With the rapid development of mobile careers, especially the development of CDMA and third-generation mobile communications technology, the system is getting higher and higher. In the mobile communication system, in order to ensure a certain range of signal coverage, we usually use the power amplifier to enlarge the signal, thereby emitting through the radio frequency front end and the antenna system. In the base station of CDMA or WCDMA, and TDSCDMA, if a general high power is used (usually working in the AB class), the spectrum regeneration effect will be generated due to nonlinearity, for better solving spectral regeneration and EVM (error vector) Availability) The problem must be used in linearization techniques. Not only that, the cost ratio of the amplifier in the base station amplifier accounts for about 50%, how is it effective, and low-cost solution is very important.
1. Presentation of superlinear amplifier solutions
Traditional linear methods of conventional settlement are mostly methods of power retraction to ensure that the utility amplifier is guaranteed to operate in linear range, which does not affect the overlay and communication of the signal. Figure 1 shows a variation of the three-order interception, 1dB compression point, and three-order intermodulation with input power.
Figure 1, decibel compression point output power
As can be seen from the figure, the traditional solution is to reduce the input power, if the input power is 1 dB, the system's intermodulation component will be good 2dB, and in order to ensure linearity, for CDMA or WCDMA amplifier We can only use 100W amplification pipe to get 5W power. However, since the tube is designed to be 100W, its static work point is still very high, and the static current is still very large. Therefore, the overall current of the amplifier is very large, and the current means that the efficiency of the amplifier is very low, and there will be a large part of the heat can only be released to the pipe and board, which is both a waste of energy, more importantly. Resulting in lowering the service life of the chip. In terms of interest, it is very expensive to provide such high-power amplification pipes.
Based on these considerations, the intermodulation of simple power rollback can be achieved, and as the power is further increased, it is still solving problems. Therefore, a feedforward predistortion design is proposed to simultaneously solve linear, efficiency, and cost problems.
2, feedforward predistramn real power]
Currently mature and popular superlinear solutions include feedforward technologies, predistortion techniques (including simulated predistortion and baseband predistortion), feedback techniques, etc. Considering that the high-use feedforward technique is higher to reduce the power amplifier and improving too much efficiency, simple use of predistortion techniques can improve linear and efficiency but does not reach superlinear requirements. In combination with the disadvantage of the two technologies, a feedforward combination of predistortion techniques are proposed here. Detailed principle is shown in Figure 2.
Figure 2 Feedforward Predictive Scheme
As shown in FIG. 2, the input signal first passes the delay line preparation and the output signal through the directional coupler, thereby detecting the case where the disabled is sent, and the distortion signal is generated in the premission unit (PD), thereby improving the main Linearity of amplifier. At the same time, the output coupling of the main power is canceled with the delayed main signal. Remove the main signal, only the error signal is reserved, and the detection of the decentralizer is used as a reference control, and another Aspects are sent to the error amplifier, which is large and the mutual movement, thereby further improving intermodulation. Here, if the improvement effect is still unsatisfactory, the ultra-linear requirement is 70dBC, the feedforward ring can be increased at first, 3 or 4 levels, thereby improving the improvement effect.
The above is only an open-loop scheme, in view of the factors such as input power, temperature, etc., it is possible to design a closed-loop control link, so that the attenuator and phase shifter in the system can change according to the change of environmental parameters. Automatic tracking changes, automatically adapt to regulation, thereby ensuring the overall linear requirements.
The realization of closed-loop is first established to guide changes in each constant, including input power, output error power, ambient temperature, main signal and error signal, etc., including input power, output error power, ambient temperature, main signal and error signal to determine each parameter. Changes. At the same time, the system requires the reflection speed of the adaptive algorithm to be within 20ns, and once the parameters change, the overall intermodulation can be tracked in time. Avoid phenomenon of short-term intermodulation.
The implementation methods of the system and the core technical issues will be introduced separately in each unit.
2.1, predistortion generating unit (PD)
The predistortion part uses analog predistractaria. The program has passed the previous test, and the intermodulation of 600kHz double audio signal can improve the 15DB, and the ACPR can improve 10dB above the 1.28 MHz modulation signal. The overall block diagram of the predistortion generating unit is shown in Figure 3.
Figure 3 Predictive generation block diagram
The input signal is divided into two channels through the 3DB bridge, and the 0 ° terminal is sent to the synthetic 3DB bridge after delay line; -90 ° side as the generating end of the error signal, then pass through a 3DB bridge, 0 ° The end produces a distortion signal, and the mutual adjustment component is very large by adjusting the bias of the amplifier FP2189, and the phase-90 ° terminal is first adjusted by the attenuator to adjust the amplitude through the attenuator adjustment. Very good FP2189 produces a very good intermodulation signal so that the primary signal is different from the error signal from the -90 ° port by the synthetic bridge, so that the part of the intermodulation signal is only removed from the main signal. Retain error signals. The phase difference between the main signal and the distortion signal is achieved by means of the main signal phase of the delay line by adjusting the attenuator and the phase shifter, thereby achieving the main signal and the distortion signal phase phase difference between-190 ° side by means of another 3DB synthetic bridge. It means that the distortion signal is inverted, so that the interpolation component of the subsidiary signal is in the process of the main megadownload.
The result of the above predistortion signal is the first technical difficulty of the project, but after two months of test, the feasibility of this method has been completed, as mentioned above, the 600 kHz dual audio signal intermodulation of 2.14 GHz It can improve the 15DB or more, and the ACPR can improve the 10dB or more for a CDMA modulation signal of 1.28 MHz.
2.2, feedforward unit
The above premature programs have been trial verified, which can improve intermodulation 15dB, ACPR to improve 10dB, which can be used to introduce a large power in small pipes. However, this does not reach a super linear requirement (ie, -70dBc), so the method of introducing feedforward is further improved. If the feedforward method is used, the power requirements for error amplifiers are high, because the intermodulation products of the primary power are high, so that large power must be launched in the error amplifier can be removed. Therefore, the price of error amplifier is added and the efficiency is reduced. However, if the pre-missing unit is increased in front of the main function, it can greatly reduce the requirements of intermodulation products, mitigate the error amplifier, and improve efficiency.
The principle of the feedforward section is shown in Figure 4.
Figure 4 Feedforward Schematic
The input signal will be larger, due to the nonlinearity of the primary power, there will be interpolated components. The enlarged main signal is subjected to the input signal by the coupler, thereby enabling the amplified main signal only with the distortion signal, and the distortion signal is enlarged by the error amplifier, so that the amplitude is the same as the intermodulation product of the main signal. By the adjustment of the phase shifter and the attenuator, it is similar to the main signal phase to phase -180 °, thereby canceling the intermodulation product in the main signal, further improving the linearity of the amplifier.
In the actual design, if the first-level loop offset effect is not ideal, the superlinear requirement of -70dBC can not be achieved, and the loop can be considered to further disappear.
2.3, closed-loop adaptive unit
The predistortion unit and the feedforward portion can greatly improve the degree of linearization of the overall amplifier, but through experimental demonstration, their countermerts will be affected by the amplitude and phase of the signal. If the dense two signal phase differs exceeds 2 °, and the amplitude of the two signals exceeds 5dB, the improvement effect will be very poor. However, since the amplifier itself will receive a number of factors such as a variety of environmental temperature, the improvement of the strong changes in the input signal, therefore, our power amplifier can be required to automatically adapt to the applications of various environments. Therefore, in order to meet the requirements mentioned above, the entire system must have an adaptive unit, and automatically adjust the various parameters according to the change of the environment, thereby ensuring the power amplifier in superlineability.
The closed-loop adaptive unit will be the difficult point of the entire project. On the one hand, there is a lot of controlled parameters, and the acquired input signal is very small. At the same time, the overall mathematical model is difficult to establish, and the relationship between the input and output cannot be described in the overall mathematical model. On the other hand, the time delay from the input signal to the output signal is about 20 ns, which requires the entire algorithm to complete the calculation of the input to the output in 20 ns, that is, the real-time performance of the algorithm is very high.
The starting point of the algorithm will be based on the table, through the test, obtain a large number of movable voltages and attenuators acquired in an in-environment, with this as a sample, design a neural network machine. This neural network is trained by a large number of samples, which has the control voltage of each attenuator and phase shifter based on factors such as environment variables and input power. Eventually the neural network machine will be implemented in FPGA.
The acquisition of the test data will use the PC to design a program that can operate the signal source and the spectrometer, put the power amplifier into the high and low temperature tank to do a high and low temperature test, and adjust the size of the input power, and design a retrieval algorithm. The process automatically completes the test process, obtains a large number of test samples, thus training the neural network on MATLAB. And eventually realized in FPGA.
The above is just a hypothesis that closed-loop adaptive algorithms will be difficult to overtally, it will spend a lot of time to collect information here, trying various programs, and finally propose and design a most suitable implementation. Therefore, there may be a lot of time and effort here.
3, super linear amplifier product implementation plan
Figure 5 Predictive + main power plant implementation
Figure 6 Predictive + Feedforward + Main Power Plant
4, super linear amplifier key technical issues
Bandwidth problem.
As the frequency increases, phase and delay will change, which requires the frequency response characteristics of each module in the entire frequency band.
Question.
The fundamental starting point of the predistreative and feedforward is the problem of signaling, that is to say that the two signals are 180 ° in the phase, if the phase phase difference depends on the error exceeds 2 °, it will greatly reduce the cancellation. That is to say, it is the key to the determination effect.
Attenuator and phase shifter
Attenuators and phase shifters are the main adjustment units of predistortion and feedforward. If the attenuator is attached with additional phase shift or the phase shifter, the entire system will not be controlled. The time delay of the two is also the key to determining the delay line. Through the experimental argument, the attenuator and phase shifter with the bridge can meet the requirements, but additional phase shift and attenuation are too large, and the consistency is poor, which will be unfavorable to production. Therefore, further use of some integrated attenuators and phase shots for testing.
Adaptive algorithm
As mentioned in the closed loop control algorithm, the corresponding time and corresponding speed of the algorithm are a very critical technical indicators. At the same time, because the input parameters are too small, the entire mathematical model is difficult to establish, the research and development of the adaptive algorithm will be the bottleneck of the entire project, which will spend a lot of time and energy here.
Our other product:
