[time-nuts] Chronometer contest sponsored by IEEE Spectrum
cfharris at erols.com
Sat Dec 1 18:43:57 EST 2007
Hal Murray wrote:
>> The thing is watches rates are dependent on how you personally wear
>> them. Some people's wrists are warmer than others, some people take
>> the watch off at certain times of day, others don't... If you wear
>> the watch normally, and measure your personal offset, you can adjust
>> the crystal to compensate for that change. The watch won't be any
>> more accurate at any given instant, but it will stay spot-on over the
>> course of a week.
> What's the ballpark for a personal offset?
I am more of a mechanical watch guy, and there it depends on the quality
of the watch. Usually less than a couple of minutes per day. After
regulating for personal error, I typically can keep to within a minute
per week on a good watch.
The same is true for quartz. I have had some watches that were nearly
perfect on, or off the wrist, and others that lost many seconds
>> If you don't have a reciprocal counter, such as a 5370B, you can also
>> use a good fractional /N synthesizer, such as the 3666[a,b,c] and an
>> oscilloscope. Put the 3666 on the horizontal axis, and the coil on the
>> vertical axis. Adjust the synthesizer until you get a good stable
> Would it be easier to see any drift if the phase was set to make a line
> rather than a circle?
Personally, I prefer to connect the synthesizer to the trigger, and set
the timebase for 1 cycle of 32KHz, and watch the waveform slip.
> A small shift in a circle just makes a not-round circle. A small shift on a
> line makes a slight opening.
> Or is the drift so huge that this question isn't interesting. (just wait a
> bit and the picture will change)
You are watching cycles slip by, so it moves pretty quickly.
More information about the time-nuts