h2bench

# Review note: H2Bench

## Session 2026-09-25: recommendations accepted

On 2026-09-25 Amish wrote: "i accept all your recommendations, go with them across all repos." Every item with a recommendation is now "Decided by Amish, 2026-09-25: go with recommendation", recorded in `docs/decisions/0002-recommendations-accepted.md` (HBN-DDR-002 v0.1). HBN-DDR-001 moved to v0.2 with the same wording.

### Decisions applied and what changed

- **D1 to D8 (wording only):** rigid 2 L tank, PEM, generic 12 W fuel cell, gas-law measurement, bench supply as the source, H2Guard costed separately and shipped with the bench, tank kept above 20 kPa gauge with purging, vocational training first, canopy hood with a single high point.
- **D9 and D10, budget:** `budget_usd` $450 to $850 in `project.yaml`, with the bench power supply in the kit (BOM line 4). R14 restated as $850 including the supply. README cost lines updated.
- **D11, relief:** 350 to 325 kPa gauge. Inventory at full lift and 15 °C 10.5 to 9.9 L (R5 at risk to met); at the relief setting 9.6 to 9.0 L; stored pressure energy at the relief setting 776 to 710 J; room release 0.032 to 0.030 % by volume.
- **D12, supply cut:** a pressure switch on the manifold opens at 310 kPa gauge and a relay (rated 10 A DC or more) breaks the electrolyzer supply, hardwired, independent of the logger firmware. Membrane load at the relief setting 357 to 332 kPa (317 kPa at the cut); R10's target restated from 300 to 332 kPa. BOM line 10 $45 to $60, line 14 $30 to $35: total $865 to $885 ($800 to $820 without the supply). Mass 25.2 to 25.3 kg.
- **D13 and D14:** vent needle valve 3 min or more and the plug-in RCD (line 18) confirmed; already in the design.
- **D15, H2Guard interface:** listed below as a cross-repo action.
- Files: `cad/src/model.py` (cut switch on the manifold; STEP and STL re-exported, clash check 0), `cad/src/sheets.py` (HBN-DWG-001 Rev P1 to P2: relief, cut, mass notes), `cad/src/concept_media.py` (key figure), `docs/04-calcs/sizing.py` and HBN-CAL-001 v0.2 with `results.csv`, HBN-PRB-001 v0.4, HBN-PRC-001 v0.4, HBN-REQ-001 v0.4, `bom/bom.csv`, `bom/bom-notes.md`, `README.md`. All PDFs, drawings and media regenerated; hero, blueprint, exploded and drawing checked by eye; the footer now reads designmolecule.com.
- README: "What sparked the idea" rewritten around William Grove's 1842 gas voltaic battery, whose current he used to electrolyze water; the earlier text about a portfolio review was removed. No pitch or problem rewording was recommended, so none was applied.

### Requirement status (HBN-CAL-001 v0.2)

Met 10 (was 9), not met 2, at risk 3 (was 4), not verifiable at TRL 3 1.

| ID | Result | Status |
| --- | --- | --- |
| R9 | Oxygen crossover unknown, no deoxidizer; fuel cell listing asks for 99.995 % | **Not met** |
| R14 | $885 full, $820 without the supply, $800 minimum; against $850 | **Not met** |
| R8 | 1.1 s trip; H2Guard valve 24 V against the 12 V solenoid; supply-break output undefined | At risk |
| R10 | Stack must hold 332 kPa over the O2 side; no rating in hand | At risk |
| R13 | 25.3 kg against 25 kg (22.3 kg without the supply) | At risk |
| R16 | Vent 3.0 min; setup needs hardware | Not verifiable at TRL 3 |
| R1 to R7, R11, R12, R15 | R5 now 9.9 L worst case | Met |

### Still awaiting Amish (no recommendation was made)

- Curriculum and age group for the first worksheets.
- Class demonstration or group practical first.
- Aluminium or stainless tank.
- Mass (R13): accept 25.3 kg, carry the supply separately, or a 10 mm deck.
- Cost (R14): where to find the $35 over the $850 budget. **Decided by Amish, 2026-09-26: budget set to $885 to cover the priced BOM; see "Session 2026-09-26: budget approved".**

### Cross-repo actions

- **H2Guard** (not edited from this repo): settle (1) the valve voltage, 24 V in H2Guard against H2Bench's 12 V tank solenoid; (2) which H2Guard output breaks H2Bench's 9 A DC supply line (the new 310 kPa gauge cut relay could share that break point); (3) hood extraction, 60 m³/h through the H2Bench canopy against H2Guard's 150 m³/h room exhaust; (4) first user, university teaching lab in H2Guard against vocational training here.

### TRL

`trl: 3` and `trl_target: 3` are unchanged. TRL 4 remains on hold by Amish's instruction: no build, test, purchasing, supplier trials, PCB or firmware beyond a sketch was started.

## Session 2026-09-25: TRL 3

### What was done

- `docs/decisions/0001-trl2-review-decisions.md` (HBN-DDR-001 v0.1): the eight TRL 2 recommendations adopted for TRL 3 work under Amish's 2026-09-25 instruction, open for his review, and the items that stay open.
- `docs/04-calcs/01-sizing.md` (HBN-CAL-001 v0.1) with `docs/04-calcs/sizing.py` and `docs/04-calcs/results.csv`: electrolyzer and fill, inventory and pressure, fuel cell, energy balance, measurement uncertainty, hood and venting, interlock, water and drier, envelope and mass, cost and logging, with a results table for R1 to R16. The script reads the tank and bench dimensions from the model and the prices from the BOM.
- `cad/src/model.py`: parametric build123d model (frame, deck, back panel, canopy with duct, supply, reservoir, 4-cell electrolyzer with membranes and ports, separator and drier, arrestor, 2.000 L tank sized from its volume parameter with guard, manifold with relief and vent needle valve, regulator and solenoid, 75 x 47 x 70 mm fuel cell, load, meters, H2Guard stand-in, 6 mm gas lines). Exports `cad/step/` and `cad/stl/` (assembly, frame and hood, tank module, electrolyzer); clash check finds no overlaps.
- `cad/src/sheets.py` and `cad/drawings/HBN-DWG-001.svg`, `.pdf`, `.png`: general arrangement at Rev P1, marked "CONCEPT, NOT FOR FABRICATION" and "PRELIMINARY, NOT FOR FABRICATION". HBN-DWG-001 was free because the concept blueprint uses HBN-DWG-010.
- `bom/bom.csv` (18 lines, all priced with a supplier type) and `bom/bom-notes.md`. Line 18 (plug-in 30 mA RCD, $20) is new; lines 6, 10, 12 and 16 were updated.
- `cad/src/concept_media.py` now builds from `model.py`; all media regenerated (hero, blueprint, cutaway, exploded, flow, `model.glb`, `viewer.html`), checked by eye, and the temporary `media/_views*` folders deleted. The kit's cutaway cuts at the mean Y of all parts, which missed the gas train, so the script renders its own cutaway on the gas train centreline.
- `docs/01-problem.md`, `docs/02-concept.md` and `docs/03-requirements.md` moved to v0.3 with the adopted choices, HBN-CAL-001 numbers and a status column from the CAL.
- `project.yaml`: trl 3, trl_target 3, trl_evidence updated; `budget_usd`, pitch and problem unchanged. `README.md`: TRL badge, links, Concept numbers and cost line.

### Requirements (HBN-CAL-001)

Nine met, two not met, four at risk, one not verifiable at TRL 3.

| ID | Result | Status |
| --- | --- | --- |
| R9 | Oxygen crossover unknown, no deoxidizer; fuel cell listing asks for 99.995 % | **Not met** |
| R14 | $865 full, $800 without the bench supply, $780 minimum; against $450 (and $850 proposed) | **Not met** |
| R5 | 9.6 L at the relief setting, but 10.5 L at full relief lift on a 15 °C day | At risk |
| R8 | 1.1 s trip to valve closed from H2Guard's TRL 2 figures; 24 V and 12 V valve mismatch | At risk |
| R10 | Stack must hold 357 kPa over the O2 side at the relief setting; no rating in hand | At risk |
| R13 | 900 x 450 x 755 mm, but 25.2 kg against 25 kg (22.2 kg without the supply) | At risk |
| R16 | Vent 3.0 min; setup needs hardware | Not verifiable at TRL 3 |
| R1, R2, R3, R4, R7 | ±1.8 % worst stage; 46.7 min cycle; 70.2 W; 10.8 W net; 0.8 % LFL in 30 m³ | Met |
| R6, R11, R12, R15 | Pressure ratings, water quality, logging, guard | Met (design review) |

Corrections to TRL 2 numbers: the 7.9 L "at the relief setting" was the tank alone at 300 kPa gauge; with the lines it is 9.6 L at the relief setting and 10.5 L at full lift. Mass rises from about 18 to 25.2 kg. Cost rises from $845 to $865 with the RCD line. The fuel cell's 146 mL/min was the reacted flow; it draws 154 mL/min. Bench height is 755 mm, not about 770 mm. R10's 300 kPa target does not cover a relief lift (357 kPa needed). R4 has only a 0.8 W margin.

Key numbers: 70.2 W electrolyzer, 256 mL/min, 17.7 min fill of 0.189 mol (4.5 L); 10.8 W net fuel cell for 29.0 min, 5.2 Wh; round trip 25.1 % (HHV); 646 J stored pressure energy at 300 kPa gauge; hydrogen per fill measured to ±1.5 %; drier lasts about 110 fills.

### Decisions recorded (HBN-DDR-001)

Decided by Amish, 2026-09-25: go with recommendation (HBN-DDR-002; recorded in this session as adopted for TRL 3, open for his review): rigid 2 L tank at up to 300 kPa gauge (gasbag with pump as the fallback if R10 fails); PEM, not alkaline; generic 12 W class fuel cell with a polarization curve as a purchase condition; hydrogen measured by pressure, volume and temperature; bench supply in constant-current mode as the source; H2Guard costed in its own project and shipped with the bench (applied to the R14 scope); tank kept above 20 kPa gauge with pressure-cycle purging; vocational technician training as the first users. No pitch or problem rewording was recommended, so none was applied.

### Still awaiting Amish (as of this session; see the recommendations-accepted session above for the current status)

- **Budget:** Decided by Amish, 2026-09-25: go with recommendation, `budget_usd` $850 (HBN-DDR-002 D9). At the time: about $850 recommended at TRL 2; `budget_usd` stayed $450. The full kit is $865, $15 over even the recommended figure; without the bench supply it is $800.
- Whether the bench power supply is in the kit (BOM line 4). Now Decided by Amish, 2026-09-25: go with recommendation: in the kit (D10).
- Canopy hood with a single high point (not in the TRL 2 review list, so not adopted; kept as the working design). Now Decided by Amish, 2026-09-25: go with recommendation (D8).
- Curriculum alignment, class demonstration or group practical first, and aluminium or stainless tank (no recommendation made). Still proposed, awaiting Amish.
- Engineering proposals from this session: relief at 325 kPa gauge (329 kPa gauge or less keeps R5 at full lift); a 310 kPa gauge cut of the supply so the relief valve is a backup; vent needle valve set for 3 min or more; plug-in RCD line. Now Decided by Amish, 2026-09-25: go with recommendation (D11 to D14).
- Mass (R13): accept 25.2 kg, carry the supply separately, or use a 10 mm deck (saves 0.77 kg). No recommendation; still proposed, awaiting Amish.

### Cross-repo notes (H2Guard, not edited)

H2Guard is not one of the batch's shared components, but H2Bench depends on it, so its REVIEW.md and documents were read. Consistent: H2Guard's trip to valve closed within 2 s (HGD-REQ-001 R4) matches R8, and its inventory rule (300 L for 30 m³) is met thirtyfold. Conflicts to resolve with H2Guard: (1) its valve is a 24 V normally closed part, while H2Bench's tank solenoid is 12 V; (2) it is not stated which H2Guard output breaks H2Bench's 9 A DC supply line; (3) H2Guard models its fan as a 150 m³/h room exhaust, while H2Bench assumes 60 m³/h through its canopy duct; (4) H2Guard recommends a university teaching lab as first user, H2Bench vocational training.

### Safety concerns

- Hydrogen is flammable from 4 to 74 % with a 0.02 mJ ignition energy and a nearly invisible flame; the bench must never run without H2Guard detection, extraction and interlock.
- Venting a full tank faster than about 90 s can raise the duct concentration past H2Guard's warning; the needle valve is set for 3 min or more.
- At the relief setting the tank holds 776 J of pressure energy; only 1 MPa rated metal parts, guard in place.
- The relief lift loads the electrolyzer membrane to 357 to 392 kPa; an unrated stack could fail and mix hydrogen into the oxygen side. The 310 kPa gauge supply cut is proposed.
- Oxygen crossover also puts some oxygen into the stored hydrogen; far below a flammable mixture, but unquantified.
- Mains supply near water: RCD (line 18) and earthed supply. Warm stacks: about 20 W of heat each. Users under 18 only under supervision.
- Research and teaching prototype, not certified laboratory equipment.

### Other notes

- No TRL 4 material exists (`build-log/` holds only its README; `electronics/` and `firmware/` are empty). None was created.
- Citations checked with WebFetch on 2026-09-25: the Fuel Cell Store 12 W listing (7.8 V at 1.5 A, 13 cells, 0.45 to 0.55 bar, 99.995 %, 0.18 L/min, 40 %, 55 °C, 5 to 30 °C, 275 g, $576) and the Ecofin Namibia report (130,000 workers by 2040, attributed to the NUST vice-chancellor; still a news report of a statement, not a study). The MSE listing confirms a single-cell 300 mL/min stack at $1,396.95 but states no output pressure, so the TRL 2 claim "up to about 0.4 MPa" was removed from HBN-PRB-001. Generic stack prices (BOM lines 6 and 12) remain unverified estimates.
- The `model.py` assemblies copy each shape, because a build123d shape can have only one parent compound; without this the STEP export and drawing silently lost parts.

### Recommended next step

TRL 4 is on hold by Amish's instruction, so the next step is a review, not a build: Amish to decide the budget (and whether the supply is in the kit), confirm or reject the relief, supply-cut and RCD proposals, and settle the H2Guard interface points with that project. For the record only, TRL 4 would need a named electrolyzer with a back-pressure rating of at least 357 kPa and an oxygen-in-hydrogen figure, fuel cell supplier acceptance of the gas, an H2Guard design at TRL 3 with a defined interlock output, a lab test report (TST, environment: lab) and build log entries.

## Session 2026-09-25: /populate to a strong TRL 2

### What was done

- `docs/01-problem.md` (HBN-PRB-001 v0.2): problem with cited sources, users, operating environment, constraints, out of scope, prior work and open questions. No co-design checklist existed, so none was kept.
- `docs/03-requirements.md` (HBN-REQ-001 v0.2): 16 measurable requirements (R1 to R16) with targets, estimated status and assumptions.
- `docs/02-concept.md` (HBN-PRC-001 v0.2): lesson sequence, 15 numbered components, energy and gas balance, nine proposed design choices, safety section and open questions.
- `cad/src/concept_media.py`: massing model of the bench (15 BOM parts, plus the electrolyzer cell block shown as unnumbered parts) on an existing lab table, with a 1.75 m person placed on the floor as a context part.
- `media/`: `hero.png`, `concept-blueprint.png`, `.pdf` and `.svg`, `model.glb` and `viewer.html`, `exploded.png` with callouts 1 to 15, `cutaway.png` (electrolyzer membranes and hollow tank), and `flow.png` (energy per lesson, all values labeled estimates).
- `bom/bom.csv`: 17 rows, numbered to match the exploded view (rows 16 and 17 are not modeled), with indicative USD prices; `bom/bom-notes.md` updated.
- `README.md`: hero image and links line; Concept rationale, Burning platform, both "Where it could be used" tables and "What sparked the idea" expanded with cited figures; Problem, Concept, Key components and Safety brought in line with the concept.
- `docs/pdf/`: branded PDFs of the three controlled documents.
- `project.yaml` is unchanged. The pitch ("a few liters of hydrogen at low pressure") and problem remain accurate.

### Key results (estimates, to be checked at TRL 3)

| Quantity | Estimate | Requirement |
| --- | --- | --- |
| Electrolyzer input | About 70 W (4 cells, 9 A) | R3 met |
| Fill time, 70 to 300 kPa gauge | About 18 min, 20.8 Wh in | R2 met |
| Hydrogen made per fill | 0.189 mol, about 4.5 L, 15.0 Wh (HHV) | |
| Electrolyzer efficiency | About 72 % (HHV) | |
| Tank inventory at 300 kPa gauge | About 7.9 L, 0.66 g | R5 met |
| Fuel cell | About 10.8 W net for about 29 min, 5.2 Wh | R4 met |
| Round trip | About 25 % (HHV basis) | Near the low end of the published 27.4 to 48 % range |
| Full release into a 30 m³ room | About 0.03 % by volume, under 1 % of LFL | R7 met |
| Hydrogen measurement uncertainty | About ±2 % | R1 expected met |
| Bench size and mass | 900 x 450 mm, about 770 mm tall, about 18 kg | R13 met |
| Parts cost (H2Guard excluded) | About $845, or about $780 without the bench supply | **R14 not met** |

Requirements not met or at risk:

- **R14 (cost) not met:** about $845 against $450. The two stacks (about $400 together) and the pressure parts drive the cost.
- **R9 (hydrogen purity) not met on paper:** the 12 W class fuel cell specifies 99.995 % or better; the purity after the separator and drier is unknown.
- **R10 (electrolyzer back-pressure) at risk:** a generic stack must be confirmed to hold 300 kPa gauge on the hydrogen side.
- **R8 (interlock) at risk:** depends on H2Guard, whose README is still at scaffold stage; set point and response time are not yet defined.

### Proposed, awaiting Amish (items 1 to 9 now Decided by Amish, 2026-09-25: go with recommendation, HBN-DDR-002)

1. **Budget.** Options: (a) raise `budget_usd` to about $850 (bench power supply included, H2Guard costed separately); (b) keep $450 as the bench-only cost with both stacks bought separately by the school; (c) shrink to a 5 W class fuel cell and a smaller electrolyzer, which cuts cost but weakens the measurements. Recommendation: (a). `project.yaml` is unchanged.
2. **Storage method.** Rigid 2 L tank at up to 300 kPa gauge (recommended), metal hydride canister, or near-atmospheric gasbag with a pump.
3. **PEM rather than alkaline electrolysis** (recommended, for student safety).
4. **Fuel cell source.** Generic 12 W class stack with a published polarization curve (recommended) or a branded stack at $482 to $576.
5. **Hydrogen measured by pressure, volume and temperature** rather than a mass flow meter (recommended).
6. **Bench power supply as the source**, with solar as a later option (recommended).
7. **H2Guard included in the kit price or not.** Recommendation: cost it separately but ship the two together.
8. **Tank kept above about 20 kPa gauge between lessons** with pressure-cycle purging of a new tank (recommended).
9. **First users.** Vocational technician training or secondary school practical first; recommendation: vocational training, where supervision and safety culture are stronger.

### Safety concerns

- Flammable gas with a very low ignition energy and a nearly invisible flame; the bench must never run without H2Guard detection, extraction and interlock.
- Pressure: 2 L at up to 300 kPa gauge (about 0.65 kJ of stored pressure energy); only rated metal vessels and fittings, relief at 350 kPa gauge piped into the hood, tank guard.
- Air ingress into the tank could form a flammable mixture inside a closed vessel; purge and minimum-pressure rules are part of the design.
- Oxygen and hydrogen outlets must stay separate.
- Mains-powered supply near water: earthed supply, RCD or GFCI, reservoir kept away from the supply.
- Warm stacks (about 20 W of heat each, fuel cell stack limit about 55 °C).
- Users include students under 18: supervised use only, with teacher-held control of the power supply.

### Problems and notes

- H2Guard's README is still the scaffold. H2Bench cites only what it states (sensor, relay controller, spark-free fan, normally closed supply valve, alarm). The interlock outputs H2Bench needs (power supply cut, tank solenoid close) should be confirmed with that project.
- Generic stack prices (items 6 and 12) could not be verified from a primary source; they are marked as estimates in the BOM.
- The Namibia figure in the README comes from a news report of a university statement, not a published study.
- The legacy `cad/src/model.py` placeholder is untouched (TRL 3 work).

### Recommended next step

Review this note and the media, then decide items 1, 2 and 9. If approved, run `/advance-trl3` to check the energy, gas law, uncertainty and release calculations in a CAL note, confirm electrolyzer back-pressure and purity with suppliers, and produce the parametric model and drawing sheet. The design is not ready for TRL 4.

## Session 2026-09-26: sources strengthened

README sources were checked with WebFetch and the weaker ones replaced or supported by primary or reputable secondary sources.

- **What sparked the idea:** Encyclopedia.com entry on William Robert Grove replaced by Grove's own paper, "On a gaseous voltaic battery," *Philosophical Magazine* 21 (1842): 417 to 420 (doi:10.1080/14786444208621600, metadata confirmed through Crossref). The 1843 *Philosophical Transactions* link now uses its DOI resolver, and the abstract confirms hydrogen and oxygen on platinum in dilute sulfuric acid and the decomposition of water. The unsupported phrase "the forerunner of the hydrogen fuel cell" was removed.
- **Namibia row:** Ecofin Agency replaced by *The Namibian* (21 September 2025), which quotes the NUST vice-chancellor on a talent gap of up to 130,000 workers by 2040, plus NUST's own IGNITE GH2 project page (685 graduates and 40 instructors).
- **Concept rationale:** the $482 to $576 fuel cell price now carries links to the two listings (Fuel Cell Shop, Fuel Cell Store).
- **Burning platform:** IEA skills survey wording corrected to match the source (some 700 energy companies, unions and training institutions).
- All other links (IEA Global Hydrogen Review 2025, European Commission, US DOE, PIB, IEA policy database for Chile, Oxford Institute for Energy Studies) were re-fetched and support their claims. No budget change.

## Session 2026-09-26: budget approved

Amish wrote, in chat on 2026-09-26: "i approve all the budget items." The open cost item (O5) is decided: budget set to $885 to cover the priced BOM (HBN-DDR-002 v0.2).

- `project.yaml` `budget_usd` $850 to $885; README budget and cost lines updated.
- R14 target $850 to $885; status **not met to met**, with no margin ($885 with the bench supply; $820 without; H2Guard excluded).
- Requirement counts (HBN-CAL-001 v0.3): 11 met, 1 not met (R9), 3 at risk, 1 not verifiable.
- Documents: HBN-PRB-001 v0.5, HBN-PRC-001 v0.5, HBN-REQ-001 v0.5, HBN-CAL-001 v0.3 (`sizing.py` budget constant 850 to 885, script re-run, `results.csv` regenerated), HBN-DDR-002 v0.2; `bom/bom-notes.md`; PDFs rebuilt. No media shows the budget, so none was regenerated.

## Session 2026-09-26: product appearance model and photoreal renders

Amish chose this repo for the first batch of product renders on 2026-09-26. This session adds an appearance model for photoreal renders; the render images (`media/render-hero.png`, `media/render-exploded.png`) are produced separately by the portfolio render step.

### What was done

- `cad/src/product_model.py` (new): `product_parts()` returns 116 parts (106 shell, 9 internal and the lab table top as context), each with a colour, a render material, its BOM line, a group and an explode offset. It imports `PARAMS`, `levels()`, `tank_cyl_len()` and `build_parts()` from `model.py`; the gas lines and the electrolyzer membranes are taken from `build_parts()` unchanged. It also defines `TITLE` and `RENDER_VIEWS` (hero, exploded and a front view from the student side).
- What it adds over the massing model:
  - Bench: T-slot grooves on the aluminium frame, a filleted HDPE deck and drip lip with a teal edge stripe and a name plate.
  - Back panel: the printed energy-chain schematic (flow line, seven stage nodes, captions), panel screws, a hydrogen warning sign, and lit logger and stage meter displays.
  - Canopy: clear polycarbonate sheet, aluminium edge trim, slotted posts with foot brackets, and a duct collar with a flange.
  - Components: a lab supply with lit voltage and current readouts, knurled knobs, binding posts and a switch; a HDPE reservoir with a teal cap and graduations; a clear deionizer cartridge showing its resin; teal electrolyzer end plates with tie rods, nuts, grooved titanium cells and brass ports; a clear separator (water visible) and a clear drier (indicator gel visible); a stainless arrestor with hex ends and a flow arrow; a brushed tank with a red shoulder band in a powder-coated guard; a brass manifold with a dial gauge, a red-capped relief valve, a needle valve knob and the cut switch; the regulator knob and solenoid coil; fuel cell plates, end plates, fan and finger guard; the electronic load with a lit readout and a lit lamp; and the H2Guard head, fan housing and controller with lit status lights.
  - Safety labels: a flammable gas label (GHS diamond) on the tank (wrapped), electrolyzer, regulator, fuel cell and H2Guard fan housing, and a warning sign on the back panel (BOM line 17).
- `README.md`: hero image now points to `media/render-hero.png`; an "Exploded render" link was added to the links line.
- Self-check previews (not in the repo) were made with the kit renderer; clear parts are left out of those previews.

### Differences from model.py (Proposed, awaiting Amish)

All main dimensions, positions and interfaces are as `model.py`. The differences are appearance only:

1. **Canopy edge trim and post foot brackets.** A 10 mm aluminium trim wraps the canopy edges and 6 mm foot brackets sit under the two posts; neither is in `model.py` or BOM line 3. Post extrusions stop 5 mm short of the canopy underside, inside the trim. Recommendation: adopt both in `model.py` at the next CAD update, since a bare 3 mm polycarbonate edge is hard to handle safely.
2. **Red shoulder band on the tank.** It follows the common convention of a red shoulder for flammable gas cylinders. Recommendation: keep it for the render, and confirm the marking against the vessel supplier's standard before any purchase.
3. **Product colours in place of the concept palette.** Examples: the load and lamp are light grey rather than yellow; the meters are black with lit readouts rather than green; H2Guard is white with a teal band rather than red. Recommendation: use the product palette for renders and keep the concept palette for the numbered concept media, where the colours help the callouts.
4. **Detail inside the model.py envelopes.** The lamp is a holder with a bulb, and the fuel cell has end plates and a fan guard. The separator and drier are clear bodies with end caps, and the tank has a 5 mm base fillet sitting in a shallow recess in the cradle. All stay inside the model.py envelopes. Recommendation: accept as appearance only.
5. **Context.** A 1,020 x 560 mm section of the existing lab table top replaces the full table and the 1.75 m person, so the bench fills the frame. Recommendation: accept for product renders; the concept hero keeps the person for scale.

### Status

This is an appearance model only: no tolerances, no fabrication detail, no PCB layouts. `trl` stays 3 and TRL 4 remains on hold. No change to `model.py`, the BOM, the budget or any requirement status. Safety concerns are unchanged from the sessions above.

## Session 2026-09-27: kit 1.5.0 and image quality

- Kit 1.5.0 synced: STANDARDS v1.5 (sections 12 to 15: product renders, storefront images and image quality, public release, authorship and signing), `.kit/cards.py`, `.kit/image_qc.py`, `.kit/release_gate.py`, issue templates, and the `/render-product` and `/release` commands. `CLAUDE.md` now matches `.kit/CLAUDE.md`.
- Every `media/render-*.png` recaptioned from its original render with the new layout: the title, concept label and repository sit in a band above the render and the view note in a band below it, each line wrapped to the image width, so no text overlaps other text or the render or runs off the image. `media/card.png` and `media/social-preview.png` regenerated with the same rules.
- `python .kit/image_qc.py` and `python .kit/release_gate.py` pass. trl stays 3.