Ever own a watch, particularly a digital watch, which always seemed to gain or loose about the same amount of time each month?

Each time you set the watch, why can't the watch realize it's been too slow or fast since the last time it was set and attempt to regulate itself?

To do this properly, the time piece would need to maintain a small amount of memory to keep track of the last time it was set and by how much it was off. Using this info, just like a phase locked loop (PLL) the watch could narrow itself closer and closer to an acceptable accuracy.

Unfortunatly, the precision of a watch usually varies based a tiny bit based on temperature. Different watches run at different speeds on different people. A wristwatch could have a temperature sensor in it to try and correlate automatically the changes in temperature to the changes in speed.

The wearer would have to set the watch using the same synchronized time source each time. The more often it was set, the more data points the watch can collect in order to set up the PLL. Such synchronized time is widely available: the talking clocks which you can telephone, the hourly beep you hear on some radio stations, and even software exists which uses NTP to accuratly set the clock in your computer.

Perhaps it would be useful to have a "regulate" mode and a "set" mode where the wearer would put the watch in regulate mode for the first few months or when he/she wants to reregulate the watch.

Such regulation would essentially ignore changes in minutes greater than say 7 minutes. All of the time zone changes I've encountered are on 15 minute boundries (between India and Nepal for example).

Drawbacks to this idea are that it could be quite complicated for the average consumer to use. Consumers who didn't fully understand the ramifications of setting their watch could end up with watches which ran very fast or very slow.

Another drawback is that this is not a zero cost to manufacturers. Not only is there some software to develop, but some additional memory within the watch will also be necessary. Memory in these small packages is apparently not cheap. It consumes power and space on the chip, thus adding cost.

Even if this idea is unworkable automatically, I would love to see a digitial watch which allowed me to regulate it by bumping up/down some digital register within the time set mode. This would at least allow one to manually regulate a digital watch. I've never seen such a setting on a digital watch, though I have seen the tiny set screw in the back of the case which is quite near impossible to set with any precision. I would imagine that something like this would be fairly trivial for a company like Casio to add to all their new watches.
-- mgrant, Apr 29 2001

Accuracy vs. Precision http://www.continge...alysis.com/acc2.htm
Description of accuracy versus percision [mgrant, Apr 29 2001, last modified Oct 04 2004]

More on Accuracy vs. Percision http://www.timezone...B/messages/139.html
The Confusing Language of Watch "Accuracy" [mgrant, Apr 29 2001, last modified Oct 04 2004]

Sources of synchronized time http://www.cis.udel...software.html#Wrist
Software to synchronize time [mgrant, Apr 29 2001, last modified Oct 04 2004]

Toasty... http://www.atomictime.com/main4.html
Radio controlled wristwatches, wall clocks, alarm clocks, weather stations, etc. [StarChaser, Apr 29 2001, last modified Oct 04 2004]

RDS is another source of time synchronization http://searchnetwor...7_gci843844,00.html
Would work well in the UK at least. [krelnik, Oct 19 2002, last modified Oct 04 2004]

If you combine the time-signal-controlled watch (using the LW signal transmitted by [among others] the National Physical Laboratory in UK) with a GPS, you've solved this. The time signal tells you when it is, GPS tells you where you are. Casio already sell these separately.
-- angel, Apr 29 2001


I totally agree, adding gps receiver to pick up the signals from a satellite solves the problem, but at great expense in both size and power requirements. When a gps receiver can be built into a watch with neither of these penalties, nearly every digital watch will have one (like an alarm!)
-- mgrant, Apr 29 2001


Just wait for a Bluetooth watch that automatically adjusts itself whenever it nears a PC.
-- Alcin, Apr 29 2001


Good grief, I thought Vernon had returned...

This is baked...watches and clocks that automagically synchronize themselves to the <US> National standard time by radio.
-- StarChaser, Apr 29 2001


There's no need for GPS *and* WWWV receivers; the GPS satellites already transmit an extremely precise time code.

But I don't think you understand, StarChaser; [mgrant] is proposing a technique by which watches could be made more accurate without adding expensive, power-consuming radio receivers. It would be done (almost) "all in software", as it were.

Assuming he's right about the ways you can model sources of error in a quartz crystal oscillator, it should work OK. Perhaps the calibration could be done in the factory beforehand? It is, of course, ridiculous overkill for a wristwatch.
-- egnor, Apr 29 2001


i'm almost sure these watches actually exist already
-- dekoi, Apr 29 2001


Considering that the watches use normal watch batteries and don't have any mention of short life expectancy, I wouldn't call them 'power hungry'. They are more expensive than a normal watch, but that's the price one pays for not having to reset one's watch every two years...Me, I've got a 10$ watch I bought at Sears five years ago that I have yet to change the battery or the minutes on. The hour has to change because of the dumbass daylight 'saving' time, but the minutes continue with no problem.
-- StarChaser, Apr 29 2001


Alcin "Just wait for a Bluetooth watch that automatically adjusts itself whenever it nears a PC." What a bloody nightmare. When was the last time you saw a PC that kept accurate time?

I'm for a simple solution - you don't change the actual accuracy of the watch but are able to specify how many seconds it should add/subtract every day (say at 0300hrs). You'd just use trial and error until it was right just like we had to do with wind up watches.
-- Gordon Comstock, May 01 2001


If you run an ntp (network time protocol) client on your pc say once a day or at startup, you can obtain quite accurate time keeping. There are a bunch of freeware ones out there if you search for them, I have no particular recomendation.
-- mgrant, Jun 16 2001


Sorry to have missed this first time around. Back around 1958 I had a clock in my car that worked this way -- if you set it ahead, the running rate was slightly increased, etc. It converged quite well, except for twice a year when it had to be reset for daylight time.
-- clowenstein, Oct 18 2002


The Coordinated Universal Time, UTC, that designates the atomic time scale only "approximates" the rotational time of the Earth. That sounds like a lot of engineering to execute theoretical perfect UTC time.

The Earth's rotation wobbles significantly for its size for use as a timing device. Somebody somewhere accounts for these wobbles. (I suspect some monk perhaps...)

Still, that being the case, it's plausible that a large earthquake or volcanic eruption will also add its own signature to the rotational wobbles so a person would still have to intervene resetting their watch for those too.

...And If we were to be sideswiped by a large meteoroid impact?

I doubt that I'll pause to check my nano second log...
-- hollajam, Oct 18 2002



random, halfbakery