We've all probably come across these kinds of electronic stopwatches at some point:

They almost always claim to count centiseconds (although this one counts in milliseconds), but do they really have such precision internally? And even if they do, does their display actually refresh to keep up with its counting speed, or does it use something slower, like 60 Hz (or less)?

  • 1
    \$\begingroup\$ Read the manufacturer's documentation and specs. Also note that every manufacturer's models are most likely different from other units. \$\endgroup\$ May 13, 2021 at 2:15
  • 4
    \$\begingroup\$ Most RTCs use 32768 Hz as the internal clock frequency, so the clock resolution will typically be about 30 microseconds. Those cheap passive LCD displays have relatively slow refresh times though, so they're incapable of anything like that update rate. You'd have to look at the spec sheet or take it apart to know what any specific model uses though. \$\endgroup\$ May 13, 2021 at 2:16
  • 2
    \$\begingroup\$ Internally, they are probably using a 32.768kHz oscillator. A counter counts this to keep track of the time. Whether the counter literally keeps track of every pulse, or whether it runs through a divider first before being incremented/recorded into the counter (i.e. a divide ratio of 1024 means it is counting every 1ms.) is up to the designer. I have no idea what the refresh rate is on a reflective monochrome LCD. Presumably it's faster than an LCD monitor, but whether they bother to run it that fast is another matter. If I were them I would run it slooowww. \$\endgroup\$
    – DKNguyen
    May 13, 2021 at 2:17
  • 1
    \$\begingroup\$ @DKNguyen Dividing by 1024 would be only 32Hz. This is kind of an interesting question. If they have an MCU they can take the number and convert it with some math. If it's a simple ASIC it might be easier to use a 32.000kHz crystal. \$\endgroup\$ May 13, 2021 at 2:45
  • 1
    \$\begingroup\$ @jsotola "Time reading" \$\endgroup\$
    – DKNguyen
    May 13, 2021 at 3:08

2 Answers 2


99% of the watches uses a 32768kHz crystal with a suitable divider (i.e. 15 bits for a 1Hz final clock).

How do they do the centisecond? Dithered counters. If you, for example, divide by 2 and on the next cycle divide by 4 the net result is one third of the frequency. With a suitable state machine for sequencing the dividers you can obtain any wanted ratio. Of course duty cycle and jitter are the most horrible things seen on this planet but who cares?

As for the display refresh rate, you have two frequency to handle: the segment clock (needed for the LCD to not burn off) and the commons duty (used to multiplex the 'digits', if there is a common for each digit, for example). The whole frame cycle is a full sequence of common drives. In short, it can be whatever is a compromise between contrast, flicker and power consumption. In ULP microcontrollers (like the MSP430) often you even low the refresh while the display is 'idle' and raise it when there's something interesting to see. I suppose that 32Hz and 64Hz would be popular choices (for obvious reasons).

  • \$\begingroup\$ Excellent answer. This video: youtube.com/watch?v=c7tb1QEdL_Y leads me to believe that 16 Hz is a more common choice than either 32 or 64 \$\endgroup\$
    – ayane_m
    May 13, 2021 at 6:23
  • \$\begingroup\$ Well, 16Hz is slightly below the flicker rate but if the crystal is slow to turn over you wouldn't notice it. I've seen FSTN panels with reaction time of 50ms so it's plausible. The MSP430x4 series guide has all the scary math for driving a glass \$\endgroup\$ May 13, 2021 at 6:25
  • \$\begingroup\$ Many false assumptions and conclusion here \$\endgroup\$ May 13, 2021 at 10:40
  • \$\begingroup\$ This stopwatch does not measure in centiseconds, it measures in milliseconds \$\endgroup\$ May 13, 2021 at 23:25
  • \$\begingroup\$ the refresh rate of the LCD has no contributing error on the accuracy \$\endgroup\$ May 13, 2021 at 23:26
  • This stopwatch does not measure in centiseconds, it measures in milliseconds
  • the refresh rate of the LCD has no contributing error on the accuracy of the time interval counter. This is only limited by eye-on-target to hand coordination, without ever looking at the display until it has stopped.
  • a 1kHz PLL counter clock has a spec accuracy of 1ms in 10hr yet an untrained human error of pushing a button is far greater
  • The 1kHz PLL divided by 3-decade counters can lock onto a 1 Hz clock divided from 32kHz and retain the same accuracy.
  • the LCD display refresh rate has nothing to do with the time interval clock.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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