| Abstract [eng] |
This master’s final degree project investigates noise in sigma-delta analog-to-digital converters, its origin, its influence on modulator operation, and the possibilities of reducing it when designing a low noise converter for biomedical signal acquisition. The work reviews sigma-delta operating principles, key performance metrics, and the main noise improvement strategies. The parameters SNR, SNDR, DR, ENOB, and FoM are discussed, and the influence of modulator order, oversampling ratio, and differential implementation on noise performance is analysed. The literature review shows that, for low frequency and high resolution applications, a second-order fully differential continuous-time sigma-delta modulator with a one bit quantizer provides a reasonable design compromise. In this work, a sigma-delta modulator with a 1 MHz sampling frequency and a 1 kHz signal bandwidth is designed using IHP SG13G2 technology. First, ideal and behavioral models are developed in MATLAB and Verilog-A, followed by the DT-CT transformation and the calculation of active-RC integrator coefficients. The behavioral model shows approximately 109 dB SNR and nearly 18-bit ENOB. Next, a fully differential folded-cascode operational amplifier with common-mode feedback and a dynamic StrongARM comparator with an output buffer and RS latch are designed at transistor level. Transient and spectral analyses of the complete transistor-level modulator are then performed. The transient analysis confirms correct formation of the one bit PDM output, while the spectral analysis demonstrates noise shaping toward higher frequencies and an SNR of about 97.35 dB. This corresponds to 15.88 ENOB. Taking into account the total modulator power of 406 µW, a FoMs of 161.3 dB is obtained. The modulator layout is also designed in KLayout. Comparison with narrowband sigma-delta modulators reported in the literature shows that the designed modulator achieves competitive performance for biomedical signal digitization applications. |
