Two Axes

A dispatch from the bench, late August 2026. On August 27 I ran the current STL set through a mesh audit. This is what the geometry says the drivetrain is, with the numbers.

---

The power path

Two axes. The first is the motor shaft: the friction clutch clamps onto its M4 thread, and the pinion is the part the clutch drives. The second is a fixed printed axle that carries a 28-tooth spur at one end, a 2-start worm in the middle, and the manual stub at the other. The spool for cord blinds slides onto that stub. The worm drives a 20-tooth worm wheel on the blind's own tilt rod.

Motor → clutch → pinion → 28T spur → worm → worm wheel → rod. That is the whole train. Two printed gears, one worm, one wheel.

The spool lives on the same axle as the worm, which is the point. A cord pull turns the axle the motor turns, through the same worm, so manual control never has to fight the motor or the gearing. Everything I tried that sat downstream of the worm killed the cord. This does not.

Axle, left to right: square stub with the slot, journal, worm between its two cones, journal, 28T spur. One print, no supports.

The worm, measured

• 2 starts, module 0.8, lead about 5.1 mm per revolution.
• Tip diameter 30.58 mm, thread depth 1.61 mm.
• Lead angle 3.21°.
• Threaded band 4.06 mm. The rest of the envelope is the two cones.

Two starts usually cost you self-locking. Here they do not, because the worm is large: Ø29 pitch diameter against a 5.1 mm lead gives that 3.21° lead angle, and for PETG on PETG that self-locks across any plausible friction coefficient with 3.5× to 6.8× margin. The blind holds position with the power off. That was the founding requirement, and it is not marginal.

The cost is efficiency, 12 to 22%. A 10:1 worm stage at 15% efficiency multiplies torque by about 1.4×, not 10×. That is the single most misleading number in any worm drivetrain, so it is worth saying plainly.

The 4.06 mm threaded band is short on purpose. Longer thread means taller cones, and taller cones need supports, and supports on a worm come off the thread flanks, which are the working surfaces. The band stays.

The ratio

Pinion to axle spur is roughly 1:1, sized by where the motor sits relative to the axle. Worm to wheel is 2 starts into 20 teeth, 10:1. Total 10:1.

At the Hotec motor's 60 RPM that is 6 RPM at the rod and 5.0 seconds for a full 180° tilt. The target was 4.6 s, so 20 teeth is the right count; 26 would have been 6.5 s. Centre distance from worm to rod: 22.48 mm, which is the fastest way to confirm the tooth count without counting.

Rod torque at motor stall works out to 0.47 to 0.86 N·m. A slat stack needs maybe 0.15 to 0.35, and less than that, because the TiltRod drives the input of the headrail's own worm rather than the slats. Torque is not a constraint anywhere in this train. The consequence is that the clutch is the only thing protecting the printed wheel from the motor, which is why the clutch gets its own two posts.

Worm on the axle, wheel on a rod stub, meshed at 22.48 mm.

What the audit corrected

I described the worm as "8 mm wide between two cones." The band that is actually cut with thread is 4.06 mm. I quoted the tip at Ø30.99; the mesh measures 30.58. Neither changes anything, but the record should say what the part is, not what I remembered.

The TiltRod, by TiltForge, is a retrofit smart blind motor — repairable, open, Matter-native, and the cord always works. Founders pricing on the first 100 units; first units ship November 2026. Check your headrail at tiltforge.com/fit or follow the build at tiltforge.com/#waitlist.