β: This English translation is in beta — the Traditional-Chinese original is the authoritative version.

Build Report

This page honestly records the build results and their limitations (current state). Re-check anytime with python scripts/check_site_quality.py. (Note: this page deliberately avoids bare math dollar signs in prose so Markdown does not parse them as formulas.)

Want the version-by-version history? The full v1→v8 record (including every audit, correction incident, and deployment note in detail) has moved to its own page → Changelog. This page keeps only "the current answers."

1. How many PDFs were read?

The source folder contains 5 PDFs (scripts/extract_papers.py scans all of them and dumps plain text).

2. Title / author / year of each paper

id	Filename	Title	Authors	Year	Relation to ISF
paper_001	general.pdf	A General Theory of Phase Noise in Electrical Oscillators	A. Hajimiri, T. H. Lee	1998	Core foundation
paper_002	jitter_ring.pdf	Jitter and Phase Noise in Ring Oscillators	A. Hajimiri, S. Limotyrakis, T. H. Lee	1999	Ring extension
paper_003	BHongGenTheor-I_JSSC2019_Postprint.pdf	Injection Locking and Pulling…Part I	B. Hong, A. Hajimiri	2019	Advanced (injection)
paper_004	BHongGenTheor-II_JSSC2019_Postprint.pdf	Injection Locking and Pulling…Part II	B. Hong, A. Hajimiri	2019	Advanced (APF)
paper_005	Hajimiri_ISCS_98.pdf	Design Issues in Cross-Coupled Inverter Sense Amplifier	A. Hajimiri, R. Heald	1998	Unrelated to ISF (honestly flagged)

Important honesty note: the filename Hajimiri_ISCS_98.pdf looks like an ISF paper, but the content is actually cross-coupled sense-amplifier design — unrelated to oscillator phase noise / ISF. It is honestly explained only in the paper_005 deep-dive, and used solely as a conceptual bridge for "regeneration / positive feedback".

3. Which equations were successfully converted to LaTeX?

The core equations of [P1] (general.pdf), Eq.(1),(9),(10),(11),(12),(13),(15)–(24), were all converted to LaTeX via high-resolution page rendering → manual line-by-line comparison, and used verbatim in the teaching pages and extracted/*.json; there are also step-by-step algebraic expansions (downconversion integral, factor-8 summation, L≈½S_φ small-angle PM, jitter high-pass kernel, flicker 1/f³ corner, the three classes of Parseval terms) plus two derivation appendices (Floquet/PPV, Leeson↔ISF). See changelog (v1, v2 entries) for how these were built up over time.

4. What is the current status of each equation (verified / still TODO)?

The [P2] ring constants (Eq.16 Γrms's N-scaling, Eq.23 FOM prefactor) and the [P3]/[P4] injection & APF equations (generalized Adler Eq.(30)/(35), APF Fig.5/Eq.(18)–(22)) have all been verified verbatim against the original PDFs; the current authoritative versions on the site are:

• Eq.(16): $\Gamma_{rms}=\sqrt{2\pi^2/(3\eta^3)}\cdot\dfrac{1}{N^{1.5}}$ ⇒ $\Gamma_{rms}\propto N^{-3/2}$ ($\Gamma_{rms}^2\propto N^{-3}$; the radical covers only the constant $2\pi^2/(3\eta^3)$ — $N^{-1.5}$ sits outside it). [P2] Eq.(16), p.794.
• Eq.(23) FOM: $\mathcal{L}\approx\frac{8}{3\eta}\frac{kT}{P}\frac{V_{DD}}{V_{char}}(f_0/\Delta f)^2$. The prefactor is $8/(3\eta)$ ($\eta$ is the stage-delay proportionality constant of Eq.(14), $\approx 1$); $\gamma$ enters only through $V_{char}=\Delta V/\gamma$; the $V_T=0$ lower bound Eq.(25) is $\frac{16\gamma}{3\eta}$.
• Eq.(8)/(11)/(12): $\sigma_{\Delta t}=\kappa\sqrt{\Delta t}$ (Eq.8), $\kappa=(\Gamma_{rms}/q_{max})\sqrt{(\overline{i_n^2}/\Delta f)/2}$ (Eq.12).
• Eq.(17)/(18) per-stage noise $4kT\gamma\mu C_{ox}(W/L)\Delta V$ and Eq.(21) power $P=2\eta N V_{DD}q_{max}f_0$.
• [P3] Eq.(26) $\tilde\Gamma=\Gamma/q_{max}$; generalized Adler Eq.(30),(33); lock range Eq.(35) $\omega_L=\frac12 I_{inj}|\tilde\Gamma_1|$.
• [P4] amplitude decay $\tau_0=2Q/\omega_{osc}$; Eq.(26) ideal-LC fundamental quadrature; Eq.(27) amplitude-corrected Adler.

The evolution of these constants (including two caught misreadings, and who fixed what in which version) is recorded in the three-misreading-incident summary below and in the v3/v4/v7 entries of the changelog.

The following remain flagged ⚠️ / TODO (external literature or secondary details, not core ISF/injection physics, deliberately retained):

• External literature (not among the 5 source PDFs): formal volume/issue/page numbers for Leeson 1966, Demir et al. 2000 (PPV), Kärtner 1990 (volume/issue/page numbers and equation notation have been checked; the period/cycle-to-cycle jitter kernels have been derived in-house from first principles and Monte-Carlo verified at jitter_kernels, no longer relying on external conventions).
• [P2] exact axes of the Fig.17 symmetric-voltage plot; [P4] stage-allocation details of the dual-modulus prescaler (Sec. VIII).
• [P5] (sense amplifier, unrelated to ISF, deliberately not transcribed).
• [P4] exact defining equation and Fourier expansion of the APF (Sec. III-D, p.2127); Fig. 3 subplot captions.

Use python scripts/check_site_quality.py to scan for all TODO: markers.

5. Summary of three misreading-correction incidents

This site has had 3 incidents where an audit mistakenly "corrected" already-verified content into an error and mislabeled it "verified" — each was only caught by going back and cross-checking the original PDF at high zoom, a numerical anchor, and independent algebra:

1. Ring FOM prefactor: v2 mistakenly changed it to $8/(3\gamma)$ (mislabeled "verified verbatim") → v4 restored $8/(3\eta)$ against PDF p.796.
2. The D mapping in the Lorentzian linewidth: v3 took the variance growth rate κ² as the diffusion constant D → v5 adjudicated with MC + analytic fit and restored $D=\kappa^2/2$.
3. [P2] Eq.(16) Γrms's N-scaling: v3 misread the radical's scope and "corrected" v1's originally-correct $N^{-3/2}$ to $N^{-3/4}$ → v7 restored $N^{-3/2}$ after triple-checking the paper's body text ("$1/N^{1.5}$" wording), the $\eta=0.75$ numerical anchor, and App.B's algebra.

Full incident details (exactly what was wrong, how it was caught, and which pages were updated site-wide after the fix) are in the v2/v3, v4, v5, and v7 entries of the changelog.

6. Which figures are conceptual simulations regenerated from the papers?

All figures are conceptual simulations regenerated in Python (not bitmaps extracted from the PDFs), reproducing the mechanisms of [P1]/[P2]. See figure_index for the mapping.

7. Which figures are only toy models (not transistor-level)?

The vast majority are toy / conceptual models (explicitly flagged as not transistor-level). A few purely mathematical figures (jitter integration, Leeson↔ISF overlay, design sweeps, PLL transfer, BER bathtub) are computed from formulas and agree with the analytic expressions.

8. Which chapters are complete?

All 7 major chapters (00 Overview / 01 Paper Map / 02 Foundations / 03 ISF Core Theory / 04 Simulation Labs / 05 Paper Deep Dives / 06 Design Insights / 99 Appendix) are complete, with verified equations, step-by-step derivations, worked examples, interactive widgets, and problem sets with full solutions. See the changelog for how the content scope evolved.

9. Which chapters still have TODOs?

Core-theory TODOs have all been closed (see item 4 above). The remaining TODOs are all deliberately retained "external-literature scope" flags (e.g. the source of period-jitter kernel conventions, standard LC-VCO design lore) and transistor-level exclusion statements (see items 3–4 in Section 12), which do not affect the correctness of the core theory. Use python scripts/check_site_quality.py to scan all TODO: markers.

10. Does npm run build succeed?

Yes (Docusaurus 3.10.1, bilingual zh+en): 0 broken links, 0 KaTeX warnings. A page-by-page scan of math rendering passed (no residual raw LaTeX, no KaTeX parse errors; matplotlib dollar signs inside code blocks are expected).

Historical fixes (the rendering-bug class) are recorded in the v2 entry of the changelog.

11. Does python scripts/run_all_sims.py succeed?

Yes: all 49 simulation scripts (38 lab_* + 7 fig_*) pass, producing 56 figures in static/figures/. Key validations: the simulated Lorentzian spectrum matches theory, flattening near the carrier; Allan deviation slopes for the three FM types land precisely at −1/2, 0, +1/2; PLL optimal loop BW≈6.9 MHz, σ_t≈259 fs; numerically extracted ISF vs theoretical −sinθ max error ~0.001; white-noise S_φ matches the 1/f² line over ~3 decades; jitter integration numerical = analytic (447.9 fs).

12. Current site size and example QA

Site size: 97 pages × 2 locales, 56 figures, 49 simulations, 21 interactive components.

scripts/verify_examples.py actually runs every Python worked example in docs that has a "reference answer" (# ->) and checks the numbers: of 144 verifiable blocks, 133 pass automatically, 0 mismatches, 0 errors; the rest are verifier false positives on formula constants in comments (such as the "2" in $2\Gamma_{rms}^2$) or comparison numbers, manually confirmed correct. See the terminal output of check_site_quality.py for the latest numbers on pages / figures present / required figs missing / content issues / soft warnings / open TODOs.

See the changelog for evolution details (what was added and what bugs were fixed in each version).

Honesty and TODO principles

• Toy models are always flagged as "a pedagogical toy model, not transistor-level."
• Content from external literature (PPV / adjoint / Floquet / Leeson / Demir) is always flagged as "not among the 5 downloaded PDFs, supplemented from standard literature."
• Uncertain constants/figures/citations are always written as TODO: manual verification needed ... — never guessed to fill a gap.
• [P5] is always honestly explained as a sense-amplifier paper unrelated to ISF, used only as a conceptual side note.
• Any "correction claimed by an audit" must be triple-verified in person — original PDF at high zoom, numerical anchor, and independent algebra — before being applied site-wide. This site has 3 real counterexamples where that discipline mattered (see item 5 above); the principle is not just theoretical.
• Printing errata in the papers themselves (e.g. [P2] Fig.16's ζ=2.5e5 which should read 2.5e-5, and [P1] p.191's "58.8 fF" which should read fC) are also honestly flagged, not silently "corrected" without a note.

Suggested next steps for manual verification

1. Check the exact constants and equation forms of [P2]/[P3]/[P4] against the original PDFs → done (see item 4 above and the v3/v4/v7 entries of the changelog).
2. Fill in the formal volume/issue/page/DOI for the external literature (Leeson, Demir PPV, Kärtner, Adler) → done (see the v4 entry of the changelog).
3. For transistor-level accuracy: use Spectre PSS+PNoise/PXF or the adjoint method to extract the ISF and cyclostationary α(x) from a real LC-VCO / ring-VCO, replacing the toy models. (Still deliberately out of scope.)
4. Calibrate the absolute numbers of the interactive calculators and the toy models to an actual process.

Deployment

Publicly deployed as a GitHub Pages (project page):

• Site: https://gmcycle7.github.io/isf-teaching-site/
• Source (public): https://github.com/gmcycle7/isf-teaching-site
• baseUrl set to /isf-teaching-site/; KaTeX CSS/fonts bundled via webpack (baseUrl-safe, fully offline).
• Deployment mechanism: local npm run build, then push build/ to the gh-pages branch (with .nojekyll); Pages serves from that branch. Packaged as the one-click script ./scripts/deploy.sh (needs only repo scope, no workflow scope).
• Copyright handling: the full text and PDFs of the 5 papers are not committed (.gitignore excludes extracted/raw_text/ and *.pdf); the footer and each page state that copyright belongs to the original authors and the content is for teaching purposes.
• CI (optional): .github/workflows/deploy.yml is ready; pushing the workflow file requires a token with workflow scope (gh auth refresh -s workflow) to enable push-to-deploy.

Individual deployment incidents (e.g. how a stuck large deployment was resolved) are in the changelog.

---

Full version history (v1→v8, with every audit and correction in detail) → Changelog