How a quartz watch works

Originally published at:


Showing the cycle of flip-flops with actual flip-flops was inspired :slight_smile:


That’s quite a good description of digital timekeeping.

It is amusing to me that you can spend tens of thousands of dollars a a mechanical watch that’s not as accurate as a $10 quartz watch. But it does look fancy.

Also, quartz watches can be easily made to be good to a minute a year (a second per week), by adding an adjustable capacitor to tweak the resonant frequency of the crystal. Then you are limited by temperature variation of the resonant frequency, which is dealt with by putting your wrist in a temperature-controlled chamber.


Mechanical watches are basically Rube Goldberg machines that tell the time. I say that as someone who has owned some pretty expensive automatics and hopes to get another later this year…


That was a particularly well organized, hence informative, science video - thank you!
(also love to know the brand of the USB oscilloscope he used…)


The good thing is that your wrist is already temperature controlled.


My Nixie tube watches tell time by running a microcontroller at 32.768 kHz, then counting timer overflows to keep track of the seconds, minutes etc.

I wonder how many flip-flops you’d have to hang on a chain to model that?



What an amazingly charismatic youtuber!
He could tell me I had terminal cancer and I would just cup my chin with my hand, lean forward and go “really? how interesting. please tell me more”


By the time the signal goes through the 14th flip-flop, the frequency is one cycle per second.

That should be 2 cps. 2 to the 15th power divided by 2 fourteen times results in 2, not 1.


Indeed sir!

1 tick
1 tock

As it should be.

1 Like

1 Like

Agreed, and I like how he got right down too business rather than cut away to a “channel intro.”

Yes, you are correct.

While the school I went to for watchmaking is closed now (as of Nov. last year), we did have a semester of quartz watch service that included coverage of history of quartz watches, bulova accutrons that used tuning forks to keep time, and all kinds of actual circuitry diagnosis and service (even thought I suck with electronics, I did well).

What you are referring to our professors referred to as a trim pot. Only older watches seem to have them, when the circuitry was considered a significant advancement so they wanted to be able to fine-tune it for the exact reason that you specified.

Most modern quartz watches do not have a trim pot, and they are essentially forced to keep time at whatever the reference crystal they are given has, within a certain range. There are high-end quartz watches from companies like Patek Philippe that I believe have not only bespoke fancy circuit boards made for them but probably have trim pots because they are designed to be very finely adjusted as befits their pedigree.

Fun fact that you and probably anyone else who works with advanced oscillators in circuitry would know, but most people do not- a quartz watch will always inevitably gain time the older it is because the canister holding the crystal is sealed, and purged with inert nitrogen or xenon gas inside, but over time the mechanical seal breaks down at a very fine level and that lack of vacuum slowly influences the oscillation frequency of the quartz crystal. This causes it to slowly gain time, and run fast.

It will happen in any time keeping device that uses a quartz based oscillator, and is unavoidable. Usually after 10 years or so it starts to be really noticable, but it can happen earlier or slightly later, but it will happen. Once it does there is no way to ever make the watch keep proper time again because it is constantly gaining beyond a reference frequency. The only choice you would have is to replace the entire circuit board or if you know what you are doing a reference oscillator canister if you are extremely skilled at micro soldering and you really know what you’re doing.

This kind of work is beyond the skill of almost all watchmakers to my knowledge but there are some who were specifically trained to do such corrective actions. They are called CEW, Certified Electrical Watchmaker, and since it was certification that only lasted for a few years about 40 years ago, they are quite rare and only a handful are still alive.

I myself am still an AWCI member (American Watchmakers-Clockmaker’s Institute) but I work as a tool and die machinist, so that is how I know these things…


Or if you’re more formal, it like to be addressed as trimmer potentiometer.


I strongly doubt that this is relevant to more than a handful of exotic specialty watches(both because it’s expensive and because it’s relatively bulky) but in the more general genre of quartz oscillator use there are the interesting devices where the design essentially abandons any hope of long-term stability from quartz immediately and focuses instead on adjustment and compensation based on a more stable input with the crystal kept around for its (usually excellent unless ‘aging’ is turning into ‘failure’) short term stability and low power consumption.

The most common flavors seem to be GPS-disciplined, if you expect to see the sky at least reasonably frequently; and rubidium-disciplined, if that can’t be assumed. Combining the second reference into the package makes both flavors a bit big for a watch; but in principle I’d imagine that you could borrow the techniques used in the disciplined oscillators to construct a watch that would just need to be periodically attached to an external frequency reference to recalibrate itself and would ordinarily consume no more space, and only negligibly more power, than one of the relatively naive models that is built either around the hope that the oscillator performs as it says on the tin or a one-time calibration at the factory.

1 Like

This may help some too- AWCI powerpoint from Witschi showing basics of how a quartz watch works.

Link is safe and from a professional organization of watchmakers.

There are also interesting things that happen with the lavet motor that runs the watch called asservicement and inhibition. Too complex to explain here, but the last second of each minute on a watch second hand often does not tick in exactly 1 second- it makes up the fast or slow remainder of the other 59.

1 Like

That’s a great idea - the microcontroller current consumption becomes really low at that frequency. For ATMega8A it’s about 50-70 uA. Compare this to to 11 mA at 16 MHz :slight_smile:

There’s also now a segment of upcoming computer scientists who think flip-flops are named after footwear.


It’s actually a trimmer capacitor. A standard tuning fork crystal uses two capacitors to match its internal capacitance. The one on the input to the inverter has more control over the resonant frequency.

An interesting aside is that when Dallas Semiconductor designed the DS1307 real-time clock chip, they decided to help the customer save a couple cents by building the two capacitors into the chip. A byproduct of this is that it’s impossible to trim the frequency of the oscillator, so you’ll never find a computer clock that keeps good time with the power off. They’re really bad, like a second a day.