A noise-cancelling technique for data acquisition systems: A case study on embedded ADCs

The noise present in embedded analog-to-digital converter (ADC) of microcontrollers (MCUs) and other System-on-chips (SoCs) is a limiting factor that can impede their use in some application. This noise can be caused due to the proximity between ADCs and digital circuits whose switching couples to the analog signals and degrades the input signal integrity. Moreover, many models of microcontrollers have more than one ADC input channel, and their inputs suffer impairments due to those noise sources. Because the noise source can be the same for all the inputs, the noise is correlated and it can be exploited to increase the signal quality in the digital domain. This paper proposes a novel adaptive noise canceler based on the normalized least mean squares (NLMS) adaptive noise cancellation algorithm. In the proposed technique, three analog inputs of the microcontroller are used, where two are used as reference and one is used for acquiring the desired signal. An experimental setup was implemented for testing the proposed scheme and the results show improvement of more than 20 dB in the noise floor, in cases of noisy power supply.