Smartfan · Prototype V2
Drag to orbit, scroll to zoom, right-drag to pan. Real sizes in millimetres: a 16 × 16 in opening, 6 in deep, a 120 mm fan, a 4 in duct with 2 in of spigot proud of the box. Pull Service from below and the grille, the filter and the fan cartridge come out into the room — the box stays in the ceiling.
The filter is what makes this tight. It sits on the room side of the ceiling line because in the void the stack reaches 150 of 152 mm. It also costs airflow this fan does not have — see the note below the model.
The turn, and the shroud that would soften it. The fan blows +Y and the duct leaves +Z: every cubic foot turns ninety degrees inside 107 mm of headroom, and in the bare box it does that by hitting the lid and finding the hole sideways. Switch Airflow on and the corners show it — two slow eddies going nowhere that the fan is still paying for. Turning shroud draws the fix: a printed elbow standing on the cartridge top. Its size is decided rather than chosen. The bend's centre of curvature is fixed by the two axes it joins, so the outer wall tops out at y = 147 and the inner wall at 47 for any radius — the height always works — but the inlet sits at 97 − R and cannot drop below the cartridge at 44. That pins R ≈ 53 mm, R/D 0.53: a tight elbow, because a good one (R/D ≥ 1) would have to start inside the fan. A plenum turn costs roughly 1.2–1.5 velocity heads and an R/D 0.5 elbow about 0.7, so the part buys back over half of one: 0.013 in w.g. at 50 CFM, 0.05 at the iPPC-3000's 100 CFM — 8 to 17 per cent of that fan's whole budget, for one printed part. Printed, not steel, for the reason the collar and cartridge are: it is not the fire barrier, the box is. All of it sits above the cartridge, so service from below is unchanged. These streamlines are sketched, not solved — no CFD was run.
The filter. The plan said no lint filter until this model asked for one, and the reason it said so is arithmetic: the iPPC-2000 has 0.155 in w.g. of static pressure in total, and the easy-duct case already spends 0.10 of it for about 20 CFM. A clean filter costs roughly 0.02–0.05 in, and a loaded one costs more — which on this fan is most or all of the remaining margin. If the filter stays, the fan should become the iPPC-3000: same 120 × 25 pocket, same 105 mm screws, same 12 V, 0.30 in w.g. and 110 CFM free air. It is louder (43.5 dBA against 29.7) and that is the trade. The three notes have been updated to match this model rather than the other way round.
Sensors in the grille. This one the plan was already asking for — room air, room light, out of the extract path. The button and both LEDs are on that same face. The button cable leaves the grille through a cable fitting, then enters the 1590BB through a second one. The sensor cable has its own fitting on the box. The grille comes off, so each cable ends in a plug.
The duct runs down the bay, and that is forced. A 100 mm duct cannot cross a joist. On a 2×6 — 140 mm deep — the largest hole allowed is about a third of the depth, roughly 46 mm, and it has to stay clear of the edges; a 100 mm hole would take the joist with it. Nor can the duct pass over the joists, because they are 140 mm of a 152 mm void. So the spigot leaves along the bay, parallel to the joists, and the fan is oriented to suit. The bonus is that it satisfies the straight-run rule for free: the bay gives far more than the 100–200 mm of settling length a bend wants.
Steel, and what “UL listed” can and cannot mean here. The air box is drawn as folded galvanised sheet because a ceiling is a fire barrier and a printed one is fuel. The collar, grille and cartridge stay printed — they are aerodynamic and serviceable parts, not the barrier. But a listing covers a finished assembly tested as a whole: a self-built unit is not listed, and modifying a listed housing voids its listing. What is achievable is listed components — the IRM-10-12 already carries UL 62368-1, the fan frame is UL 94 V-0 — inside a metal enclosure, which is the honest claim. A listed product is a separate project through an NRTL, and ETL is usually the cheaper door than UL.