Brass Gear Clock Bench

Lyon · France · Anno 2015

Domestic mechanical movements, mended one at a time and written down the same week

Entry II · Springs

Letting down a mainspring before it lets you down

A wound mainspring in a domestic clock holds enough energy to break a click spring, a wheel tooth and a knuckle in the same half second. This is the sequence I use to get the power out, and the one time I did not follow it.

Written
26 February 2024
Corrected
14 September 2024
Length
8 minutes
Bench
Lyon

What is actually stored in there

The mainspring out of an ordinary French eight-day movement is around twenty-eight millimetres wide, a shade under four tenths of a millimetre thick and roughly two metres long when it is free. Fully wound, that band of steel is holding the whole week of timekeeping plus the striking work, and it wants to be somewhere else. People think of clock springs as gentle because clocks are quiet. The spring does not know it lives in a quiet object.

Everything that goes wrong inside a movement during dismantling goes wrong because that energy found a path nobody planned. A click that is worn and lets go under load will spin the winding arbor faster than you can react. A barrel hook that fails while the plates are apart will throw the spring out of the barrel across the room. Neither of those is common, and both are entirely avoidable by not having the power in there when you take it apart.

The order of operations, written on the wall

The sequence is short and it does not vary. The clock is stopped and the pendulum lifted off its suspension. The hands come off, then the dial, so nothing fragile is in the way of my elbows. The let-down key goes on the winding square, one hand holding it firmly, the other hand ready at the click. Only then does the click get lifted, and only a fraction, so that the ratchet wheel can move a tooth at a time.

Then the key comes back in small increments, letting the ratchet slip one tooth per movement, with my wrist braced against the bench rather than held in the air. On a big eight-day barrel that can be fifty or sixty tooth-steps. It takes about ninety seconds per train and it is ninety seconds well spent. Striking trains get let down as well, and on the movements that arrive from country houses the striking spring is often the stronger of the two.

The single most useful habit I picked up is that my hand never leaves the key until the spring is fully down, and the click never gets held open with a screwdriver wedged against something. The moment you introduce a wedge, you have handed control of the energy to a piece of steel that is not gripping anything in particular.

The afternoon it slipped

In the spring of 2018 I was letting down a striking spring in a wall movement and the let-down key was slightly too small for the square. I knew it was slightly too small. I used it anyway because the right one was in the drawer on the far side of the room and I was three minutes from the end of the job. About eight teeth from the bottom the key rounded off the corners of the square and jumped.

The ratchet ran, the click chattered, the winding arbor spun and caught the side of my index finger, and the whole remaining wind dumped into the barrel in about a second. I was lucky. The cost was a bruised finger, a click spring that had to be made new, and a winding square that I had to dress and that will now always take a slightly larger key. Somebody in forty years will look at that square and wonder what happened, and the answer is that a man was three minutes from the end of a job.

The changes afterwards were small. Let-down keys live in a block screwed to the bench edge rather than in a drawer, so the correct one is never further away than reaching. And I test the fit of the key on the square with a light turn before the click is touched. If it rocks at all, it is the wrong key.

Barrels, and why I stopped opening them on the bench top

Even fully let down, a mainspring in its barrel is coiled against its own shape and will expand suddenly when the lid comes off. For years I popped the lid on the open bench and caught what happened next. That works until the day the spring is broken near the inner coil and the whole thing leaves the barrel at speed, which happened to me once and put a fine scratch across a brass dial plate that was sitting a foot away.

Now the barrel goes inside a cardboard box on its side, lid uppermost, and comes apart in there. If the spring wants to come out it can do it inside a box. The other change is that I no longer pull springs by hand at all. They go in and out with a winder, which also lets me see the free diameter of the old spring before it goes back in.

That free diameter is worth measuring. A spring that has set will have a free diameter well under two and a half times the barrel diameter, and one that tired will never deliver even torque across eight days no matter how clean the train is. I have wasted whole days chasing a rate problem that was a set mainspring the entire time.

Checking the spring before it goes back

Out of the barrel, the spring gets cleaned, dried and looked at along its whole length against the window. I am looking for two things: a blue tinge from somebody overheating it years ago, and fine cracks across the width, which usually start near the inner eye or at the hook slot. A crack at the inner eye is a spring that will break, and it does not matter how many years it has already survived.

The other check is the hook itself and the slot in the barrel wall. On a clock that has been running a century, the slot has often worn into a keyhole shape and the spring end slips under load. That produces the maddening fault where the clock stops at random after four or five days with plenty of wind left on the key, which sends people looking at the escapement for weeks.

When the spring goes back I write its measured free length, width and thickness into the job book. It takes a minute and it means that the next time this movement comes back, whether to me or to somebody else with the same figures in front of them, nobody has to guess whether the spring in there is the right one.

Pivot diameters come off a bench micrometer reading to a hundredth of a millimetre, rates off the oak board behind the bench taken at the same hour each evening, and the dates are the days the work was actually done. This entry was last corrected on 14 Sept 2024.

Back to the numbered record