Hacker News new | past | comments | ask | show | jobs | submit login
The Beverly Clock (wikipedia.org)
177 points by ColinWright on Dec 28, 2021 | hide | past | favorite | 49 comments



It's not self-correcting, though, just powered by energy harvesting from atmospheric pressure changes.

The Clock of the Long Now was supposed to be self-correcting. There's a mirror setup to focus the sun on some part which expands and provides a correction input.

That project is decades behind schedule, although Bezos is now having one built for him on land he owns.


Apropos very long running clocks: I wonder how long we've been counting the days of the week for without interruption? Ie was there ever some time of great upheaval when eg the Jewish people slipped up, and accidentally phase-shifted their sabbath?


I was originally going to write "Since Friday, 15 October 1582", since that was the first day of the Gregorian calendar.

But the previous day on the Julian calendar was Thursday, 4 October 1582. So even though there was a jump in the day number, the weekdays remained monotonic.

So maybe the correct answer is 1 January, 45 BC, the day the Julian calendar took effect, which was apparently a Friday.


For anyone interested in such oddities, here's an email from a DEC engineer in response to an alleged bug report back in the 90s that goes into a brief but very deep dive of the history of calendar, including the jewel about the jump between the Julian and Gregorian calendars (and the continuity of the days of the week):

https://neosmart.net/forums/threads/an-extended-history-of-t...

More to the point, that email suggests that there's a good chance the continuity of the days of the week predates even the Gregorian calendar and goes back to the founding of the Roman empire, as presumably Mercedonius wouldn't have given anyone a reason to suddenly change the days of the week.


Does my unix date utility take into account these occasional leap seconds? And are these just hardcoded in the library itself and updated once in a while?


Leap seconds are supported by ntpd. Leap seconds are announced in advance at this fixed URL, which also contains the list of past leap seconds: https://www.ietf.org/timezones/data/leap-seconds.list

...which can be added to ntpd's configuration if you want your system's leap second to occur exactly on time.

Even if you don't, the leap second will eventually take effect via an update from your configured remote ntp servers, or your GPS source, or wherever you get your time from.


Thanks!

I wonder whether the continuity of the Jewish week predates the Julian calendar?


Yes, it’s been 7 days since the creation.


We might be able to discover if this had happened - for example if an eclipse occurred in historical documents on a certain Friday, but we would be able to calculate that it had actually taken place on Saturday, indicating that a day had been "lost" between then and now.


Interesting query. I wonder if anyone ever realized this had happened and came to the conclusion that the ceremony around it all, now that it's not on the "right" holy day, is completely man-made.


The clock of the long now is an amazing piece of engineering.

I think we can give them some slack for being a bit behind. Building a clock inside a mountain designed to keep time for over 10,000 years is going to run into some issues.


While cool,

> While the clock has not been wound since it was made by Arthur Beverly in 1864, it has stopped on a number of occasions, such as when its mechanism needed cleaning or there was a mechanical failure, and when the Physics Department moved to new quarters. Also, on occasions when the ambient temperature does not fluctuate sufficiently to supply the requisite amount of energy, the clock will not function. However, after environmental parameters readjust, the clock begins operating again.


We’ve had one of the Atmos clocks mentioned in the article at our father’s house for years. It counts up to the year 3000, but I imagine it will need to be serviced a few times before then.


That is crazy haha


One can only imagine the strange wonderful world we might inhabit if steampunk/clockwork tech became the dominant player in all industries.


I think we got a taste of that in the late 70s when the engine bay of every automobile was a dense nest of rubber vacuum hoses.


What about late 70s engine bays needed having (more?) vacuum hoses? Is there a vehicle that's a good example of this?


The federal government had super strict emission laws for vehicles starting in the 70s. This was before computers were cheap and reliable enough to stick in average vehicles, so the way automotive engineers got it to work was to have vacuum lines to send signals to the carburetor and distributor. They probably could've done it with analog electronics, but you already have a free source of vacuum (the intake manifold), and there isn't an simple way to convert an analog voltage to a linear position, while you can do that with a diaphragm and vacuum.

For a good example, any American V8.


Peak insanity might have been this mid-80s Honda:

https://www.autoweek.com/car-life/but-wait-theres-more/a1850...

Especially this picture:

https://hips.hearstapps.com/autoweek/assets/s3fs-public/03-m...

Eventually the rubber hoses would get old, crack, and leak vacuum. Now your car starts running wonky and diagnosis is pretty much impossible.


Thanks for illustrating the nightmare that three decades later will still roust me from a sound sleep. Back in my pro wrench days, I had to remove-and-replace the head on one of those Hondas. What should have taken an afternoon soon turned into a hellish day-and-a-half of squeezing a vacuum tester.


Any carb car, or even plans. Though you can argue an array of vacucum lines exist in 90's-current cars today because of emissions.

There's also the car community idea of a clean engine bay, or wiretuck.

https://preview.redd.it/3ddncz8e2ao21.jpg?auto=webp&s=e75f4a...

to

https://imgur.com/nuZGt


not 70s but my Mercedes 300TD from 1985 made extensive use of vacuum hoses to power lock doors, run pumps and even to turn off the engine (it's diesel so no spark plugs.) I think it was built that way because electronics were more expensive?

the engine was a marvel of engineering. I got it to 400K+ miles (odometer eventually broke), and I could drive just fine even after my alternator broke and I had no electrical power whatsoever. I also could (and did) do extensive maintenance myself because there were no computers to fiddle with and everything was modular.


> I think it was built that way because electronics were more expensive?

Mercedes has been the king of Rube Goldberg vehicle systems for ages. Anyone I know with one has had numerous electrical and other reliability issues. Shame as they used to make decent vehicles that actually lasted but boat anchored them with technological clown circus.


I have an Atmos clock on the mantlepiece but as the article alludes to, although it is perpetually in motion it does not keep time very well.

I love it as a mechanical wonder but I would love it more if it were useful as a timepiece.

I've been trying to think of ways to measure it's timekeeping such that I could make the necessary adjustments to fix it, but it's also so delicate that I'd rather just leave it alone.


At least for some form of mechanical clocks: There is an app for that :)

E.g., for Android, the "Watch Accuracy Meter", which can be found in the Play Store or the APK source of your choice, uses the phone's microphone to measure the frequency of mechanical watches.


The Atmos does not have a seconds hand. All models except two do not have minute indications.

I believe that is going to difficult measuring accuracy. You can approximate something if you compare it over a long period of time, perhaps at 12 o'clock, or as much 12 o'clock as you can tell by inspecting the hands alignment.


I wonder why this isn't used to store power in a battery that's used for IoT devices?


My math might be utterly wrong, but it looks like a Teensy in sleep mode consumes 0.23 milliamps at 3.3 volts, which comes out to 66 joules per day. The OP mentions that the mechanism provides about 13 millijoules per day, which is 5,000 times lower.


Teensy is nowhere close to most power efficient options out there. A power efficient arm can be running on micro or even nanoamps in sleep mode. It will of course depend on the rest of the circuit surrounding MCU.

I use STM L4 for my projects.

https://www.st.com/resource/en/product_training/STM32L4_Syst...

On page 4 it says down to 30nA with I/O wakeup, 350nA with 32kB memory retained.


you would probably get a lot more energy in cumulative milliwatt-hours per week/month from having a tiny 2 x 1cm size photovoltaic cell such as from a cheap desk calculator exposed to ambient room light, and something like that has no moving parts.

if we are talking about actual practical IoT things a single "big" solar cell (125x125 or 156x156mm) size is quite cheap, and you'd run the output of the PV cell directly into a tiny battery charge controller circuit configured to keep one 18650 or 21700 size battery at its float voltage.


Size and power? One cubic foot of air generates 3.6 microwatt hours. This clock is a lot more efficient than an Arduino!

Also, temperature controlled environments might not have a 6 degree day to night cycle.


Too little energy and too unreliable. The article mentions that this clock stops sometimes and needs to be manually readjusted once it is back.


I guess that tire pressure sensors use batteries. Always seems like they should just be able to harvest power when you're driving, and have plenty of juice to get past the idle times.


I wonder how slowly the airtight box leaks.


This amazing clock has never been wound and yet sometimes works!


> on occasions when the ambient temperature does not fluctuate sufficiently to supply the requisite amount of energy, the clock will not function

So it has stopped.


Yes, but it was still right twice a day.


Only on those days when it stopped for a full 24h (adjusting that for leap second as appropriate).


But it is “still running” and “never been wound”!


Fascinating piece of machinery, but have to wonder why the article doesn't state whether it tells the correct time or not.


It's a mechanical watch and not self-correcting. So obviously, it'll drift over time.


A quick mention of accuracy at least would be worthy of inclusion, perhaps in relation to other mechanical clocks

e.g. this one is accurate to 1 second per year. https://en.wikipedia.org/wiki/Shortt%E2%80%93Synchronome_clo...

With that kind of accuracy the Beverley Clock would potentially only be 15 seconds out (putting aside the times it has stopped)


Accuracy is the least interesting part of the clock. The power mechanism could be paired with any random timekeeping device.


Might be the least interesting, but it's the sole purpose of a clock.


The interesting thing here isn't a clock, it's the novel power harvesting device. There simply happens to be a clock tied to the output end of the shaft.

You could get some random highly-accurate clock and put that on instead, or you could put a wheel with an arrow on it, and it would be no more or less interesting.


No, long term accuracy is not the sole purpose of a clock.

People make trade-offs depending on application. Eg if your clock loses a minute on average once a week, that's still good enough for most household applications.

Similarly, wristwatches put a premium on low weight and small size, not necessarily on accuracy. (And that's not even mentioning clocks as status symbols.)


Wasn't meaning specifically long term, to be fair I get your meaning but you'll be needing to know the accuracy/inaccuracy for it to be useful beyond a status symbol.


Oh, definitely. And I would have like the article to include that information. Even if it was just to say that it's accuracy is nothing to write home about either way.


In some sense you are right that accuracy information would only be a slight benefit, but adding a single sentence about accuracy to the article would also only be a slight cost.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: