ORIGINAL ARTICLE Comments on “A novel high input impedance front-end for capacitive biopotential measurement” Ramon Pallàs Areny 1 Received: 31 August 2018 /Accepted: 13 November 2019 /Published online: 13 December 2019 # International Federation for Medical and Biological Engineering 2019 Abstract A front-end for biopotential sensing in wearable medical devices has been recently proposed which is claimed to provide 100 GΩ input impedance by manually matching two resistor pairs in a positive- and a negative-feedback loop around an operational amplifier (op amp); the cost being that the equivalent input noise voltage doubles with respect to a simple non-inverting amplifier. The ECG acquired with capacitive (sic) electrodes through a cotton shirt is presented as a proof of the performance of the proposed circuit. It turns out, however, that the analysis ignores op amp’ s input capacitance hence the effort to achieve a very high input resistance seems futile. Further, cotton is highly hygroscopic hence not an appropriate dielectric, so that there is no proof that the electrodes tested were actually capacitive. This comment addresses these two problems and some additional conceptual and methodological inaccuracies found in the paper. Keywords Biopotential sensing . Capacitive electrodes . Voltage-loading effect . Noise analysis 1 Impedance model for capacitive electrodes At any given frequency, electrode impedance, the same as the impedance of any other material, can be modeled by a resis- tance R s in series with a capacitance C s or a resistance R p shunted by a capacitance C p . For electrodes wherein DC can flow through, the parallel model seems more appropriate be- cause, in the series model, capacitance C s blocks DC. The relationship between the parameters of the two models is R s ¼ R p 1 þ ωR p C p À Á 2 ¼ R p 1 þ ω=ω p À Á 2 ð1Þ C s ¼ C p 1 þ 1 ωR p C p À Á 2 " # ¼ C p 1 þ 1 1 þ ω=ω p À Á 2 " # ð2Þ where ω p = 1/τ p =(R p C p ) -1 . These equations show that, even if R p and C p are constant within a broad frequency range, the value of R s and C s will change at each signal frequency being consid- ered inside that range. At (angular) frequencies smaller than ω p , R p and R s and C p and C s will be very close. Therefore, a small R s at a given frequency requires a very small ω p as compared with that frequency, hence a very large τ p ; a large R p is not enough to guarantee a small R s . Consequently, the statement in [1] “If the equivalent model between the body and the electrode is simpli- fied as a coupling capacitor, the equivalent input impedance of the post front-end should be at least 100 GΩ to detect 0.1 Hz low frequency signal for the coupling capacitance as low as several pF” needs some discussion relative to the electrode model and to the ECG signal frequency. With regard to electrode impedance, R p would become re- dundant if its impedance were much larger than that of C p at the signal frequency ω, i.e., R p >> 1/ωC p , hence ω p << ω. With re- gard to ECG signal frequency, its fundamental component is usually higher than 0.5 Hz (30 beats per minute). The 0.05 Hz corner frequency in some ECG diagnostic standards is intended to prevent waveform distortion. Therefore, electrode impedance at 0.5 Hz will be capacitive whenever τ p = R p C p >> 1 s/π = 318 ms, which is quite large for biopotential electrodes. For example, a cotton-based “non-contact electrode” described in [2] had 305 MΩ shunted by 34 pF, hence τ p = 10.4 ms (ω p ≈ 96 rad/s), which means that its reactance at 0.5 Hz (3.14 rad/s) is more than 30 times larger than its resistance, and at 10–15 Hz, wherein most of ECG power is, electrode reactance and resis- tance will be close. Then, from Eq. (1), it follows R s ≈ R p /2; hence, it cannot be neglected at all. * Ramon Pallàs Areny 1 Department of Electronic Engineering, Universitat Politècnica de Catalunya (BarcelonaTech), Barcelona, Catalunya, Spain Medical & Biological Engineering & Computing (2020) 58:267–269 https://doi.org/10.1007/s11517-019-02083-x