Delta Sigma Modulators with Beamformers for Application in Medical Ultrasound Imaging

Abstract

Portable medical ultrasound solutions require a high degree of compactness and low power consumption, without sacrificing image quality. In this dissertation we investigate the feasibility of employing a continuous time delta sigma (CTDS) analog-to-digital converter (ADC) based receive architecture to simplify the front-end. An important step towards the design of a high performance CTDS ADC is to gain an overview of the trade-offs comprising the error sources involved, and how to decide each source’s contribution in the overall design specification. We provide a systematic design methodology using the Simulink environment in Matlab. With an example design specification, we show how each error source can be modeled, and determine its impact on the final performance through high-level simulations. In the quest of reducing the power consumption of a CTDS modulator’s internal high frequency sampling quantizer, we explore the use of voltage controlled oscillator (VCO) as a replacement. Because its use complicates excess loop delay compensation, we propose applying a capacitive summing solution to the preceding integrator. A prototype fifth order CTDS modulator is designed and implemented in 180 nm CMOS, and achieves a measured performance of 76 dB SNDR across 10 MHz bandwidth consuming 58 mW of power. For the special case of cardiac ultrasound, we give design considerations for per-channel digitization and beamforming. Further, we present the design of a third order single-bit CTDS modulator with an integrated mixer especially tailored for this application. Manufactured in 65 nm CMOS, it demonstrates 67.4 dB SNDR across 1 MHz bandwidth consuming only 131 μW of power, corresponding to a figure of merit of 34.2 fJ/conversion step. By adjusting the phase of each mixer, we show how to realize a highly effective digitization and narrow-band beamforming structure. Finally, we step into the digital beamforming domain and provide an analysis of the distortion mechanism in delta sigma beamforming. From a per-channel consideration, analytical equations are derived describing that during sample repetition an alias/imaging process occurs. Using the same analytical approach, the insert zero compensation method is shown to provide first order attenuation of the alias/images. Furthermore, by combining the insert zero with a low complexity filter, an architecture results that achieves power-efficient beamforming without any penalty to the dynamic range.