cycloidgen v7.10.0

Numbers — they move, on every design, and here is by how much. Nine designs

• the six presets, rollers, a ground steel 21:1 and a ring-output drive - were
  analysed by 7.9.0 and by 7.10.0 and the whole of report.json compared.

• Mass, everywhere. Each bearing seat is now bored to its bearing and has a
  lip, the shaft is located by two sleeves, and the discs are held apart by
  washers, and all of that is metal or plastic the mass model weighs. Assembled
  mass is up 0.9% to 1.5% on the presets (781.1 to 791.8 g on the 21:1), 1.0% on
  ground steel and 0.6% on the ring-output drive. mass.spacers_mass_g is new:
  3.7 to 5.5 g across the presets, 9.5 g in steel. Inertia, the unbalance couple
  and power density follow.

• The contact, on a printed disc. An FDM disc in a printed material now has
  its first-layer edge broken back 0.3 mm, so the analysis counts that much less
  of every line contact on it. On every preset: contact pressure +1.9% and the
  safety factors down by it, torque capacity -3.75% (3.51 to 3.38 Nm on the
  21:1), -2.5% to -3.2% with clearance (0.898 to 0.875 Nm), torsional stiffness
  -1.2% to -2.1% with its contact share -2.4% to -3.3%, wind-up +1.0% to +1.9%,
  PV at the ring and output pins +3.9%. Transmission error moves 1.4% on the
  10:1 and the 15:1 and under 0.6% elsewhere. Lost motion, efficiency and
  temperature do not move.

• A machined drive moves only by its mass. The ground steel 21:1 is
  identical in every other value.

No check passes or fails a design it did not before. The ring-output drive gains
the new TIE_NUTS_ON_OUTPUT_FACE warning: its tie bolts have nuts now, and on
that drive they stand on the face that turns.

Added

• Parts you can build. The model had every part's shape and none of what
  holds a gearbox together, and now has it:
  • Seats bored to the bearing in them, not to the room it left. The 6800
    shaft supports, 19 mm outside, sat in 22 mm bores on every preset, and a
    laser-cut 21:1 disc was one its cam bearing fell out of. The disc, both end
    plates, the carrier's boss and the end cap are bored to the bearing's own
    outside in the STEP, the meshes, the 3D view, the drawings and the cutting
    DXFs; a seat with no bearing keeps its running clearance.
  • A lip behind every seat, its hole at the bearing's pitch circle, so a
    bearing is held in by something other than its press fit. The end-plate
    drawings say which face each seat is bored from and how deep.
  • Two sleeves and the disc washers, locating the shaft from cam to shoulder
    and holding the discs at their designed gap. Made parts with their own
    files and lines on the bill of materials.
  • Tie bolts as screws with a size, a length, a head and a nut - ISO 4762
    cap screws named off the clearance hole (4.5 mm is an M4) at the next
    preferred length past the stack, ISO 4032 nuts, from a new
    core/fasteners.py table for M2.5 to M16.
  • A flat on the input shaft for a coupling's set screw, on a carrier drive.
  • Edge breaks and lead-ins, edge_chamfer, 0.5 mm by default: outside
    edges and every bearing-seat mouth on the solids, never on a running
    surface. Outside corners only, so it never adds a fillet into a corner a
    bearing seats against.
• A printed part comes out ready to print. Each part in stl/ stands on the
  face that needs least support, at the origin, and on FDM the bed face's edges
  are broken back 0.3 mm for the first layer's squash. The 3MF carries the same
  edge break on the same face.
• The 3D view says what you click. A picked part is tinted and its line on
  the bill of materials is read out under the viewport. X-ray sees through
  the housing and its end plates, and clicks go through them. Measure gives
  the distance between two clicked points, in the window's unit. Clashes
  tints every part a clash check is about, and selecting any check in the list
  tints that check's parts.
• Two checks. SEAT_HAS_NO_SHOULDER warns of a seat with less than a
  millimetre past its bearing - including a motor register cut into the same
  face as a shaft support, which the model has had since registers existed.
  TIE_NUTS_ON_OUTPUT_FACE warns that on a ring-output drive the nuts stand
  proud of the face the driven machine bolts to.
• An Advanced switch on the parameter panel. It opens on the seventeen
  fields a design is made with; the other forty-four are behind the switch
  (also View ▸ Advanced parameters, Ctrl+Shift+A). Hidden fields keep their
  values and are still analysed and checked, a search finds them with the
  switch off, and a selected check shows the ones it points at.

Changed

• The STEP disc is the sketch's spline, not a polygon of several hundred
  flat faces up to 55 µm off the profile: one face per flank, a vertex at every
  root and crest. A CAD package can offset and dimension it, and the file is
  about a seventh of the size. "DXF tolerance" is now "Profile tolerance", because it
  decides the STEP disc as well.
• The window is quieter. One primary action, Export ▾, in the header
  beside the readings, with a red Export blocked pill naming the checks when
  it is blocked; the row of export buttons under every tab is gone, Pin as
  reference is on the Compare tab, and the not-verified notice runs along the
  status bar. Sentence case instead of capitals throughout, toolbar toggles as
  chips, the camera views as one joined control, parameter groups that fold
  behind their headings and are remembered. A session left on a tab under its
  old capitalised name still opens on it.
• A short checks list is one line per finding, so a 1366x768 laptop shows
  four of them rather than two wrapped ones; the full sentence is the tooltip
  and the explanation panel.
• report.json gains mass.spacers_mass_g and spec.edge_chamfer, and a
  saved design gains edge_chamfer. Older design files load, with the 0.5 mm
  default.

Fixed

• The cutting DXF of the input end plate on a ring-output drive was drawn
  to the shaft support's seat where the solid has the second main bearing's,
  and the carrier's drilling template had an 11 mm hole where the carrier is
  bored to its support. The mass model had weighed the first of those wrong
  too.

---

Windows - cycloidgen_..._Setup.exe. It is unsigned, so
SmartScreen will warn on first run: More info -> Run anyway. A
signed build is planned.

Linux - cycloidgen-...-x86_64.AppImage. chmod +x and run it;
it needs glibc 2.35 or newer. If it will not start, the machine has no
FUSE 2: install libfuse2, or run the file with
--appimage-extract-and-run, which unpacks it instead of mounting it.

macOS - cycloidgen-...-arm64.dmg, Apple silicon. Drag it to
Applications, then run this before opening it:

xattr -dr com.apple.quarantine /Applications/cycloidgen.app

It is signed ad-hoc but not notarized, so Gatekeeper blocks the first
launch. On macOS 15 and later the dialog it raises defaults to
Move to Trash, which deletes the install - there is no Open
Anyway in that dialog, it appears in System Settings only after a
launch has been blocked, and right-click -> Open no longer bypasses
it. Clearing the flag first avoids the dialog entirely. On an Intel
Mac, use pip.

Anywhere with Python 3.10-3.14: pip install cycloidgen, then
cycloidgen.

The numbers this produces are preliminary sizing estimates, not a
certification. Validate against a physical prototype before anything
load-bearing depends on them.