Details

Autor(en) / Beteiligte

• Chandrakumar, Hariprasad
• Markovic, Dejan

Titel

A 15.2-ENOB 5-kHz BW 4.5- \mu W Chopped CT \Delta\Sigma -ADC for Artifact-Tolerant Neural Recording Front Ends

Ist Teil von

• IEEE journal of solid-state circuits, 2018-12, Vol.53 (12), p.3470-3483

Ort / Verlag

IEEE

Erscheinungsjahr

2018

Link zum Volltext

• Volltext

Quelle

IEL

Beschreibungen/Notizen

• Implantable closed-loop neural stimulation is desirable for clinical translation and basic neuroscience research. Neural stimulation generates large artifacts at the recording sites, which saturate existing recording front ends. This paper presents a low-power continuous-time delta-sigma analog to digital converter (ADC), which along with an 8<inline-formula> <tex-math notation="LaTeX">\times </tex-math></inline-formula> gain capacitively-coupled chopper instrumentation amplifier (CCIA), realizes a front end that can digitize neural signals from 1 Hz to 5 kHz in the presence of 200-mV pp differential artifacts and 700-mV pp common-mode (CM) artifacts. A modified loop-filter is used in the ADC along with new linearization techniques to significantly reduce power consumption. Fabricated in 40-nm CMOS, the ADC occupies an area of 0.053 mm 2 , consumes 4.5 <inline-formula> <tex-math notation="LaTeX">\mu \text{W} </tex-math></inline-formula> from a 1.2-V supply, has an input impedance of 20 <inline-formula> <tex-math notation="LaTeX">\text{M}\Omega </tex-math></inline-formula> and bandwidth (BW) of 5 kHz, and achieves a peak signal to noise and distortion ratio (SNDR) of 93.5 dB for a 1.77-<inline-formula> <tex-math notation="LaTeX">\text{V}_{\mathrm {pp}} </tex-math></inline-formula> differential input at 1 kHz. The ADC's figure of merit (FOM) (using SNDR) is 184 dB, which is 6 dB higher than the state of the art in high-resolution ADCs. The complete front end occupies an area of 0.113 mm 2 , consumes 7.3 <inline-formula> <tex-math notation="LaTeX">\mu \text{W} </tex-math></inline-formula> from a 1.2-V supply, has a dc input impedance of 1.5 <inline-formula> <tex-math notation="LaTeX">\text{G}\Omega </tex-math></inline-formula>, input-referred noise of 6.35 <inline-formula> <tex-math notation="LaTeX">\mu \text{V}_{\mathrm {rms}} </tex-math></inline-formula> in 1 Hz-5 kHz, and total harmonic distortion of −81 dB for a 200-mV pp input at 1 kHz, and is immune to 700-mV pp CM interference. Compared to front ends intended for closed-loop neural recording, this paper improves the linear input range by 2<inline-formula> <tex-math notation="LaTeX">\times </tex-math></inline-formula>, the signal BW by 10<inline-formula> <tex-math notation="LaTeX">\times </tex-math></inline-formula>, the dynamic range by 12.6 dB, the FOM by 12.4 dB and remains immune to large CM interference while maintaining comparable power, area, and noise performance.