Product · Desk lamp
One ring sets height, tilt and direction.
CAD-developed concept2017 conceptIndependent project
Designed and analysed — prototype in progress
TASK · ring at 170°
Turn the ring: drag round the dial, or ← → · Page Up / Down: the next pose
AXON is a name I've carried for more than fifteen years, first as a business idea. The lamp came in 2017: one ring that sets height, tilt and direction. I began engineering it in late 2025; in 2026 it became a frozen, checked CAD.
From STOWED to TASK the head rises 180 mm and tilts 30° while the pan holds at 0° (θ 0–172°). Then the pan turns 180° by θ 268°, as the head rises 60 mm more.
| Pose | θ | h | φ | β |
|---|---|---|---|---|
| STOWED | 0° | 160 mm | -90° | 0° |
| TASK | 170° | 340 mm | -60° | 0° |
| AMBIENT | 270° | 400 mm | +30° | 180° |
Each pose lands on its target height, tilt and direction: φ and β are off by 0.00°. h is measured in the model's frame. The ring turns 270° between its end stops. Designed · kinematic model
A cam, a gear train and a linkage were each modelled to do the same job. All three hold every pose by geometry, so holding did not decide it: the cam needs 254 N·mm at the ring against 385 and 399, and it is the only one that nests at STOWED with no collision inside the mass target. Its cost is the part count.
| Phase 1 stick models | cam | gear | linkage | Level |
|---|---|---|---|---|
| Peak ring torque, re-sized for a measured μ 0.20 | 254 N·mm | 385 N·mm | 399 N·mm | Designed · stick model |
| Collisions when nested at STOWED | 0 | 2 | 0 | Designed · stick model |
| Lamp mass needed (target ≤ 2000 g) | 996 g | 1,546 g | 2,576 g | Specified intent · stick model |
| Parts, in pieces, screws included (the cam's honest cost) | 97 | 61 | 128 | Designed · stick model |
| Printed sliding contacts under load | 7 | 4 | 2 | Designed · stick model |
The cam's 97 counts pieces, 21 of them screws, clips and retainers. The CAD has 79 pieces in 67 drawn parts and counts no screws: it draws them as holes. The 32 loose balls of the ring's race became one stock bearing; the head grew from one printed shell to 16 pieces.
| What changed | Phase 1 | CAD | Change |
|---|---|---|---|
| The 32 loose balls of the ring's race became one stock bearing, drawn as its two races | 32 | 2 | −30 |
| Screws, clips and retainers: the CAD draws screws as holes only, so it counts none; some of those joints became bonds, presses or a pin | 21 | 1 | −20 |
| The head: one printed shell became a machined can and plates, the LED, optic, dome, ballast, its bearings and spring | 1 | 16 | +15 |
| Added in the CAD: liners, ball bearings at the arm's pivots, the rack's backing roller, the dock insert, the cable and its carrier | 0 | 20 | +20 |
| Smaller changes: the ring split in two bands; four pins became two pins and the head's pivot tube; two bushings and a thrust washer became a bearing on a journal; the detent's ball and spring one plunger; the ballast disc dropped | 11 | 8 | −3 |
| Kept one for one: spiders, shoes and their pins and bearings, sleeves, tongues, column, hub, bracket, axle, arm, crank, rod, slider, floor, core, cap and foot | 32 | 32 | 0 |
| Total | 97 | 79 | −18 |
Every line of the Phase 1 list (44 lines) and every drawn part (67) is counted once. Neither count is a production bill of materials. Designed · counted on the models
Each shoe rests on a groove flank cut 10.29° steep, under the 11.31° friction angle, so the head's weight cannot turn the ring. The detents only click and the end stops only stop: neither holds a pose.
Specified intent · provisional friction
The flank is 10.29° against a friction angle of 11.31° at the provisional μ 0.20: a holding margin of 1.02°. The selection rule asks for at least 2° at the lowest measured μ, so the groove lead is chosen after the friction coupons are measured; the band holds down to μ 0.184. Lifting the detent ball takes 89 N·mm at the ring.
Problem
80.2° overhang
The two shoes' loads point opposite ways at every lead and angle, so in a one-piece drum one of the two locking flanks would print face-down.
Fix
≤ 45° overhang
Two bands, ring A printed upside down and ring B upright, each locking flank face-up, joined at a lapped seam.
Problem
0.48 mm sag
A 2.5 mm printed floor plate under a 2.0 kg lamp sinks 0.30–0.48 mm at the rim and closes ring B's 0.5 mm running gap.
Fix
16 µm
The same plate in 6061-T6, 23–34× as stiff in bending: the rim gap stays at least 0.48 mm, for 73 g more.
Problem
4.3 mm off the mesh
A 5 N push on the stowed head bent rack B's blade away from the tilt pinion: the head front moved 13.7 mm and the teeth parted past their working depth.
Fix
1.92 of 2 mm engaged
A flanged ball bearing on a dowel behind rack B, opposite the mesh: the deflection under the same push is 1.65 mm (base held), against a 2 mm target.
67 drawn parts (79 pieces) in 12 layers out from the fixed core. 12 of them change shape with the groove lead, so they are made after the friction test sets it.
Designed · counted on the model
The floor plate, the stop pin, the cord carrier and the foot are fitted last, after the ring is set at TASK: the explode shows where each part sits, not the build order.
Every figure here comes from the model and carries its level. Two targets are not met by design; they are shown, not hidden. Nothing has been measured yet: of 97 recorded claims, 40 are designed and 57 specified intent.
| Figure | Model | Target | Status | Level |
|---|---|---|---|---|
| Free play through the ring, ring only (R-12a) | 2.71 mm | ≤ 0.5 mm | not met by design | Specified intent · fit model |
| Play under a reversing 5 N push (R-12b) | 17.07 mm | ≤ 17 mm (provisional) | not met by design | Specified intent · fit model |
| Head deflection under 5 N, base held (R-12) | 1.65 mm | ≤ 2 mm | met by design | Designed · beam model |
| The same, standing on its foot | 1.68–1.97 mm | ≤ 2 mm | met by design | Designed · beam model |
| Holding margin, flank under the friction angle | 1.02° | ≥ 2° once μ is measured | decided by a test (TP-01) | Specified intent · provisional friction |
| Detent breakout against the floor, at μ 0.20 | 1.65–1.91× | 1.5–2× | met only at the provisional μ | Designed · provisional friction |
31 targets at the CAD freeze: 1 accepted by the owner, 14 met by design, 9 decided by a test, at assembly or on receipt, 5 not set, 2 not met by design; 0 measured.
The pass criteria were written down before any reading, and the analysers stay frozen until real data arrives.
Clearance and press ladders, each bracketing its fit; a ladder that does not bracket is reprinted, never extrapolated.
Specified intent · test protocol2942 g on the shoes (9.62 N per shoe). Pass: run-in μ spread ≤ 0.06 with CV ≤ 15 %; |μ run-in − μ fresh| ≤ 0.05; no cracking or transfer; the lever end moves ≤ 1 mm under hold.
Specified intent · test protocolHolding margin ≥ 2° at the lowest corrected μ; lamp ≤ 2000 g; nested gap ≤ 15 mm. The rule sets the lead; the 12 lead-dependent parts are made after it.
Specified intent · fixed before dataFree play R-12a ≤ 0.5 mm and R-12b ≤ 17 mm; deflection ≤ 2 mm under 5 N; the base walk at desk μ ≥ 0.6; the detent's breakout 1.5–2×.
Specified intent · targets
Renders of the frozen CAD in its three poses and close up, in the materials specified for the build. Two finishes are assumed until chosen: matte black for the printed parts and a clear anodise on the head's can and rear plate. Nothing is built, and no figure on this page rests on these pictures.
The specified screws, the cord's exit and the reading lines are added from their specs; nine screws whose type is not chosen yet are left out, their holes shown empty, and the index mark's size and the cord's run across the desk are assumed. The beam follows the optic maker's file measured with a similar LED, its brightness set for the picture. The desk and the room are set dressing.
Next project · Packaging · Whisky
Original whisky decanter and walnut coffret, re-engineered for a 1.5 L fill
