Mechanical Clocks that Never Need Winding

What is it about mechanical clocks? Maybe it’s the gears, or the soft tick-tocking that they make? Or maybe it’s the pursuit of implausible mechanical perfection. Combine mechanical clocks with “free” energy harvested from daily temperature and pressure variation, and we’re hooked.

Both the Beverly Clock, built by Arthur Beverly in 1864, and the Atmos series of clocks built between 1929 and 1939, run exclusively on the expansion and contraction of a volume of air (Beverly) or ethyl chloride (Atmos) over the day to wind up the clock via a ratchet. The Beverly Clock was apparently a one-off, and it’s still running today. And with over 500,000 Atmos clocks produced, there must be some out there.

Although we had never heard of it, this basic idea is really old. Clicking through Wikipedia (like you do!) got us to Cox’s Timepiece, which is powered by the movement of 68 kg of mecury under atmospheric pressure. It is currently not running, but housed in the Victoria and Albert Museum in London. Even older is a clock that we couldn’t find any info on that dates from 1620, invented by Cornelius Drebbel. Anyone know anything?

We’ve had energy harvesting on our mind lately, and the article on the Beverly Clock says that it gets 31 μWh over a day when the temperature swings by 3.3 °C. Put into microcontroller perspective, this is 0.39 μA at 3.3 V, so you’ll have to be pretty careful about your sleep modes, and an LED is out of the question. How amazing is it, then, that this can power a mechanical clock?

Thanks [Luke], [hex4def6], and [Wallace Owen] for tipping us off to these in the comment section!

27 thoughts on “Mechanical Clocks that Never Need Winding

  1. “What is it about mechanical clocks? Maybe it’s the gears, or the soft tick-tocking that they make? Or maybe it’s the pursuit of implausible mechanical perfection. Combine mechanical clocks with “free” energy harvested from daily temperature and pressure variation, and we’re hooked.”

    Aside from tickling all the geek boxes, something like this would fit perfectly into any alt-history fiction. No electricity, no miniaturization needed, reasonable level of skill.

        1. I checked out the rather bizzare clock they are building with all its artsy style and behavior but they seem to have glazed over what it takes to keep it accurate for 10000 years. As I said in a previous post, tides could be used for timing to keep it synced. They are likely doing something to address that issue, but with so much emphasis on style, I’m starting to wonder if they are missing out on some of the engineering while trying to make it pretty.

  2. “How amazing is it, then, that this can power a mechanical clock?”

    It can’t. That’s just the effect of one temperature swing lifting the 1 pound weight by 1 inches. A ratchet can extract energy out of the bellows going both ways, being a push-pull action, so during the course of a day it lifts the weight twice, or possibly multiple times if there’s a draft, or a fire is lit in the room, or a cloud passes overhead and blocks the sunlight falling in through the window etc.

    The weight is simply hanging by a long chain, so it has multiple days of reserve power, and stays winded by the average number of temperature changes. The clock does stop at times due to lack of temperature variations.

    1. “That’s just the effect of one temperature swing lifting the 1 pound weight by 1 inches” You’re right! Thanks.

      I was really thinking of the daily temperature variation as being more simple sine-wavey than it probably is. And that it gets a click on its ratchet from significantly smaller temperature changes in both directions is even cooler than I had thought.

      There’s an academic paper on the clock, but it’s paywalled and I didn’t dig too deep into it. Maybe I should. How much energy _does_ this thing harvest in an average day?

      OTOH, efficient mechanical clocks _do_ use very little energy to run. Anyone have a ballpark number on that?

      1. Obviously the diaphgram in the bellows can move different amounts and the mechanism can be designed to take a “bigger bite” if the temperature varies by a greater amount simply by having more teeth in the ratchet. It also works by atmospheric pressure variations in general, as well as temperature changes.

        But the main point was that it probably has a mechanical rectifier movement that winds the mechanism when the bellows are moving in AND out to double the available energy. Maybe it doesn’t, but that’s the way I would design it: if you have a ratchet wheel, the pawls can push it one way at the top, and the other way at the bottom, maintaining the same direction of rotation.

        As for the energy, a 1 kg weight over 1 meters stores 2.7 mWh of energy. Long case clocks were built with power reserves of 30 hours, since they needed re-adjusting every day. With more accurate mechanisms, they were built with power reserves up to 8 days. Beyond that, they started building them with electromechanical kickers instead of weights. Torsion pendulum clocks use far less energy, with the escapement changing direction every 10-20 seconds since the oscillation rate is not set by gravity, and those were built with coiled springs with enough energy to last 400 days.

        1. As for the winding mechanism itself – again not knowing exactly how they were built – as simple version is to have two wheels. One is driven by the winding ratchet, the other drives the clock, and a loose chain loops around both wheels. The chain droops down between the wheels, pulled down by the weight riding on a pulley, so when the driving wheel is rotated by the bellows, it lifts up, and the driven wheel lets it sink down as the escapement runs, thus enabling continuous or intermittent winding without disturbing the clock.

          Of course the force on the driven wheel is continuously changing because there’s a slight extra pull on the chain whenever the weight is lifted, so the escapement needs to be adjusted a little slow to account for the occasional tugs on the chain. The difference is neglible, because the escapement probably isn’t accurate enough to begin with, and the clock needs re-adjusting every few days or weeks.

      1. Park it in the hallway of your house (where most ‘grandfather clocks’ had their place of old), and you’d get more temperature changes. Unless you never have visitors and never leave the house.

    1. Don’t they? Or do you mean purely using atmospheric pressure, rather than how these use ambient pressure as a relative reference compared to the expansion of another working fluid. I think you’d need a pressure gradient (another fluid or a vaccuum) for work to be done, but I could be wrong.

  3. It’s fascinating that the Atmos clock can run for years without human intervention, but torsion pendulum clocks are not know for accuracy, and it’s doubtful that it would keep decent time for a year. The only way to make such a clock accurate is to keep it in a temperature-controlled environment, which defeats the self-winding function.

  4. Not are they relatively expensive they cost a small fortune to clean and there is a waiting list.
    I have a minute repeater, Rolex’s are junk comparatively speaking. A nice repeater will set you back 20-30 K and they go up from there.

  5. The ideas here are wonderful but they don’t go the distance on the way to perfection. The perfect clock never accumulates error over time. (like sun-dials) It would be so cool to make a passive mechanical detection system that detects the lunar induced tidal changes and then compensates for it so as to make detecting solar tidal changes possible and then use that as a source to help gently tune clocks like this. Then one would have a self winding, self regulating clock that stays true to the actual time indefinately until the bearing wear out. :-) That would be a beautiful thing. Now that would be one hell of an engineering thesis project wouldn’t it?

      1. I found it on their site. Personally the idea any optical system staying functional for 10000 years is pretty extreme. I’d still look at tides. (Definitely playing arm-chair engineer here. Its easy for me to sit here and express opinions while they build a clock. I hope it works out well for them.) LLAP :-) 🦊

    1. >”self regulating clock that stays true to the actual time”

      But what is “actual time”? The length of the day is variable, so a clock that accurately and precisely measures steady intervals of time – actual time – is going to run out of sync.

      If you’re using astronomical time as your standard, then you have to deal with the fact that a second is not a second, and all your other definitions change with it. Like the definition of a Newton as kg-m/s^2 changes subtly when the earth’s rotation slows down, and all your weights and measures are shifting around depending on what day it is.

      1. Without getting into an argument of what “actual time” is, the best explanation I could find comes from the famous quote “Time is what clocks measure.”, which is probably circular, but for the sake of this context, I’m not concerned. Large stationary clocks that serve the local area almost invariably match up with the day/night cycle with one day being exactly the time it takes for the Sun to pass through the viewers prime meridian in the sky to the point in time when it does it again. Subdivisions are then made to get hours, minutes, and seconds. I know it isn’t “scientifically” precise but it is contextually precise and would be more than good enough for everyday use. Upon review, there apparently is a solar synchronization system with this clock, so it should be every bit as accurate over the centuries as any sundial. :-) 🦊

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s