Tesseract · CoScad

Tesseract with captured joints

The current tesseract replaces the 19 mm bolted corner blocks with 12 mm split hubs and chamfered 7 mm bars. The outer envelope remains 100 × 100 × 100 mm; the centre remains a 30 mm cube. The new hubs are 37% smaller across and have about one quarter of the old block's bounding volume. The original 24-screw version is retained in examples/assemble/tesseract/bolted/ for comparison and migration tests.

How the connection works

Each bar has an integral 4.2 mm neck and 5.4 mm collar. The corner base has open channels so the bar can enter sideways, perpendicular to its length. An outer cap closes those channels. Once both halves are together, the collar cannot pass through the narrower neck opening: the frame's retention comes from the captured shape rather than friction on the bar alone.

The cap has an integral split pin, initially 3.36 mm diameter, in a 3.30 mm hole. Its slight interference holds the cap shut. The nominal 0.03 mm radial interference is a starting point for a printed fit test, not a printer-independent guarantee. There are no screws, nuts, loose dowels or adhesives. The .assemble keyword fastener peg records an integral joint for the instruction planner; it does not add a separate manufactured part.

Files and printing

The support is 35 mm tall, with four posts under the inner cube. The corner halves print with their outer diagonal face down; the cap's outer tip is trimmed at x+y+z=9 to give it about 69.7 mm² of flat contact (the original x+y+z=3 trim had only 7.4 mm²). The pin still prints vertically. next uses real mesh bounds when putting these diagonal faces on the bed. Full-length bars print lying down, and the diagonal struts retain a flat underside. The short fit-test bar stands on its broad end so its integral collar prints upward without an isolated first layer. The full-length bars require local slicer supports beneath their end collars: the collars start 0.8 mm above the bed and their necks start at 1.4 mm. The upright short coupon does not test this. The retry explicitly enables snug supports and an 8 mm brim, and verifies the resulting toolpaths.

The supported build contains eight bases, eight caps, twelve bars, eight struts, one inner cube and one removable support. Use the same filament and slicing settings for the fit test and the full build. If the cap is too tight, reduce fit in cap.coscad and reprint the fit test. For example, fit = 0 gives equal nominal diameters, and fit = -0.05 gives 0.10 mm diametral clearance. These settings may be loose; the fit test must establish retention.

Assembly sequence

  1. Place the inner cube on the temporary support.
  2. Insert the eight diagonal struts into the cube, then slide each corner base onto its exposed strut end from outside the cube. The shallow socket seats the base; it is not yet a locked frame.
  3. Seat the twelve bars in the open corner bases. Each bar enters from outside its cube edge, perpendicular to its own axis, into both bases together. There is no final bar that must be pushed axially between two closed sockets.
  4. Check all collars are seated. Press the eight caps diagonally inward, after every bar is installed. Their split pins retain the caps, and the closed hubs capture the bar collars.
  5. Lift the frame and withdraw the temporary support straight down through the bottom opening. The support is not part of the finished object.

The generator enforces the support/centre/strut/base/bar/cap prerequisites using the existing generic after= hints. Its generated plan has 36 steps and no flips. Support removal is the final manual operation above: the current planner does not generate removal operations. Its legacy face labels on peg joints are locations, not the diagonal insertion directions described here. Insertion motion is checked separately by the model regression.

Verification and limits

Run the repeatable geometry regression with Python 3.12 or newer:

python3 -m venv /tmp/tesseract-check
/tmp/tesseract-check/bin/pip install -r examples/assemble/tesseract/requirements.txt
/tmp/tesseract-check/bin/python examples/assemble/tesseract/verify.py
coscad next examples/assemble/tesseract/fit-test.assemble
coscad next examples/assemble/tesseract/tesseract.assemble
coscad plan --png examples/assemble/tesseract/tesseract.assemble

The regression separates the deliberate pin press fit from unwanted interference: it uses a clearance pin for all-pair assembly checks and sampled entry paths, then checks that the nominal pin alone has a small positive interference. It checks connected watertight parts, bar capture, cap/base/bar approach paths and downward support removal. Path positions are sampled every 0.1 mm; this is not a continuous swept-volume proof or a flexible-material simulation.

The verification report records 748 compiler checks, 4,629 sampled motion positions and the one-bed meshes.

The original A1 coupon lost bed adhesion on subsequent layers; the center base initially adhered, while the side cap and short bar failed. The revised four-piece retry is prepared offline pending restoration of the home network. No successful revised print, cap pull-off test or long-term creep test has been run. The split pin and its fit are the remaining practical uncertainty. This is a display model; its captured bars should not be treated as a rated structural connection. Avoid repeated forced disassembly until the fit test establishes how the selected filament behaves.