Trotter 2.0

Four and a half inches of table, in three quarters of a minute.

Wind the drum through the opening in the lid and a latex loop inside the chassis holds 0.86 joules. Then the machine spends them badly on purpose. A continuous rotary damper, belted to the same two shafts that carry the legs, refuses to let the energy out quickly — so instead of one violent scuttle you get 4.6 in of walking that takes 45 seconds to finish.

Nothing on this page has been printed. Every figure comes off the CAD sources, the slicer, or a deterministic offline rebuild that runs with no network and no model in the loop. The last section says exactly where that evidence stops.

Standing

At rest, each blade simply hangs off its crank pin. Nothing is wound, and nothing holds the machine up but four flat shoes.

Wind

Lift it and turn the drum 2.00 turns. About 80 mm of cord comes onto the barrel, the loop stretches to 100.3 mm, and the legs cycle freely in the air.

Inside

Under the lid: the drum on the rear shaft, a 240 mm belt carrying the front shaft at 1:1, and a second belt taking a branch up to the damper.

Release

Let go. Across the whole release the shafts never pass 7.4 rpm. Without the damper in the belt line, this is the moment it would snap.

Walk

Diagonal pairs, half a turn apart. Each shaft turn carries the chassis 3.1 in — the marks ruled on the bench are inches.

Rest

It stops after 1.55 turns with the loop still stretched. 0.17 J never comes out of the band at all.

at rest, unwound

The knee is a shape, not a joint

one printed blade per corner

Each leg is a single printed blade with two holes in it. The lower hole carries the crank pin, so the top of the leg is dragged round a small circle. The upper hole is a long slot, and a guide sleeve bolted through the frame sits in that slot and never moves. Between them they decide everything the leg does: the blade has to swing as the pin goes round, and it has to slide up and down its own slot while it does.

The bend halfway down is not a knee. It is a fixed kink in the blade that puts the shoe under the chassis instead of out beside it. There is no second joint anywhere in the leg, and no spring, no cam and no return.

Which is also why this page can show you the real thing move: the pose the CAD exported is a valid mechanism pose — the two diagonal pairs are half a turn apart, and every ankle sits within three thousandths of a millimetre of the line through its own pin and guide — so the film drives the assembly rather than imitating it.

7 mmthe crank pin, off the shaft axis. The entire stride is levered out of this one small circle.
32 mmthe fixed guide sleeve, above the shaft. This is the point the blade pivots and slides against.
58 mmpin down to ankle. Pin, slot and ankle lie on one straight line inside the blade.
14 mmof sliding per turn: the guide runs between 25 and 39 mm from the pin and comes back.
12.6°each way, and that is the whole swing of the blade.
14 mmof lift at the top of the step, over a 39 mm reach at the shoe.

Side elevation of Trotter 2.0 showing both left blades: the long slot near the top of each blade, the crank disc it hangs from, and the flat shoe on its free ankle. — Elevation, drawn straight from the assembly. The long slot shows clearly at the top of each blade, the crank disc under it, and the free ankle pivot above each shoe. The near blade and the one behind it are half a turn apart, which is what a diagonal pair looks like at any instant.

Where 0.86 joules go

59% of it into a damper

The elastic is a specified part, not a guess. From a 20 mm natural span the loop is stretched to 100.3 mm, and it has to measure 11 N at 40 mm and 14 N at full span. That curve is where 0.86 J comes from, and 1.50 of the 2.00 drum turns sit inside the characterised part of it.

Almost none of it becomes movement. The damper takes the largest share straight out as heat; the brass journals and the guide slots take the next; and a third share never leaves the band at all, because the machine comes to rest before the loop returns to its natural span. What is left is the tan sliver, and the tan sliver is the walking.

0.51 J
Rotary damper. Bled off as heat through the 140 mm belt on the front shaft.
0.17 J
Slots and journals. Crank pins, brass sleeves and the long guide slot in each blade.
0.17 J
Still in the loop. The walk ends before the band does; this is what stays wound.
0.01 J
Motion. The chassis is barely moving at any instant. It simply keeps moving.

Winding drum 6.4 mm effective radius

The loop pulls a 0.8 mm braided cord off a barrel groove wound in a single layer, so the radius stays put as it unwinds.
Rear shaft 5 × 97 mm ground steel

Cut down from 5 × 100 mm stock and cross-drilled 2.3 mm for the crank bolts, with filed flats for the pulley grub screws.
Primary belt 2GT, 240 mm

Rear to front at 1:1 over matched 20T pulleys 99.5 mm apart, tensioned by a 623ZZ idler in a slot. Both shafts turn together, the same way.
Damper branch TD62L1-2000

A second 140 mm belt over 40T pulleys drives a continuous two-way rotary damper rated 200 mNm at 30 rpm. This is the part that turns a snap into a walk.
Four cranks 7.5 N peak pin load

Printed hips bolt through the shaft crossholes. Front-left and rear-right point one way; the other diagonal points the other. Get that phasing wrong and it paces instead of trots.
Shoes 26 × 62 mm rubber

A 2 mm rubber sheet under each 34 × 70 mm printed shoe, on an ankle that is deliberately free. Nothing holds the sole level but the floor, which is why it had to grow to 62 mm across.

Nominal case of the release simulation: solid-CAD masses, free base, free ankles, 0.65 sole friction, no more than 0.12 in the slots. Across 8 runs — nominal, low and high damping, high friction, each at two timestep schedules — forward travel came out between 4.6 and 4.7 in every time, with no sideways drift. Conditional evidence, not a prototype record.

25 printed pieces, 9 types

about 27.7 h on the plate

Each type prints flat on a named datum in its supplied pose. Nothing is scaled.
Part	Qty	Filament
frame	2	navy
floor	1	navy
bearing_cover	4	navy
lid	1	tan
drum	1	tan
crank	4	dark grey
leg	4	dark grey
spacer	4	dark grey
foot	4	grey
pieces in total	25

Trotter 2.0 from above and behind, showing the tan lid with its finger opening, the navy chassis, dark grey blades and grey shoes. — The colours in the table, printed. The opening in the tan lid is where a finger reaches the drum; the shallow raised panels around it are the lid's own stiffening.

On the printer

PETG, 0.4 mm nozzle, 0.20 mm layers — 0.16 mm around the small cross-bores if your slicer will do it — 4 perimeters, 5 solid layers top and bottom, 100% infill. Every piece fits a 220 × 220 × 250 mm build volume.

100% is not bravado. The simulated mass distribution is the solid CAD volume at PETG density, so lighter infill changes what the machine weighs and where it carries it, and the gates would have to be run again against weighed parts.

And what you buy

Trotter 2.0 is not an all-printed toy. The printed shell carries a drawer of metal: two ground shafts, 625ZZ bearings at 5 × 16 × 5 mm, a 623ZZ idler, two 2GT belts, four matched pulleys, the damper, 12 cut brass sleeves, 41 socket-head screws, 8 grub screws, 41 nuts and 17 washers.

The load paths run through metal on purpose. Every sleeve takes the clamp load itself, so no screw is ever tightened down onto a plastic blade.

Plan view of Trotter 2.0 with the lid and near frame made transparent, showing the tan drum, both shafts, the long primary belt, the second belt and the damper. — The same assembly with the lid, the tub and the near frame dropped away. Drum at one end, the long belt down the middle, and the damper standing above the front shaft on its own short belt. Everything that turns is in one plane, and it all turns at one speed.

Where the evidence stops

8 simulated release cases

The design was rebuilt and checked offline: no network, no model in the loop, sealed source and artifact hashes, and deliberately broken variants that the same gates have to reject. That buys a great deal. It does not buy a walking prototype, and both halves belong on the same page.

Checked in software

720 crank phases of static equilibrium on free ankles — worst margin 4.3 mm over a 62 mm contact line.
8 full release simulations with a floating base and no prescribed body motion. Forward travel 4.6 to 4.7 in every run.
Energy accounted end to end, with the dry-friction losses realised to 99.97% of the specified Coulomb work.
Damper radial load 8.4 N against its 13.4 N rating; peak guide normal force 2.6 N.
Wrong crank phasing, a disconnected crank, open or inverted meshes and stale sources are each rejected by the same gates.

Not checked yet

No prototype exists. Nothing here has been printed, assembled, or walked across a real floor.
The elastic curve is assumed: 11 N at 40 mm and 14 N at 100.3 mm. Measure the loop you actually buy with a force gauge.
Sole friction is assumed at 0.65 on a dry floor, slot friction at no more than 0.12. Both are conditional on lubricated hardware.
Printed fit, support scars, journal wear and fatigue life all need a real print. A watertight mesh says nothing about any of them.

The numbers

PETG, 0.4 mm nozzle

Standing, overall: 6.8 × 6.9 × 4.5 in
Printed weight: 8.6 oz
Printed pieces: 25 in 9 types
Material: PETG
Nozzle and layers: 0.4 / 0.20 mm
Infill: 100%
Build plate: 220 × 220 × 250 mm
Print time: about 27.7 h
Wind: 2.00 drum turns
Walk per wind, simulated: 4.6 in over 45 s

One wind, 45 seconds, and 0.01 J of it is the part you actually watch.