Communitygithub.com

CATEular/Eularcat-PersonalWebsite

深读模拟与混合信号 IC 论文,整理可追溯的电路分析、实测指标、引用和论文配图,输出 Markdown / Obsidian 笔记,适用于 PMIC、Buck、放大器、比较器、ADC / DAC、PLL 与基准电路等文献,不用于直接替代电路设计或仿真验证

Eularcat-PersonalWebsite 是什么?

Eularcat-PersonalWebsite is a Claude Code agent skill that 深读模拟与混合信号 IC 论文,整理可追溯的电路分析、实测指标、引用和论文配图,输出 Markdown / Obsidian 笔记,适用于 PMIC、Buck、放大器、比较器、ADC / DAC、PLL 与基准电路等文献,不用于直接替代电路设计或仿真验证.

兼容平台~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/CATEular/Eularcat-PersonalWebsite/tree/HEAD/skills/analog-ic-notes

在你喜欢的 AI 中提问

打开一个已预加载此 Agent Skill 的新对话。

文档

Analog IC Notes

Transform a real paper into a reusable, traceable analog IC note that can be re-read, compared, and referenced. Follow the circuit deep-reading method from DC-DC Buck Reader, selecting analysis dimensions based on the actual paper type.

Input & Paths

Use the PDF, paper link, or connected reference tool provided by the user. First confirm the paper title, authors, year, journal/conference, and DOI. Advisor is not counted as an author.

When only an abstract is available, organize only the information within the abstract. If the full text or figures are missing, explicitly note that; do not infer circuit details.

Copy config.example.json to your own config.local.json and fill in the input, notes, and images directories, or use the environment variable ANALOG_IC_NOTES_CONFIG to point to a different configuration.

{
  "paths": {
    "papers": "./papers",
    "notes": "./notes",
    "images": "./notes/assets"
  },
  "language": "zh"
}

Relative paths are resolved relative to the configuration file directory. Supports ~ and absolute paths for the current system. Personal configuration must not be included in the shared skill package.

First resolve the actual paths. When no configuration is provided, use papers/, notes/, and notes/assets/ under the current working directory.

python scripts/paths.py --config config.local.json

Paths explicitly specified by the user in the current conversation take priority. Do not modify the original paper, do not overwrite existing notes. Before modifying an existing record, first review its content and preserve the user's analysis.

Reading & Organizing

  • First verify the source, then build the main text around the research problem, architecture, key circuits, measurement evidence, and limitations. Adapt the template to the paper; fixed number of sections is not required.
  • Use the general paper template to save bibliographic information, parameter tables, key figures, personal understanding, and open questions.
  • Distinguish between "original text from the paper", "own derivation", "simulation results", and "measured results". Reference specific page numbers, figure numbers, table numbers, or equation numbers.
  • Extract essential short original quotes and attach your own explanation. If the original sentence is not available, do not generate English-like simulated citations, and do not extensively copy the main text.
  • Record test conditions and definitions for parameters. Mark missing items as "not reported". When comparing across papers, first verify supply voltage, bandwidth, load, area definitions, and FoM definitions.
  • Maintain paper independence. Do not actively link to user's private PDKs, commercial chips, internal projects, or personal directories. Related literature must have real sources.

Analysis Dimensions by Type

Literature TypeKey Records
PMIC / Buck / Charge Pump / LDOTopology, control, supply and load, switches and drivers, efficiency, ripple, transient response, and reliability
Amplifier / Comparator / ReferenceInput/output range, gain and bandwidth, noise and offset, power consumption, stability, and PVT conditions
ADC / DACArchitecture, sampling rate, resolution, linearity, dynamic test conditions, power and area
PLL / ClockArchitecture, frequency range, phase noise, jitter definition, locking behavior, and power consumption

Only use dimensions applicable to the paper. Do not force efficiency tables and load step metrics from Buck onto other circuits.

For Buck papers, further reference the Buck deep-reading template and PMIC terminology reference. Formulas in references are only analytical clues; the specific definitions and applicability conditions must follow the original paper.

Figures & Formulas

First read the figure extraction guide, then use scripts/extract_figures.py. Dependencies are listed in requirements.txt.

python scripts/extract_figures.py paper.pdf --config config.local.json --paper-id paper-slug

Prefer image extraction based on actual PDF page rendering. Preserve original polarity, colors, and aspect ratios. Do not invert colors if the photo background is dark.

For vector circuit diagrams, use page cropping. Record the original page number and figure number. First check that figures, labels, and captions are complete, then write into the note.

Place images in a dedicated subdirectory for the paper, using relative Markdown image links. Use $...$ and $$...$$ for formulas, clearly distinguishing paper formulas from your own derivations.

Output Checklist

  • Title, authors, year, and DOI match the source. Advisor information is stored separately.
  • Every key numerical value and claim has a traceable source. No fabricated innovations, original quotes, or measurement results.
  • Figures open correctly with proper orientation and colors. Formulas and tables are readable.
  • No personal absolute paths, account information, private processes, or internal project identifiers.
  • The original paper remains unchanged. All related references are traceable. The main text is written in the language requested by the user.

Public release is only by explicit user authorization. Exclude config.local.json, local PDFs, and other personal configurations when sharing.

相关技能