科研 Skill

神经记录前端电路设计

neuro-frontend-design

Designs or reviews analog front ends for neural recording headstages, covering noise budget, CMRR and ADC resolution. Use when choosing an amplifier or debugging a noisy front end.

发布者 Codar安全审核通过v0.1.0更新于 2026-11-18

SKILL.md

Neural recording front-end design

Start from the signal (amplitude and band) and the electrode (impedance and offset), budget noise at the input, then choose gain, filters and ADC so the amplifier, not the ADC or the electrode, sets the floor.

0. Load the tool reference

Read references/chips-and-tools.md: ADS1299 and Intan RHD2000 specifications taken from the current datasheets, ADS1299 register configuration, simulation tools (KiCad/ngspice, LTspice, TINA-TI) and layout rules for microvolt inputs.

1. Requirements

SignalTypical amplitudeBand
EEGabout 10-100 uVabout 0.1-100 Hz
ECoG / LFPabout 50 uV-1 mVabout 0.1-300 Hz
Spikesabout 50-500 uVabout 300 Hz-6 kHz
EMG (surface)about 0.1-5 mVabout 20-500 Hz

Set an input-referred noise target well below the smallest signal (often a few uV rms or less across the band) and a DC input range that covers electrode offsets (hundreds of mV are possible with wet electrodes).

2. Noise budget with the bundled script

python scripts/noise_budget.py --en-nv 30 --fc-hz 10 --in-fa 10 --z-mag-kohm 10 --z-real-kohm 5 \
       --band 0.5 100 --gain 24 --adc-bits 24 --vref 4.5

Inputs: amplifier voltage noise density (nV/sqrtHz) and its 1/f corner, current noise (fA/sqrtHz), electrode impedance magnitude and real part in the band, bandwidth, gain, ADC bits and reference. It integrates each source over the band and prints input-referred uV rms per source, the total, approximate peak-to-peak (x 6.6), and which source dominates.

3. Design rules

  • Input impedance far above electrode impedance (100x or more) to limit attenuation and common-mode-to-differential conversion from impedance mismatch.
  • CMRR depends on electrode impedance mismatch as much as on the chip; use a driven right-leg or bias electrode.
  • AC-coupled or wide-input-range DC-coupled inputs to survive offsets; check settling time after saturation.
  • Anti-alias filtering matched to the ADC (delta-sigma converters relax the analog filter but check their digital filter latency for real-time use).
  • Choose gain so the largest expected input plus offset uses most of the ADC range without clipping; then check ADC noise referred to input is below the amplifier's.
  • Reference chips: ADS1299 (EEG class, 24-bit) and Intan RHD2000 (spike class, about 2.4 uV rms input-referred) are common baselines; quote datasheet values for your configuration.

Output

Requirement table, noise budget per source with total, gain and ADC choice, input range and impedance check, and the dominant noise source with how to reduce it.

示例提示词

帮我给这款EEG放大器算噪声预算,选ADS1299的增益和采样参数。

所属插件

神经科学面向神经科学研究:EEG/MEG、fMRI、BIDS、spike sorting、连接组、脑机接口解码评估、神经刺激安全与公开神经数据检索。

安装该插件会一并获得同领域的 45 个 Skill / MCP。

同插件的其他科研 Skill

常见问题

神经记录前端电路设计能做什么?

Designs or reviews analog front ends for neural recording headstages, covering noise budget, CMRR and ADC resolution. Use when choosing an amplifier or debugging a noisy front end。

如何安装神经记录前端电路设计?

在科搭 Codar 左侧点击「插件」,在「组件」页签搜索「神经记录前端电路设计」并安装;在 Claude Code 或 Codex 中可以发送本页的安装提示词,或下载 Skill 包后解压到 ~/.claude/skills 或 ~/.codex/skills。

神经记录前端电路设计需要哪些权限?

该科研 Skill没有声明额外权限;执行具体操作时仍由 Agent 按自身规则请求确认。

神经记录前端电路设计经过官方审核吗?

是。SpeedSkills 收录的每个版本都会记录内容摘要与权限声明,当前版本已通过安全审核。