Pendulum Clock with Atomic Precision

u19925 writes "Now you can get atomic clock precision out of your grandma pendulum clocks. Here is how it works: There is a camcorder fitted inside the clock which monitors the pendulum swing. It has an atomic clock signal receiver. It compares the pendulum swings with the atomic signal hearbeat. The camcorder also has an arm. If the pendulum clock drifts, then it uses its arm to push or pull the pendulum to make correction." It's not an April Fool's joke, but it is rather impractical.

Comments

Regularity?

[rjch]

Now if we could just invent something that would push or pull Grandma when she's not regular enough...

Why not magnets?

[FyRE666]

Why not just control the swing with a couple of magnets mounted at the ends of the pendulum's arc? It would surely be cheaper and easier to maintain than a camera and mechanical arm ;-)

Re:Why not magnets?

[Zeinfeld]

Why not just control the swing with a couple of magnets mounted at the ends of the pendulum's arc? It would surely be cheaper and easier to maintain than a camera and mechanical arm

These are not grandfather clocks, they are large public clocks and the movements are very old. The objective is to avoid human contact since people tend to break them advancing or retarding them for summertime.

So this is not a Rube Goldberg device, it is a piece of conservation technology :-)

The Westminster Tower Clock, with its famous bell 'Big Ben' is kept accurate by a warden who runs (ok shuffles, most jobs of that type go to aged war veterans) up a flight of stairs and adds or removes pre-decimal pennies from the pendulum bob. Ah you cry, but the time taken by a pendulum does not depend on the weight, well yes but the pennies slightly raise the center of gravity of the bob you see...

The camera bit sounds a little over the top, surely an led and a receiver???

Re:Why not magnets?

[scott1853]

okay, the real question is why do this at all?

Geez Batman, how are we going to figure this one out?

Quick Robin, to the article!

There is a growing shortage of people who are familiar with the workings of the large mechanical clocks on churches and public buildings, as routine maintenance tasks such as winding the clocks become automated. Yet they still need to be put forward an hour in spring and moved back again in the autumn without damaging their fragile mechanisms, some of which are 250 years old.

Re:Why not magnets?

[unitron]

"The camera bit sounds a little over the top, surely an led and a receiver???"

I'm guessing that you mean an opto-isolator type circuit. Other possibilities, a magnet embedded in the pendulum and a hall-effect device near each end of the path of the swing or a coil attached to a circuit which will detect the inductance change caused by the pendulum's proximity, or just let the pendulum brush across a contact pair and complete a circuit.

Re:Why not magnets?

[wwwillem]

We are used to /.-ers commenting without having read the article. That's even the fun of slashdot. But it appears that we have reached new levels: even people submitting a topic don't read their source anymore and everybody else follows like lemmings.

The New Scientist article doesn't mention any camera's, camcorders (why should you record this anyway, it's over in a second :-), or such. According to the article, the guy just uses a couple of IR sensors. That's a whole lot cheaper than camera's.

Still, this whole project is of course nuts. You love clocks (like I do !!) and than you have the honor to wind them every day, every Sunday at noon, or ..... That's just the fun of having old clocks.

Anybody can read the time from his cellphone. And using a GPS for the time-reading ... why doesn't the guy put a GPS on the pendulum and measure the frequency that way. That's cool!!!

I was going to do this

[Zeinfeld]

Actually this was a project I had in mind, but it can be done with much less kludge.

My plan was to put a magnet on the pendulum and then put the regulation mechanism on the reverse. This would measure each swing of the pendulum from the emf induced in a coil on the back of the clock. This would also be used to advance or retard the pendulum if necessary.

Latency?

[shibbydude]

Wouldn't the latency of the net connection/camera/lever defeat the whole purpose of atomic precision? I mean, anyone can just reset thier clock once and a while to the "technical" standard time. Is this really accurate?

Re:Latency?

[AndroidCat]

The most accurate clock I have is an ancient "digital" flip-leaf that depends on the powerline 60 Hz for its time base. It's not very precise because the line freq does drift from time to time. However, once it drifts too far, the the power company applies a correction to bring it back into range. Accurate, but not precise.

My computers and other appliances use crystal clocks which are very precise. But they slowly drift, and no correction is applied (except when I net sync my computer) so they drift and keep on drifting. Precise, but not accurate.

The net connection/camera/lever arrangement may not be as precise as an atomic clock, but it will be very accurate. See how it works? ;^)

No you werent

[stratjakt]

You never even thought of it until just now.

C'mon, admit it.

RTFA people

[glenebob]

Doesn't anyone read the friggin articles?

Nowhere in the article are the words 'camcorder', 'grandma', 'arm', etc.

It wasn't designed to fit into a grandmother clock and it certainly doesn't use a camcorder. It uses an infared sensor to sense pendulum location and a 'piston' to modify pendulum swing, and it is being used to automate maintenance on large clocks in churches, etc. It can also set the clock ahead and back an hour for daylight savings time.

Gotta be the worst case of can't be bothered to RTFA I've ever seen.

Now, anyone who thinks it would be better to replace the clocks in Big Ben with some modern electronic thing... well... probably ought to be shot. This doesn't seem like a bad way to get those big clocks to operate a good long time without human intervention.

This is awesome...

[Polyphemis]

Sweet! Now the 'tick-tock-tick' my grandma hears every day of her life ominously counting down to her impending death can be atomically accurate! Thanks Slashdot!

Not as impractal as it first looks

[Chronos56]

I feel that some informed comments are required for this topic. I am a clockmaker, yes I work on 200 year old clocks all the time; and I have to say that this is a really neat merging of modern and antique technology.

The clocks that are being regulated are tower clocks, they are observed by hundreds if not thousands of people a day. It would be nice to know that they are on time. It would also be a crime to rip out the old pendulum movement and replace them with an electric movement. Another feature is that the old antique system can run for several days in the event of a power failure, it just won't be quite as accurate.

The movements in these clocks are heavy cast iron units with large gears and very heavy pendulums. Using a magnet system to attempt to influence the timing rate would probably prove ineffective. However using some sort of system to raise or lower the pendulum by just a couple of millimeters will affect the timing rate by several seconds a day.

These clocks used to be wound once a week by hand and the time would have been reset at that time. These days most of these clocks have been converted to an automatic winding system, thus they see much less hands on maintenance, automatic systems for regulating the clock become much more attractive.

As a side note, the tower clock in London, commonly known as "Big Ben" ("Big Ben" is really the name of the bell that is used to count the hours) is regulated by adding or removing one or two old English Pennies, the one that were about the size of an old American Silver Dollar. The clock is regulated to be as on time as possible on the Queens Birthday and on New Years Eve.

Going even further afield some of you might get a kick out of the elaborate astronomical clocks that were designed in the 1800's. These were astonishing pieces of engineering that have been known to take an astronomer to figure out all of the settings required to set the clock.

I guess my passion for my vocation is showing, I hope that I was able to add something of interest.

Chronos

Ultra-accurate mechanical clocks

[Wesley Felter]

Danny Hillis designed the Clock of the Long Now to keep time accurate to the second for 10,000 years, and it's completely mechanical.

Conceptually....

[wowbagger]

Conceptually, this is the same trick almost ALL accurate clocks use - it's called "discipline".

Consider where I work - we have a very accurate 10 MHz reference to sync all our RF gear to. We need that ref to be tracable to the National Bureau of Standards. Now, it would be somewhat impractical to check with the Bureau 10 million times a second (anybody want to run a fiber from Boulder to Wichita just for the time sync?).

Before I go on, let me point out the difference between precision, accuracy, and repeatability.

• Precision is how many decimal places you can put on a reading. Saying "it's 12:00" is not as precise as saying "it's 12:00:00.0001".
• Accuracy is how close you are to the right answer. Saying "it's 12:00:00.001" is not as accurate as saying "it's 13:00" if in truth it really is 13:01.
• Repeatability is a measure of how close you are to your previous readings - if you say "it's 12:00:00.0001", "It's 12:00:01.035","It's 12:00:01.002" when read several times in one millisecond, then you aren't very repeatable.

From a metrological standpoint, having more accuracy than repeatability is useless. Having more precision than accuracy is also useless. (Ignoring tricks like averaging for the moment.)

Back to the example. What we do is to have a very high precision and stable oscillator (we used to use a rubidium standard). It has a long term stability of about 10E-9 and a short term stability of 10E-12. In other words, over a short period of time the thing will drift not more than one part per trillion, and over the long term (days) it will drift about one part per billion.

Now, that is running next to a GPS receiver that gives us a time tick synced to the Bureau. Every second the GPS time is compared to the local time standard, and an error value is computed. That error value is averaged over a long period of time, and used to gently tweak the rubidium standard. Thus, over the long run the drift is reduced to level of the cesium clocks, about 10E-13.

So we have atomic clock accuracy but rubidium clock precision and repeatability.

Now, if you used the same sort of technique on a pendulum clock - measure the error between the clock and the GPS, average, filter, and apply - you would have atomic clock accuracy with pendulum clock precision. Granted, I would not want to use the clock's time for reporting astronomic phenomena where the precision must be very high, but for normal use this would be quite good enough.