Drivetrain
How the encoder knows where your slats are.
An absolute encoder needs a shaft end, and a blind's tilt rod does not give you one — it runs straight through the housing. A 12-tooth spur fused beside the worm wheel drives a 16-tooth crown gear onto a vertical axis, which creates that shaft end. The crown sits under the rod, beside the worm, teeth up at the spur — and one diametric magnet sits in a pocket down its middle, just above the AS5600 on the board below. Drag the rod: every rotating part carries the same reference mark, so the ratios are something you watch rather than something you take on trust.
Tilt rod
20T wheel + 12T spur · the bar is the slat plane
0°
Crown gear
seen from above · magnet in the center pocket
0°
- Rod
- 0°
- Crown
- 0°
- Encoder
- 768 / 1536
- Motor
- 2.50 rev
- Per count
- 0.117° at the rod
- Wrap headroom
- 2.67×
TiltForge
Direct LAN control
Position 50% · open
encoder 768 / 1536 counts
State IDLE · Calibrated yes
Or drag either dial. A full sweep runs 5.0 s, the real tilt time.
A simulation of the page the firmware serves, driven by the gear ratio — not a live device.
Why these numbers
- 12 into 16 is a 0.75 step-down. A full 180° tilt turns the crown 135°, which is 1,536 of the encoder's 4,096 counts. That leaves 2.67× headroom before the absolute reading wraps, so no count in the travel is ambiguous.
- 0.117° per count at the rod. Finer than a blind needs — backlash and clutch hysteresis will limit you long before the encoder does.
- The magnet is one diametric disc. Ø4.7 × 3 mm, magnetized across its diameter, so the field turns with the crown. It started as a pair of 3 × 6 × 1.8 mm blocks laid poles-opposed to fake the same field, then a Ø6 × 2 disc; the real disc reads cleaner. It has to be diametric: an axial disc of the same size gives a reading that looks alive and means nothing.
- The printed part sets the air gap. The pocket runs Ø6.25 × 2.27 mm down from the recess inside the tooth ring and ends on a 0.20 mm printed floor, so the magnet sits just above the crown's bottom face. The gap is that floor plus the crown's running clearance over the sensor. The firmware prints the AGC on boot; on the bench the disc reads 80–100 on the 0–128 scale, graded good, and that beats any calculation.
- The board goes in inverted. The AS5600 is on the board's bottom copper, which is a fab fact, not an assembly one: installed, that face points up, at the magnet just above it.
- The pocket locates the magnet. Ø6.25 around a Ø6 disc leaves 0.125 mm of slack a side, inside the AS5600's ±0.25 mm misalignment budget, so the print sets where the field centers, not a glue step. Calibration learns the ends of travel, so which way the disc faces only moves the sensor's zero.
- The worm is why it holds. A 2-start module-0.8 worm, Ø26.6 at the tip, sits under the 20-tooth wheel: 10:1 with a 4.1° lead angle — self-locking, so the blind stays where you left it with the power off. Five motor turns cover the whole tilt, in 5.0 seconds. It is not in the drawing because you cannot see it looking down the rod, which is the view the drawing takes.
The part that decides the accuracy
The crown gear is the most important printed part in the set, because its tooth quality is the angle noise floor. Print tolerance, the printed journal, the PCB stack-up and the footprint RSS to 0.274 mm of misalignment against the ±0.25 mm AMS allows — about ±1.5° at the rod. Invisible on a window blind, and measured rather than hoped for.
The long version, including the trap where the sensor die sits 0.295 mm off the board's centerline, and how the magnet found its place (written before the switch to one disc), is in Where the Magnet Goes. The board itself is at Rev D.