**1. Introduction**

Currently, an increase of the life expectancy in the population of developed countries is taking place. Therefore, new habits for healthy lifestyles are being adopted, many of them trying to implement preventive health programs and early detection of diseases, as the most effective way to improve the effectiveness of treatments and therapies and ensure, as far as possible, a high quality of life and a healthy aging. The Internet of Things (IoT) that allows data to be collected and analysed at any time and from anywhere, is called to play a fundamental role to offer a solving strategy in healthcare [1]. In IoT-based healthcare, sensors and devices are developed for a variety of objectives, such as monitoring the medical conditions of people, assisting in the treatment of diseases, and providing access to patient information. In this context, wearable devices are seamlessly connected to improve information delivery and the care-giving process in healthcare services [2]. Given the large-scale challenges caused by chronic diseases, very low cost and effective wearable devices for telemedicine have become of higher importance.

Electrical bioimpedance (EBI), or simply bioimpedance, joins the attributes to become a promising sensor technology in the IoT environment. EBI is a well-established physical concept in which an object's impedance to an applied alternating current over increasing frequencies can be measured, to assess tissue composition [3]. In addition to being economic, lightweight, easy-to-use, and noninvasive, bioimpedance can be used for a wide range of clinical applications, ranging from examine body composition in healthy people to monitoring various types of diseases such as diabetes, hypertension, and others. Therefore,

**Citation:** Corbacho, I.; Carrillo, J.M.; Ausín, J.L.; Domínguez, M.Á.; Pérez-Aloe, R.; Duque-Carrillo, J.F. A Fully-Differential CMOS Instrumentation Amplifier for Bioimpedance-Based IoT Medical Devices. *J. Low Power Electron. Appl.* **2023**, *13*, 3. https://doi.org/ 10.3390/jlpea13010003

Academic Editors: Orazio Aiello and Andrea Acquaviva

Received: 28 October 2022 Revised: 7 December 2022 Accepted: 25 December 2022 Published: 30 December 2022

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

in recent years, a pronounced trend towards the integration of EBI in wearable systems has been observed.

In practice, for detecting some transient physiological events, bioimpedance spectroscopy (BIS) is used. As with any spectroscopy technique, BIS implies the measurement of the bioimpedance spectrum in a determined frequency range, for which a sequential sweep of analysis varying the frequency is carried out. Typical frequencies in BIS are in the range from several hundreds of Hz to 1 MHz, also known as the *β*-dispersion range. Therefore, the use of such a broad signal spectrum puts several challenges for the full integration of wearable bioimpedance-based devices into the clinical health care system. In particular, a CMOS integrated BIS system in the IoT horizon requires a great circuit optimization not only in size but also in energy consumption.

The block diagram of a bioimpedance-based IoT system for medical applications is illustrated in Figure 1. The source of power, which can be a battery or an energy harvester, is controlled by a power management unit (PMU), which optimizes and regulates the signals used to supply the rest of the blocks. The bioimpedance under test, *ZBIO*, is excited by an AC signal, usually a current in order to avoid any damage on the biological sample, and the resulting voltage is acquired and conditioned by the analog front-end (AFE). Then, signals are efficiently processed in the digital domain, by a digital signal processor (DSP), and can be locally stored or transmitted by means a wireless protocol. The user interface allows control of the operation of the overall system.

**Figure 1.** Conceptual block diagram of a bioimpedance-based IoT system for biomedical applications.

The IA is a critical constituent block of the system previously described [4–29]. Indeed, an appropriate signal acquisition is required, which includes a demanding performance in terms of differential-mode (DM) signals amplification, common-mode (CM) signals rejection, and noise, among others, whereas the overall power consumption has to be kept to a minimum extent, which is particularly a challenge in applications that require the processing of signals contained in a wide frequency range and with a relatively large amplitude. The indirect current feedback (ICF) technique results suitable to design a monolithic IA with low-voltage capability [5,22,29]. In addition, a single-stage ICF IA provides compactness and the possibility of achieving operation over a broad frequency range [11,12,22,26,29].

The overall performance of an analog system in general, and of an IA in particular, can be enhanced by adopting a fully differential (FD) implementation [23,25,30]. There are well-known advantages associated to this solution, such as the extension of the signal range, due to the availability of two output terminals, the increase of the linearity, thanks to the ideal cancellation of even-order harmonics, and the decrease of the effects of undesired noises coming from the supply, which can be considered as CM signals. There are also disadvantages related to the use of a FD circuit, such as the increase of the circuitry to obtain a fully symmetrical structure, with the consequent increase in area and power consumption, or the need of a CM feedback (CMFB) network, to control the CM component of the output signal. Therefore, all the pros and cons must be considered and a design tradeoff has to be established.

A FD IA, relying on the ICF technique and suitable for bioimpedance analysis in an IoT biomedical application, is presented in this contribution. An analysis of the main characteristics of the proposed circuit is provided, which is confirmed by means of simulated

and experimental results. In addition, the solution is compared in terms of circuit structure to other differential IA previously reported [29], whereas a performance comparison with similar solutions in the literature is also carried out. The circuit has been designed and fabricated in 180 nm CMOS technology to operate with a single-supply voltage of 1.8 V. The experimental characterization illustrates the robustness of the proposed solution. The rest of the manuscript has been organized as follows. Section 2 deals with the block diagram and the transistor level implementation of the IA, whereas different design considerations are discussed in Section 3. Measurement results are reported in Section 4 and conclusions are drawn in Section 5.
