Seaport, printing several parts in a row by shoving each one off the plate
Arachnid's Seaport mode cools the bed, pushes a finished part off with the body of the toolhead and starts the next on fresh sheet. The geometry that makes it work, and the parts it refuses.
- Published
- 3 min
- To read
- Arachnid
- Project
A printer without a belt or a robot arm needs a person to take each part off the plate before the next one can start. Seaport is Arachnid's way of removing that person for small parts. It prints one part, cools the bed until the plastic lets go, shoves the finished part off the plate with the body of the toolhead, and starts the next part on a different patch of sheet. If a layer of the first part has welded itself to the plate, the second part is not printed on top of it.
arachnid a.stl b.stl c.stl --set print.seaport=true
The geometry is the feature
The only tool for pushing is the toolhead, so everything depends on what is at each height under the nozzle. Measured up from the nozzle tip on a Bambu A1, the first 2 mm is empty space, from 2 mm to 5.5 mm is the sock, and the head body starts at 5.5 mm.
For the body to reach a part of height H, the nozzle has to sit at H minus 5.5 mm or lower. That puts the nozzle tip inside the height range of the part for the whole push. There is no Z at which the body touches the part and the tip is clear of it.
The only thing keeping the nozzle off the plastic is sideways lead. The leading face of the head runs 25 mm ahead of the tip, so the body meets the part 25 mm before the tip does. That gives two situations Arachnid cannot make safe, and in both it refuses the job and says why:
- A part shorter than 7 mm
- Only the sock and the nozzle are low enough to touch it, and the body cannot reach it at all.
- A part further from the ejection edge than the lead
- The part would still be on the plate when the nozzle arrives. The refusal includes the distance it would need.
What the push looks like
The push runs at 8 mm/s with 300 mm/s² acceleration and the nozzle switched off. Before it, the printer waits for the plate to reach temperature on the way down, with M190 R, and then dwells for five minutes. The dwell is there because PLA keeps contracting after the plate reads its target.
Where the nozzle rides matters as much as how fast it moves. The obvious implementation sits the nozzle just under the top of the part so the body is certain to make contact. On a Benchy that puts the contact point 97 percent of the way up the hull, which uses the whole height of the part as a lever against its own base. Arachnid pushes low instead. With the nozzle riding 0.6 mm above the sheet, the underside of the head body sits at about 6 mm, so the same shove acts through a much shorter lever.
Checking whether the part survives being dropped
Ejection ends with the part falling, so before it prints anything Arachnid looks at the part it actually sliced and not at its bounding box, which cannot see a narrow neck. It flags four things:
- Slenderness
- Height against the square root of the footprint. A tall, narrow part tips over instead of sliding.
- A waist below the push point
- The narrowest cross section anywhere, and whether the push lands above it. That section carries the whole moment.
- A thin shell
- One wall, or infill below 10 percent, is a skin around air.
- Mass
- Under 3 g the risk is the landing and not the push.
A draft Benchy with one wall and 8 percent infill is flagged on both shell counts. The same model at the quality setting, with three walls and 15 percent infill, is not flagged at all.
Where it fits
Seaport only runs when the geometry allows it, and it says so when it does not. Parts it cannot push safely are refused with the reason, and a fragility warning appears before anything prints. The Arachnid project page lists the rest of the engine, including strength mode, multicolour flush volumes and sending jobs over the local network.
Written by Caisson AI. Questions or corrections go to hello@caissonai.co.