Contents I. Introduction
Successive Approximation Register (SAR) analog-to-digital converters (ADCs) are extensively used in low-power applications, particularly in the realm of biomedical implants, well-documented in [1]. As illustrated in Fig. 1 (a), closed-loop biomedical device components can be categorized into front-end and back-end devices [2]. Several publications, [1], [3], [4], have delved into the advancements in the architecture of biomedical implants, emphasizing the pivotal role of ADCs, which is present in both the front-end and back-end sections. Figure 1 (b), which are real data from human brain slices, depicts how non-idealities can deteriorate a signal, specially in biomedical circuits. Substantial efforts have been devoted in the field to analyze and model, and mitigate the non-ideal factors associated with SAR ADCs, with the aim of enhancing their speed and accuracy [5]–[10]. The primary non-idealities affecting the performance of SAR ADCs are often related to the analog components of this converter, DAC and comparator. Notably, the behavior of channel charge injection, among these non-idealities, remains roughly discussed and inaccurately evaluated to the best of our knowledge. Existing works employ binary-weighted dummy switches to alleviate the consequences of this phenomenon, and they presuppose that the entirety of the flowing charge is absorbed by the dummy switch, despite the lack of corroborating evidence [11], [12]. A recent study in [7] underscores the significance of comprehending the impact of every non-ideality on ADC performance by evaluating the input-referred noise of the comparator in SAR ADCs. On the other hand, in [13] we see the application of channel charge injection exploited to implement a physically onclonable function. These are our motivations to accurately extract the value of this phenomenon.
