Feedstock Hopper: Interfaces
The hopper is a set of modules joined at three defined interfaces, not a set of parts. Anything that satisfies an interface can be swapped in, which is what lets one hopper serve a bare MK3S, an enclosed one, and machines we have never seen – someone with a different printer makes one small adapter, not a new hopper.
| Interface | Between | Varies with |
|---|---|---|
| A – hopper | body <-> hub | never |
| B – outlet | hub <-> outlet module | the hose |
| C – mount | hub <-> machine plate | the machine |
Why nothing connects to the plate
Section titled “Why nothing connects to the plate”The obvious layout – a plate carrying the hopper socket above and the outlet below – cannot be printed. A bayonet socket only accepts from one end, so a plate with a connection on both faces always has one facing the bed: a recess facing down is unprintable, and flipping it puts the other connection into the bed instead. You can have quick-release on one face of a plate, not both. The imported mount had exactly this problem, which is why it printed badly in every orientation.
So the connections move off the plate and onto a hub: a short tube with a bayonet socket at each end, bolted down to the plate from underneath. The body twists into the top, the outlet twists into the bottom, and the plate stays a flat piece of material with holes in it.
| Part | Prints |
|---|---|
| hub | standing on its bottom face. A skirt flares out to 95 mm at the bed and tapers in to the 72 mm tube, so there is no overhang anywhere – both sockets are internal bores. |
| plate | flat. A clearance hole, four holes for the hub, and the machine’s own pattern. No features on either face, so it can equally be laser-cut, acrylic or plywood. |
| outlet | neck upward. Converges to the hose socket, which sits on the bed. |
Two things fall out of bolting the hub down rather than clamping it:
- The hopper stays retained with the outlet removed. If the outlet were what held the stack together, twisting it off to clear a jam would leave 4.24 kg of full hopper loose on the plate.
- The hub cannot rotate when the body is twisted into it, so no separate anti-rotation feature is needed.
The plate – the only machine-specific part – is then also the simplest and cheapest thing in the stack to get wrong and reprint, which is the right place for the uncertainty to sit.
A – hopper
Section titled “A – hopper”The existing coupling, unchanged: 58 mm neck in a 72 mm socket, 18 mm deep,
25 degree turn, keyed. 50 mm pellet bore. Sized by the converging-outlet flow
rule – see design-notes.md.
B – outlet
Section titled “B – outlet”The same coupling as A, not a smaller one. That looks wasteful until you follow the bore: the hose is 20 mm, so the path has to converge from 50 mm to 20 mm somewhere, and a converging section is exactly where material arches. Putting a second, smaller coupling in the hub would move that convergence into the part that is bolted down. Keeping the coupling identical lets the hub run a straight 50 mm bore and puts the whole convergence inside the outlet module – which is the part you can twist off in seconds.
The jam-prone geometry belongs in the quick-release part. That is the reason for the choice, and reusing the coupling also means one interface to verify instead of two.
Quick-release here is the point: clearing a jam by twisting the outlet off beats undoing fasteners, and swapping hose types becomes a small reprint.
As built
Section titled “As built”| Part | Size | Prints |
|---|---|---|
| hub | 95 skirt, 72 tube, 36 tall, straight 50 bore | standing on its bottom face |
| plate | derived (134.6 x 134.6 for the MK3S), dia 65.0 hole, 4 x M4 at dia 79 | flat |
| outlet | dia 64 over the pins, 83 tall | standing on the hose socket |
Verified by intersection rather than by eye: every coupling in the stack seats free and captures at an identical 115.31 mm3 when displaced 3 mm, a bolt passes cleanly from under the plate through the hub’s fixings. The outlet’s bore widens monotonically from the hose end up to the coupling, so in the flow direction it converges with no ledge.
The hose-into-socket fit was settled on a printed coupon: a 21.5 mm tube with a 3.5 mm rib at 8.5 mm pitch, right-handed, with 0.2 mm of clearance on both the bore and the groove. The groove is a swept polyhedron that ends on flat caps, so the sweep runs a full lead past the socket at both ends; a cap inside the part would leave solid material where the hose has to start.
C – mount
Section titled “C – mount”Ours to define, so it is a documented number rather than an inherited one. A plate satisfies it with three things:
| Feature | Size | Why |
|---|---|---|
| central clearance hole | dia 65.0 | must clear the outlet neck’s pins at dia 64.0, not the dia 58 neck |
| hub fixings | 4 x M4 on a dia 79 circle | outboard of the hub’s 72 mm tube, inboard of its 95 mm skirt; through-bolted, nut in a dia 8 counterbore |
| machine pattern | machine-specific | the only part that touches reality |
Bearing is a non-issue – the hub seats on a 722 mm2 annulus, which is 0.05 MPa for 4 kg (see loads) against roughly 50 MPa for PETG. Plate bending around a 65 mm hole is the thing to size for, and plate thickness is a parameter for that reason.
Plates:
- MK3S frame (
hopper_plate.scad,plate_variant = "mk3s") – clamps the printer’s own 370 x 370 mm frame – measured as 6.3 mm steel, 40.5 mm across the top bar. Built; see below. - Panel (
hopper_plate.scad,plate_variant = "panel") – bolts through a drilled flat sheet, for the Original Prusa Enclosure or anything similar. Built. The pattern is parametric because the panel is drilled to suit – there is no fixed pattern to match. Drill the panel dia 65 for the outlet neck and use washers or a backing plate: four M4 heads on thin sheet under a sustained few kilograms will dimple it long before anything breaks. - Stand – not built; there is no
standplate variant. It would decouple the load from the printer entirely. Worth having, since 410 mm of body on a 370 mm frame puts 4.24 kg at around 800 mm on a machine that slings its bed in Y.
The MK3S plate clamps rather than hooks
Section titled “The MK3S plate clamps rather than hooks”A cantilever hook stays seated only while its load holds it down, which suits a spool of filament but not a full hopper. This plate straddles the frame and pinches it.
The hopper axis is offset 47 mm from the frame plane, not over it. The outlet hangs below the plate and has to pass beside the frame rather than through it. There are two minimums and the second is the one that bites: the outlet clearing the frame needs 38.6 mm, but the saddle – jaws and fillet included – is wider than the bar it grips and must also sit clear of the plate’s own clearance hole, which needs 46.8. Both are asserted, and the error says which one it was.
The saddle’s roof carries the vertical load, sitting on the frame’s top edge, so the clamp screws only resist sliding and tipping rather than holding the weight in friction.
That roof is also a riser, and its height is solved rather than chosen. The outlet hangs below the plate and beside the frame – in the column the gantry rises through – so at full Z the toolhead reached it. The mount therefore stands off the frame by whatever puts the outlet’s mouth clear of the top bar, which needs nothing measured on the machine: nothing that travels can go above that bar. At the shipping outlet that solves to a 62.2 mm roof and a 98 mm tall plate, leaving the saddle’s jaws the only thing reaching down the bar, which they must do anyway to grip it. The clearance above the bar is the only handle; the lift follows from the outlet’s height, so a different hose re-solves it instead of leaving a hardcoded riser quietly wrong.
The braces follow the riser rather than the saddle. They stop at the frame’s top edge, so the saddle’s jaws stay the only part reaching below the bar, and they run inboard far enough to sit under the whole hub footprint rather than stopping at its axis – the bolts furthest from the saddle are on the longest lever, so ending at the axis braced the half needing it least. Tying their height to the saddle instead was fine while the saddle was short; once its roof became a riser it made each brace a near-vertical fin hanging 30 mm below the bar, which is the collision the riser exists to avoid.
Stiffness sizes this part, not strength. A bare flat plate is well inside yield but deflects half a millimetre at the hub, and with 450 mm of hopper above that, half a millimetre becomes nearly five at the top – on a machine that slings its bed in Y. So the saddle spans the full plate width rather than a stub in the middle, and braces turn the flat cantilever into a T-section. Worth roughly fifteen to twenty times the stiffness for a few grams.
The plate sizes itself. Its width follows from the offset, the saddle’s own width and the hub skirt; the saddle follows the plate; the braces follow the clearance hole and the plate edge; the clamp screws follow the saddle. A margin is the only handle. That is not tidiness – a plate width and a saddle length set independently is precisely how the saddle ended up shorter than the braces meant to sit on it. At the loaded 4.24 kg on a 47 mm offset, the couple across a 30 mm grip is about 65 N.
Verified: the frame clears the plate, hub and outlet; the slot accepts 6.2 and 6.5 mm and rejects 6.8. It prints upside down – plate face on the bed, saddle legs up, slot opening upward – with nothing overhanging.
