Scratch-building

A lift bridge over the canal that has to work twice an evening

The canal was in the plan from the first evening, and for two years it was crossed by a fixed girder bridge that none of us believed in. Replacing it with a lift bridge that genuinely rises took a winter, and the mechanism underneath is the part I would do differently.

At the bench: 5 November 2024
Typed up: 19 January 2025
Reading: 8 minutes
Room: Amsterdam, Noord-Holland

Stop 01 of 04

Why the bridge had to move at all

A polder branch crossing a working canal inherits a problem: the water is a route and the railway is an interruption. Every fixed bridge we found in photographs of the real thing sat high on long approach embankments, which is exactly the thing an attic layout cannot afford in length. A bridge that lifts sits low, the approaches stay short, and the crossing becomes an argument between two kinds of traffic rather than a piece of scenery. On an operating evening that argument turns out to be considerably more interesting than a viaduct.

There is also the question of what a model railway is for. Ours is run to a sequence with four people standing round it, and a bridge that must be raised for a barge and lowered again before the afternoon goods gives one of those people a job that is not driving. The fixed girder we started with was tidy, plausible and completely inert; nobody ever looked at it twice. On the first evening after the lift bridge went in, every single person wanted the job of working it.

Stop 02 of 04

Drawing it from the wrong side of the canal

I spent three Sunday mornings riding out along canals to the north, sketching bridges from whichever bank had a bench on it. All the useful detail is underneath: the pivot boxes, the bracing on the counterweight beams, the little cabin that is almost always on the same side as the road. From a distance a lift bridge is two triangles and a deck, and it is perfectly easy to model that and end up with something lifeless. Up close it is a dozen small decisions about where the mass sits, and a photograph taken from the opposite towpath flattens every one of them.

The drawing that finally worked was made at full size on the back of a roll of lining paper, with the deck length taken off the model and everything else scaled to it by eye. Working at full size meant I could cut card templates straight from the sheet and hold them over the canal before committing any brass. Two of the first templates were obviously wrong in place and not wrong at all on paper, which is as good an argument as I know for not drawing a model bridge at drawing-board scale.

Stop 03 of 04

Brass, wire and a counterweight that lies

The deck and towers are soldered brass section, chosen because they have to take handling twice an evening for years rather than sit still for a photograph. The counterweight beams are brass too, ending in a hollow box that looks like concrete and contains lead. The mechanism is deliberately stupid: a length of steel wire inside a brass tube runs down one tower to a crank under the baseboard, and a hand turns it. We tried a small motor and gearbox first. It was slower than a hand, it hummed all evening, and when it jammed it kept trying.

The counterweight is a lie, and admitting that early saved a great deal of grief. On the real thing it balances the deck. On a model the deck weighs almost nothing and pivot friction is the dominant force, so a correctly proportioned counterweight simply sits there looking correct while the wire and crank do all the work. Anyone who puts their head under the baseboard will see the fraud immediately. In two years of Tuesday evenings and a handful of visitors, nobody ever has.

Written on the card

Build the deck for handling, not for the photograph
At this scale pivot friction beats counterweight balance every time
A hand-turned crank is quieter and more forgiving than a motor
Make the visible half honest and the hidden half simple

Stop 04 of 04

Making it survive two hundred lifts

Wear shows up in three places: the pivots, the wire, and whatever the deck lands on. The pivots are brass tube running on brass rod, which is ordinarily a poor bearing, but they turn perhaps ten degrees a few times an evening and a drop of light oil twice a year has kept them tight enough. The wire stretched during the first month and then stopped. The rail joints at each end of the deck were the real trouble, and they had to be rethought twice before they stopped producing a stall in the middle of a move.

The deck rails now finish a hair short of the fixed rails and the electrical connection is made by two sprung contacts underneath, rather than by rail ends touching each other. The deck itself lands on two brass pads that set the height exactly, so the rails never take the impact of it coming down. Alignment gets checked with a short four-wheel van pushed slowly across, and if that crosses without a sound the bridge is right. The test takes fifteen seconds and has caught every problem before an operating session did.