Highly accurate timepieces have intrigued me even since I was a young child - I have no idea why - just one of those fascinations...!

One common approach to keeping a timepiece accurate is to have it automatically check itself against a reference clock located in Boulder Colorado, by picking up low frequency radio waves which are broadcast from that facility (station WWVB).  These "atomic" clocks and watches incorporate a very sensitive radio receiver that (in North America) picks up the WWVB time signal each night and uses it to set the clock or watch to the correct time - to within a small fraction of a second. (Note that there are different transmitters, designed to different specifications, that provide coverage in Europe and Japan.)

The only claim these timepieces have to the word "atomic" is the fact that the reference clock in Colorado bases its timekeeping on an extremely precise oscillation within Cesium atoms.  There's nothing "atomic" about the clocks and watches themselves; rather, they are a combination of radio receiver technology and the everyday "quartz" oscillator timekeeping technology.

During the stretch of hours (or days - if reception conditions are poor) when the radio-controlled clock doesn't pick up the signal, the timekeeping accuracy is that of an ordinary "quartz" clock, so it may drift a bit and be a few seconds fast or slow by the time the signal is received again.

However, some radio-controlled clocks (not all models, just some models), use the radio time signals not just to set the clock to the right time, but also to adjust the clock's rate. For example, if the clock set itself to the right time at 1am on Monday, and found that it was 1 second fast at 1am on Tuesday, not only would the clock set itself to the correct time at 1am on Tuesday, but it would also slow its own rate down just a tiny bit, so that when 1am on Wednesday rolls around, the clock might be only a small fraction of a second off!  Once this kind of a clock has had a chance to regulate itself in this way, it might go quite a few days without accumulating even one second of error, even without receiving the radio signal!  Note that the clock should be in a place where the temperature is quite stable in order to get the best results.

There are some variations in the actual technique used for doing the regulation.  Some timepieces (e.g. the Arcron SynTime line) measure the error that accumulates from one day to the next, and then compenstate.  Others (e.g. some Casio models) seem to do the calibration by measuring a precise time interval during a single reception of the signal (which can last several minutes).  In either case, the timepiece ends up adjusting its own rate such that any remaining error in the rate is probably on the same order of magnitude as what might be caused by temperature variations (particularly for watches, which go through significant changes in temperature).

Note that this self-regulating feature doesn't affect the frequency of the quartz oscillator.  Instead, these models use a digital technique to compensate for the oscillator's error.  Typically, the oscillator is deliberately set to run a little bit too fast.  In addition, once per minute or so (I'm not sure of the exact interval), the timepiece drops a few of the pulses coming out of the oscillator.  The number of pulses that get dropped can be adjusted to fine-tune the overall rate of the timepiece.  The long-term effect of this is to slow the timepiece down just slightly so that it's timekeeping becomes very accurate.  (By the way, this means that once per minute the timepiece will indicate a second that's just a bit longer than the other 59 seconds, but the difference is too small to be noticeable in normal use.)

To me, this self-regulating approach seems like the "right" way to design a radio-controlled timepiece - it's a nice touch, even if I can't come up with a single reason why I would need it :-) .

So, in case others are interested too, I thought I'd start a list of radio-controlled clocks and watches that have this feature.  For some reason, the clock and watch manufacturers usually don't even mention in their ad copy whether or not a given timepiece has this feature.  Even the models that include it often don't have it mentioned in their advertisements.  However, it is sometimes mentioned in a user's manual.

---

I'm sure that the list below is incomplete!

If you know of a radio-controlled timepiece that fine-tunes its own rate and isn't listed below, please feel free to contact me so I can add it to the list (and please include in the email the reason why you believe the self-regulating feature is present - this might be a description of an experiment you tried, or a quote from the user manual or ad copy, etc.)  Thanks in advance to contributors to the list!


Arcron "SyncTime" watches and clocks (watches also sold as the "Maximilian" brand)
Casio Waveceptor watches (the following models):
    WV55HA-1AV digital display, using module 2583.
     WV56HA-1AV digital display, using module 2587.
     WV50H-1AV     digital display, using module 2406.

Oregon Scientific desktop clock model RM116E     (Thanks go to a visitor for mentioning this model. My tests confirmed his results.)

Radio Shack desktop clock 63-968 (a nice clock, but I've found the battery life to be short)

 

---

I have not tested the following models myself, however, a visitor told me that these also are self-regulating:

Arcron 4320M-5

Radio Shack 63-970

Radio Shack 63-970A   (may have fewer errors decoding weak signals than the 63-970)

Timex (ACCL) 3534T

 

---

A more detailed perspective

A few visitors to this page have emailed me with comments indicating that they have checked out the accuracy of their own clocks rather precisely, so I thought I would add a few comments on factors that can affect the accuracy of self-regulating timepieces.

First, though, here's a description of the experiment I've used to detect and observe the self-regulating feature in action:

• Install the clock's batteries
  (remove the batteries if necessary so that you can perform this step - it resets the clock to its natural (unregulated) rate)
• Before it can receive the radio signal, turn off its reception of the signal
• Manually set the clock to the correct time, and carefully note any initial error in the clocks readout
  (it might be a fraction of a second fast or slow due to manually setting the time)
• Let the clock run for several days or however long it takes to accumulate a large enough error that you can measure the error accurately (e.g. several seconds of error).
• Note how much time has passed since the clock was set, and how much error has accumulated.
  From this, you can calculate the clock's natural rate
• Enable reception of the time signal
• Wait until the clock has received the signal several times (this may be a few days, or longer if reception is poor)
  By the time this step is completed, the clock presumably has had a chance to regulate itself.
• Note the date, time, and current amount of error in the clock's readout
• Disable reception of the signal
• Wait long enough for a significant error to accumulate (at least a second or two)
  (this may be a long wait if the self-regulation worked really well!)
• Note once again the date, time, and amount of error accumulated.
  By comparing this checkpoint with the previous checkpoint, you can measure the clock's regulated rate.
• Enable signal reception, for normal use

Once the clock (or watch) has regulated itself (i.e. adjusted its rate as well as set itself to the correct time), the clock's rate will probably be much more accurate than its natural rate, however, it still may not be perfect, due to several factors that can affect the performance of the self-regulation.

First of all, the clock's adjustments of its rate occur in discrete steps.  Hypothetically, if we assume the oscillator frequency is 32,768 cycles per second, and if regulating adjustments are performed once per minute (by dropping a certain number of these cycles), that would mean that once each minute the clock's timekeeping could be adjusted in increments of 1/32768 of a second.  Since this adjustment gets repeated 1440 times per day (the number of minutes in a day), that means that the adjustment increment on a daily basis is about .044 seconds, or about 1.32 seconds on a monthly basis.  Since the error should not need to be more than half of the adjustment increment size (i.e. rounding to the nearest increment), the self-regulation should be able to achieve accuracy to within .66 seconds per month, all other factors being steady (such as battery voltage and temperature and aging of the crystal).

If you're 'lucky', the accuracy might be significantly better.  I.e. if the clock's natural rate just happens to produce a monthly error that's an exact multiple of the adjustment increment size (on a monthly basis), then the self-regulation could, in theory, compensate perfectly and produce exact timekeeping.  On the other hand, the worst case should be a monthly error 1/2 the size of the adjustment increment size (on a monthly basis).  The clock's temperature is probably the biggest variable in determining whether the regulation algorithm will be lucky or unlucky, since this has the largest effect on the oscillator's frequency.

As you can see, the oscillator frequency, the time interval between rate-trimming adjustments, and the clock's operating temperature are all key parameters affecting self-regulation.  So, if you have one of these clocks, the way to get the best performance out of it might be to try different locations (with different temperatures) to see if you find one where the numbers work out just right and the compensation ends up spot-on (until the battery runs down a bit or the seasons change or ...!).

I don't know what parameters the designers of these clocks actually use - the above is just a hypothetical example - 'your mileage may vary'.


Home