# CHUNGUS PROVENANCE LEDGER — what is real, what is read, what is made up
Living document. Updated at EVERY ship (COORDINATION.md rule 4). The coherence
gate asserts this file exists and that every ASSERTED item below is current.
An item leaves this table only by being solved, never by being forgotten.

Status: SOLVED (primary source / derived and checked) · READ-FROM-SOLVER
(computed by a kernel, page just displays it) · ASSERTED (we made the shape up
— treat as decoration until solved).

## ASSERTED — the dubious list (read this first)
| # | Item | What's fake about it | Path to real | Since |
|---|------|---------------------|--------------|-------|
| A2 | K ladder weights at high KCl | Boltzmann K(4p)/Na(3p) proxy at single T_exc; spectrum panel says "K synth pending". η at K-rich mix is provisional | K synthesis in kernel (cross-species pooling Na*+K, two-radiator RTE) — task #15 lineage | v6 |
| A3 | Absolute radiance scale (W) | EMIT_AREA·scale anchored to deGroot phenomenology fits, not our machine. Spectrum SHAPE is defensible; the watts multiplying it are not | Living Optics camera calibration (band-ratio + radiance map) | v1 |
| A4 | Reject temperatures T_pv=330 / T_exh=600 / wall model | Sankey Ṡ=Q/T entries use assigned node temps | Sol [CALC] / measurement | v5 |
| A5 | f_flue = 0.045 exhaust floor + low-n closure | [CALC] from stoich H₂/O₂ flue enthalpy — derived but unmeasured; kernel still saturates radiant fraction at 1 at EVERY n_alk (page floor patches the average). USER-VISIBLE 2026-08-25 [OP]: at n_alk→1e20 the thin gas still 'radiates' full Pin through K NIR lines → η inflates to ~41% — artifact, not physics; MPP argmax optimizes over this distorted low-n landscape | Kernel-side exhaust closure (ASK to Sol, COORDINATION.md) | v6.1 |
| A9 | Recuperator wall solver params | recup.js SOLVES Tw(z) (damped Newton, closure ≤2e-5·Pin): absorbed cavity radiation + film-limited emissivity + carrying/supply-limited salt evaporation [OP 2026-08-25 three-term spec]. ASSERTED params: turns=3/Dh=3mm/tWall=2mm (sweep: FLAT — radiation-dominated), eps_film=0.2 (molten salt vis/NIR-transparent), alpha_evap=0.1, wetFrac=0.15, cavity T_ref=1500K, mdotSaltMax=saltSup slider (THE live design knob — delivery tracks it linearly; design surface is the chem loop, not roll geometry). v6.8 UNIFIED: solver IS the recuperator — ledger rec=solved Qc, roll intercept Qroll debited from view channel, salt-loop row=Qevap (condensation return open), Tcore preheat=0.11·(TcOut−300) [T Cantera anchor], recup slider=cold-side effectiveness coldFac. Residuals: cavity T_ref 1500K fixed, per-node resWall ~1e-2 (energy closure exact, node residual not at 1e-6), condensation/Cl loop open | JANAF eval >1400K; film-ε source; cavity T_ref from solved field; salt loop closure | v6.8 |
| A7 | Axial emission ∝ local release | FIXED v6.5: cell axial profile now reads the SOLVED β release law w(fr)=B·e^(−B·fr)/NB (β=0.3 near-uniform, matching [OP] "persists down most of the tube"). Old render drew the kernel core→wall SHELL coordinate as z → fake 32× spread (operator caught it). Remaining asserted: emission ∝ local release, first-order — no axial RTE coupling | Axial RTE / 2-D (r,z) solve (na_grid2d family) | v6.5 |
| A8 | Vmpp one-diode params | FIXED v6.5: Vmpp DERIVED = 0.626[T STC] + nkT/q·ln(C) + (−1.9mV/K)[T]·ΔT — old hand-fixed 0.68 was exactly this at design point (58 suns, 330 K), now moves with Pin/mix (operator: "voltage is most definitely a function of intensity"). Remaining asserted: ideality n=1, cell T=330 K (A4), lumped C (per-ring ~90 mV spread not modeled) | Per-ring Vmpp; measure cell T; datasheet ideality | v6.5 |
| A6 | Sub-bandgap recycle R_bg≈0.94 [CALC] | Now DERIVED: R_mirror·exp(−2P·Σα_FCA·t), α_FCA=1.8e-18·N·λ² [T Baker-Finch/Green]. Remaining asserted: Maxeon geometry/doping params (N_dope 1e19, t 3µm, W 160µm, R_mirror 0.97, P=2). v6.8: EQE-tail double-count FIXED (1150-1250nm booked electric AND recycled; now sub-BG = w·(1−EQE) only) | Confirm params vs Maxeon datasheet / measure sub-BG R with LO camera; λ-resolved + full reabsorption loop in kernel | v6.4.1 |

## READ-FROM-SOLVER
- Spectrum shape: Voigt + self-reversal + fEsc — kernel (solve/synth)
- η, electric W, all Sankey flows — ∫L·EQE spectral integral + ledger() closure, Σ=Pin exact
- n_alk AUTO — argmax η over kernel (v6.2, no P&O)
- End-fed axial profile — kernel gradient (na_grid2d axial-front family)
- T_wall — flux balance from drawn-spectrum W (WradEff)

## SOLVED [T]
- Line positions (NIST) · Na₂ 805/882 (deGroot 3.21) · K₂ 1.096 µm (Huennekens
  1984) · H₂O 1.38 µm edge (HITRAN) · Maxeon EQE (datasheet) · quartz ε, k

## Retired fudges (do not resurrect)
- A1 Swiss-roll axial profile (v1–v6.2) → SOLVED v6.3 (2026-08-24): kernel buildPath
  β-source (physics.js) — β=3 end-fed front, β=0.3 distributed swiss injection
  (single-ended per real geometry [OP]); cosine + swissW blend deleted; both modes
  read sol.gradient. Gate asserts axialBeta follows burner + spread stable under recup.
- Display-only mix slider (v1–v5) → mix enters synth() (v6)
- Fixed ETA589/819/NIR constants → spectral integral (v6)
- P&O "MPP tracking" chasing cosmetic noise → kernel argmax (v6.2)
- Sankey stacked rectangles → real flow geometry + return loop (v6.1)
- Caption overdrawn on T-field (read as radial asymmetry) → moved out (v6.1)