier is offset through loop 1, leaving only the inverted third-order intermodulation component. After the third-order intermodulation component is amplified by the auxiliary amplifier, it is offset by loop 2 to offset the intermodulation component generated by the nonlinearity of the main amplifier, thereby improving the linearity of the power amplifier.
Feedforward technology provides the advantages of higher calibration accuracy without the disadvantages of instability and bandwidth limitation. Of course, these advantages come at a high cost. Since the power level is high during output calibration, the calibration signal needs to be amplified to a higher power level, which requires an additional auxiliary amplifier, and the distortion characteristics of the auxiliary amplifier itself should be above the indicators of the feedforward system.
The offset requirements of the feedforward power amplifier are very high, and the amplitude, phase and delay must be matched. If there are power changes, temperature changes and device aging, the offset will fail. For this reason, adaptive offset technology is considered in the system so that the offset can keep up with the changes in the internal and external environment.
|