Quantization Noise Minimization in $\Sigma\Delta$ Modulation Based RF Transmitter Architectures

This paper describes an optimization method for minimization of quantization noise in ΣΔ-based RF transmitters. The aim of the method is to enable the use of reconstruction filters with wider passband, or alternatively, a lower switch-rate. The method uses a general representation of the ΣΔ converters in combination with a differentiable approximation of the quantizer. Based on this, a Monte Carlo-based algorithm is developed around the damped Gauss-Newton iteration. As a result of the suggested algorithm, the residual quantization noise after reconstruction filtering is significantly decreased. Finally, simulations using a bandlimited signal with a Gaussian distribution are used to demonstrate the capabilities of the suggested algorithm when applied with the proposed ΣΔ modulator representation. The resulting performance is compared to several cases of ΣΔ converters designed using traditional methods, demonstrating significant improvements in terms of reduced reconstruction normalized mean square error. This implies that the transmitter efficiency can be improved with minor changes in the modulator implementation.