Methods of Timing

I use two methods for timing my tuning fork and quartz watches, though I am considering a third. For a thorough description of various methods, check out this thread on WUS. For my rambling musings on the wider question of timing watches, follow this link.

My main method is the stopwatch method. That is to say, I take a watch and compare the difference in time between it and a time reference using a stopwatch. I then track the change in this difference over time and plot the results into an Excel spreadsheet.

I record the five most consistent readings taken against time.is and the five most consistent readings taken against time.gov. This helps to iron out the small errors that each reference source has, as well as the errors in my use of the stopwatch. I end up with an average of ten readings that are not accurate enough on their own to tell me the SPY value of the watch. Over the course of a number of days and weeks, however, the average smooths out and a fairly reliable SPY value can be determined.

The ultimate test of SPY, however, is a full one-year test. To measure the off-set from UST at one moment in time, and then again 365 days later.

The progress each watch makes during the year, however, is interesting and instructive (particularly as seasonal changes can speed up or slow down a movement).

While there seems to be a fair scope for error in the stopwatch method, real world results are accurate. After all, the SPY value cannot be more or less than the change in measured off-set from UST over time. So if my watch was 0.01 seconds off on the first of January, and 1.01 seconds off on the first of July, then whether it's exactly right to five or six decimal places doesn't really matter. In six months is has drifted by pretty much a second and I could project that it is likely to be around two seconds off on the year.

An evolution of this approach, for dealing with large numbers of watches, would be to arrange the watches around the computer monitor and video them against a display of a reference source (e.g. time.is). You could then go through the video in slower time and quite quickly and accurately determine the off-set of each watch. This is a set-up that I am seriously considering, as many of my watches simply don't get stopwatched as often as they should.

My other method of timing is based on measuring precise performance at a given temperature.

I use an ex-lab Torrey Pines Echotherm chiller/heater to set a watch to a particular temperature, and then use a MicroSet timer to measure the drift in off-set between the watch's stepper motor pulse and a GPS reference signal over a fairly lengthy period of time. I am not an expert in this sort of technology, and am grateful for the advice and wisdom of the members of the WUS community. A thread discussing my experiments can be found here.

Here, a Certina DS-2 Chronograph is being timed at -3C.

And here is the MicroSet timer:

I use this set-up to determine the performance of a movement across a range of temperatures, but it should be noted that this does not directly correlate to a watch's 'real world' performance.

For my mechanical watches I have tried using the stopwatch method but I find that the 'Watch Check' app. from the Android store is easier. This app. is not accurate enough to help with HAQ timings, but it's great for mechincals. It takes its reference time from an NTP source and requires the user to tap a button when the watch reaches a particular time. A watch can be tracked over any period of time, as long as it doesn't stop.