Research On Key Technologies For UPWM Type Digital Class D Audio Power Amplifier

With the development of the portable digital electronic products, the digital class D audiopower amplifier has been a hot topic in research because it has the characteristic of highefficiency and is easy to link with digital audio signal sources. The UPWM type digital classD audio power amplifier uses UPWM technique to convert the digital audio signals into theswitching signals for driving the power stage. Compared to other digital class D audio poweramplifiers, the UPWM type digital class D audio power amplifier has the advantage of simplecircuit implementation and can take the form of switching signals with a constants and lowpulse repetition frequency. However, at present the UPWM type digital class D audio poweramplifier has not completely supplanted the traditional linear audio power amplifiers in thefield of audio yet. There are three main reasons account for that. The first reason is that theexisting UPWM type digital class D audio power amplifier cannot have low cost, highefficiency and high fidelity at the same time due to the non-optimized design for functionmodules. The second reason is that UPWM is inherently nonlinear and causes the distortion inthe baseband of the audio signal. Although a number of methods have been reported to correctthis distortion, these methods cannot have low computational complexity combined with theeffectiveness of the correction. The last reason is that the power supply noise in the powerstage will degrade the output performance. The feedback technique can be used in the UPWMtype digital class D audio power amplifier to make it have a better power supply noiseimmunity capability, but the design for the UPWM type digital class D audio power amplifierwith power stage signal feedback is very difficult. For solving the above problems, Iresearched the related key technologies for UPWM type digital class D audio power amplifierand got four innovative achievements.(1) A high-order multi-bit digital Sigma-Delta modulator noise transfer function designmethod based on multi-variable optimization theory and a new stability decision scheme isproposed. The effects of the CSD encoding for all coefficients and scaling the state variableson selecting the noise transfer function are taken into account in this method, so low-costdigital Sigma-Delta modulators with optimized quantization noise shaping can be designed bythis method. Experimental results show that the eighth-order6-bit digital Sigma-Deltamodulator designed by this method can achieve121dB signal-to-noise ratio under thecondition that the oversampling ratio is8and the maximum-stable-input-amplitude is higherthan0.9FS, and its signal-to-noise ratio is better than that of the digital Sigma-Delta modulators designed by traditional methods.(2) A pseudo-natural sampling algorithm based on the3rd-order and1st-order Lagrangeinterpolation methods and a position judging method for trailing-edge PWM, and apseudo-natural sampling algorithm based on the decomposability of the asymmetricdouble-edge PWM process and the Lagrange numerical differentiation method forasymmetric double-edge PWM are proposed respectively. Experimental results show that theproposed algorithms can basically eliminate the harmonic distortion produced by UPWMwith modest computational complexity and are better than other kindred algorithms.(3) An improved UPWM nonlinear distortion precorrection method based on correctionfactor is proposed. The proposed method not only can correct the nonlinear distortionproduced by UPWM but also can solve the original method’s problem that the energy of thecorrection factors adding to the digital Sigma-Delta modulator is too big that can cause theintegrator outputs to saturate and oscillate and can degrade the performance of the poweramplifier system. Based on this method, an FPGA-based switching signal modulator isdesigned and tested. Experimental results show that the energy of the correction factorscalculated from this method is far less than that calculated from the original method and thatthe proposed method can basically eliminate the nonlinear distortion produced by UPWM.(4) A new power supply error correction method for digital class D audio poweramplifiers with half-bridge single-ended output architecture and a new power supply errorcorrection method for digital class D audio power amplifiers with H-bridge differential outputarchitecture are proposed, respectively. The methods use the feed-forward technique to build apower supply noise measurement circuit for acquiring the digitized power supply noise andadd a different power supply error block in the switching signal modulator to correct thepower supply error produced in the power stage. As the proposed methods can correct thepower supply error without changing any block in the original open-loop digital class D audiopower amplifier, therefore they can be easily used in any open-loop digital class D audiopower amplifier. Experimental results show that for the half-bridge single-ended digital classD audio power amplifier with the corresponding method, its PSRR increases from6dB to45dB, and its PS-IMD decreases from-12dB to-50dB (the frequency of the power supplynoise is50Hz); the PS-IMD of the H-bridge differential digital class D audio power amplifierwith the corresponding method decreases from-6dB to-45dB (the frequency of the powersupply noise is50Hz). Therefore, the methods can make the digital class D audio poweramplifiers have a better power supply noise immunity capability.