2
\$\begingroup\$

Building a timegrapher is quite straightforward, it is nothing more than a piezo transducer connected through to an amplifier. However, in order to actually utilize the results, one need an accurate clock to benchmark the results against.

How should I go about acquiring this signal? The end result should, I suppose, be a input signal that I can feed into the program I use to analyze the sounds from my watch.

I understand that there are USB-powered GPS units that generate a very precise 1 pulse per second (PPS) output. But how do I feed this into my software?

What other (preferably low-cost) options are there?

  • Something built using an NTP-server? But will there be problems with lag?
  • Building a radio receiver that uses a time signal. I have understood that this can be a bit messy (interference, etc).
\$\endgroup\$
  • \$\begingroup\$ Are we talking about a clock that gives actual time (hrs:mm:sec) or a clock that produces precise pulses? Either way there are plenty of real time clock modules for sale on ebay and lots of designs for crystal controlled precision oscillators. \$\endgroup\$ – JIm Dearden Jan 15 '15 at 13:18
  • \$\begingroup\$ @JImDearden: He's talking about a test instrument that's used to calibrate wristwatches. \$\endgroup\$ – Dave Tweed Jan 15 '15 at 13:40
  • \$\begingroup\$ @JImDearden: I only need the pulses, the actual hrs:mm:sec are irrelevant for this use \$\endgroup\$ – Figaro Jan 15 '15 at 14:08
  • 1
    \$\begingroup\$ @Figaro Thanks for that info - you might be interested in this hack of a quartz clock mechanism to produce an accurate 1 second signal pulse. josepino.com/electronics/one_second_timebase \$\endgroup\$ – JIm Dearden Jan 15 '15 at 21:19
  • 1
    \$\begingroup\$ @Figaro It could be in that area as it uses a 32kHz crystal but I couldn't guarantee it, check out the texas instrument notes ti.com/lit/an/slaa322b/slaa322b.pdf they seem to indicate 5 - 30 ppm \$\endgroup\$ – JIm Dearden Jan 16 '15 at 10:47
2
\$\begingroup\$

Keeping in mind that Windows (and other desktop OSs) is not even close to a real-time operating system, your best bet would be to get an external GPS receiver and feed its 1 pps output into the other channel of a stereo audio input on your computer. Then, your software could correlate the information between the two channels in order to determine the frequency error of the watch you are testing.

|improve this answer|||||
\$\endgroup\$
  • \$\begingroup\$ I'm probably using the wrong search terms, but what are my actual hardware options for such a 1PPS device? I'm unable to source anything useful \$\endgroup\$ – Figaro Jan 15 '15 at 14:13
  • \$\begingroup\$ I've mostly worked with OEM GPS receiver modules that I've built into other equipment, rather than complete packaged receivers. The search phrase "gps module 1pps output" seems to give a large number of hits for such modules. You will need to add an external power supply, antenna and output wiring. \$\endgroup\$ – Dave Tweed Jan 15 '15 at 14:20
  • \$\begingroup\$ @Figaro A shortwave receiver tuned to WWV or similar time broadcast station could also be used as a reference. \$\endgroup\$ – Phil Frost Jan 15 '15 at 15:03
  • \$\begingroup\$ @PhilFrost: That isn't a effective as you might think for short-term accuracy. Sure, it's ultimately locked to an atomic standard, but the radio path delay can vary over the short term by 10s of milliseconds, requiring long integration times to get high accuracy. \$\endgroup\$ – Dave Tweed Jan 15 '15 at 15:47
  • 1
    \$\begingroup\$ 10 ms variation, measured over 100 seconds gives you 100 ppm accuracy. However, we're trying to calibrate a wristwatch to on the order of 1 second per day, which is just ~12 ppm. \$\endgroup\$ – Dave Tweed Jan 15 '15 at 16:48
1
\$\begingroup\$

Just to add to my comments above here's a couple of typical xtal clock circuits using common CMOS dividers.

enter image description here

The 4060 doesn't quite have enough divider stages to produce a 1Hz signal from a 32.768 kHz Xtal so an extra divider (4013) is required. The 4521 has lots of stages but not all are available so a higher frequency Xtal is used (4.194304 MHz).

Alternatively you could try using a simple xtal clock controlled micro controller (sat ATtiny for example running from a 32KHz xtal) or get something like an arduino nano (16MHz clock) ready made.

|improve this answer|||||
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.