Checking The Accuracy Of Fake Watches

Since [th3badwolf] realized a wrist watch is the ultimate men’s fashion accessory, he’s been trolling around eBay looking for a nice looking, but still inexpensive wearable chronometer. The Fauxlex brand isn’t normally regarded for accurate time keeping, so he decided measure the accuracy of his off-brand watches in a really clever way.

[th3badwolf] had a camera with a built-in intervalometer lying around and figured if the camera was set to take one picture a minute, the second hand would stay still while the minute and hour hands moved. An hour-long test confirmed his theory and he pointed his cameras towards his knock-off watches.

In the resulting time-lapse video available after the break, [th3badwolf] calculated that the first and third watches lose about 24 seconds a day. He attributes this fact to the watches having the same clockworks. The second watch gains nearly three minutes a day, and he’s trying to send that one back to the supplier. We’re not sure how that will end up, but at least [th3badwolf] has two reasonably accurate watches now.


82 thoughts on “Checking The Accuracy Of Fake Watches

    1. My friend, if I were you, I would not boast about having “the same fake Tag Heuer”. The moto of The Federation of the Swiss Watch Industry is: “Fake watches are for fake people.” Think about these wise words. If you can’t afford a real Tag Heuer, maybe you simply don’t need one. It’s better to have an affordable but REAL watch, that does not pretend to be anything, that is for timekeeping. Would you rather be friends with a real straight up person, of with someone who is a faker? There are plenty of respectable and affordable brands. Some 100 dollar watches are a better time keeper than Tag Heuer (real Tag Heuer), believe it or not. Why bother with fakes? It’s like living a fake life.

      1. ” There are plenty of respectable and affordable brands. Some 100 dollar watches are a better time keeper than Tag Heuer (real Tag Heuer), believe it or not.” I believe it. I also believe that if you are just looking for timing accuracy, $100 is way too much, try $30 maybe.

    1. I had some very poor digital clocks in various devices including watches, some also lost whole minutes in a 24h period, sad but true.
      Poor electronics aren’t better than poor mechanics it seems.

      1. srsly? how you can manage to make your quartz crystal inaccurate?
        hell wikipedia says a typical quartz wristwatch will gain or lose less than a half second per day at body temperature
        and again if it isnt the quartz how you can fck up the design?
        Anyway off to upgrade my arduino mini binary watch and add some thermal compensation lol.

  1. I am having this same problem I bought a Timex at JC Pennies so I am sure its not fake… its been losing time though, at first I thought I was being abducted by aliens but then I slept with another watch in my pocket and verified the watch just sucked…

    1. That happened to me also– you can send it back to Timex under warranty (I assume they just replace the guts with one that works proprerly). A Timex dealer can even give you a shipping box.

      I prefer my Seiko, though.

    2. Your Timex could be creating a localised time dilation field which just happens to exclude the other watch you take to bed.

      I remember the original casiotrons being notorious for this. I received one as a gift when I turned 18 in the 1970s, but the time lost was horrendous.

      Finally upgraded it last year when someone gave me a nice Seiko for my 21st. Best wrist watch I’ve ever owned.

  2. Indeed as west said, best to put them on a controlled heating pad or something,

    Also seeing as how this is a hacking site, id assume most know that if you pop the back off there is a small leaver attached to the spring that will allow one to make it go a bit faster or slower.

    1. If you ask me, it would be foolish to design a watch without at least some kind of adjustment mechanism. I know all too well about the mechanical movements and their assorted mechanisms, but even an electronic version should allow the end user to adjust the accuracy of the timepiece…even if you have a trim capacitor or a solder bridge somewhere in the circuit. Maybe it’s no feasible commercially, but if I were designing it, that would be a must-have feature.

  3. It’s not the accuracy that makes a chronometer such as the Rolex valuable. It is the predictability. It is okay for your timepiece to be off by a certain amount each day if you are using it to find longitude. But you need to know HOW MUCH (and fast or slow) it is going to be off between readings.

    About 20 years ago (IIRC, on sci.electronics newsgroup) there was a group of posters comparing the accuracy of their digital watches and even tweaking the oscillator to fine tune them. One guy even wore his to bed and let his body temperature act as a “crystal oven” to regulate the watch.

  4. so, what needs to be designed is an overnight ‘Watch Temperature Regulator’ aka timed heater/cooler that will compensate for the deficiencies inherent to your specific watch, keeping it on time overall.

  5. This is making me crazy. Remember Engineering Physics I? Accuracy and precision are not the same. While clever this experiment is inconclusive without knowing both the accuracy and precision of at least one of the time pieces or the intervalometer.

  6. My son brought a bunch of these home from Iraq. He gave his sister a “diamond”-studded Faux-lex. She loved it. One by one, the “diamonds” started falling out, an then one day, she took it off to take a shower and came out of the bathroom yelling: “my watch just exploded!”

    Sure enough, there were pieces all over the floor!
    They don’t build those things to last, apparently, but at the price, you can always buy a replacement!

  7. Speaking of the accuracy of cheap watches.

    There is an excellent (long) thread over at watchuseek relating to the ezChronos watch from TI. They managed to get it accurate within 2 or 3 seconds per year. Not too bad for a cheap watch. :)
    Temperature calibration is done over a 15 day period, 5 in hot, 5 in cold, 5 in room temp. Then adjusted in the settings of the watch.

      1. I was jaded too, after two failed marriages that I felt were mostly my fault. It turned out I hadn’t met the right woman yet, and married for the sake of marriage rather than love and companionship. It wasn’t a matter of fault, but of simple incompatibility.

        Now, I am with a woman who I fit with perfectly, like two puzzle pieces. We don’t have everything in common, but we complement each other perfectly and after 18 months of dating have yet to fight, argue or even really disagree. I’m truly happy with a relationship for the first time.

        So, don’t give up! The right one is out there.

        Regarding watch accuracy, the only watch I’ve ever owned that was accurate to the point that I only lost a few seconds a year was a Caravelle by Bulova. It’s their cheap line too. And, I’ve had both cheap and expensive digital watches lose/gain several seconds per week. The most accurate digital I’ve ever owned was a Casio DataBank back in the late 90s.

  8. My girlfriend sent me a fauxlex from some Chinese backwoods (only a few million people) factory town. At one point I had a Rolex dealer look at it just for grins (while I was getting a crystal replaced on another watch). He ended up having to take the back off to find any sign that it wasn’t genuine, which surprised the hell out of him.
    When I told him that it kept really accurate time, he said “well, there’s your first sign that it’s not a real Rolex.”

  9. This is a great subject to segue into discussions about calibrating your calibration devices. If you wanted to pursue this method of calibrating a watch, how would you build an intervalometer precise enough to do it?

    1. Easily. Set the watch accurately using a computer synchronized with NTP. Wait however long you feel like (24h is probably long enough). Compare watch to computer synchronized with NTP. Note how fast/slow the watch is. No need for an intervalometer.

      If you actually want to build a very accurate intervalometer I’d probably do it using a GPS’ PPS output and TTL logic or a microcontroller. Could also use an OCXO or some other frequency standard.

    2. There have long existed commercial watch timer machines for mechanical watches. You clamp the watch to a special sensor that uses a microphone to listen to the ticking. They can tell you the watch’s instantaneous rate, the average rate over some interval (e.g. the last 5 minutes), the standard deviation of the rate, and a bunch of more technical things like the swing angle of the watch’s balance wheels (which it measures from the sound!). Watchmakers use these machines to measure the adjustment and regulation of watches.

      There’s a $9 iOS app called Kello that implements a simple version of this, but it’s quite finicky.

    1. For a mechanical watch (especially an automatic) that much variance isn’t out of the question. If you own an automatic, you own it because you like the idea of it, not because you desire the most consistent timepiece.

    2. Mechanical watch movements that meet the requirements and testing of an Officially Certified Swiss Chronometer can be something like +6/-4 second per day, and that’s in a watch costing many thousands of dollars.

      Yes 24 seconds in a mechanical movement is quite believable. We have been spoilt last century with the invention of the quartz movement.

      By the way I saw the first Quartz clock in a museum. It was the size of a fridge that you’d see in a family of 10.

    3. Mechanical systems are very difficult to tune to that level of accuracy. I have a wind-up clock that I can tune using the length of the pendulum. The best I can achieve is about 3 minutes per month of loss or gain. Due to the fact that it operates on a spring, it’s gaining time for the first half of the month and losing time for the second half of the month. Getting your accuracy goal per year is impossible.

    4. 24 seconds is perfectly believable for an affordable mechanical watch. I have mechanical watches that vary from +40ish seconds/day to +1.2 seconds/day in use.

      Note, however, that these numbers are tricky to interpret, because of several reasons. First, the rate of a mechanical watch varies with position, and the video shown here only tests each watch in one position, and not the same position for all three; #1 is crown down while #2 and #3 are crown up.

      If you’re actually wearing a watch, you’re moving your hand, so the rate error you’d get is a complex, semi-random weighed average of the errors for each position. The most common positions, however, are dial up (watch set down, sitting, reading the watch), crown down (walking) and crown right (sitting), so the crown up position tested for #2 and #3 is one of the least representative of actual usage.

  10. I was curious what the net effect of a timepiece’s drift would be, so I lazily attacked the math for it.
    Losing one second per hour, or 24 seconds per day doesn’t sound like a lot, but in fact works out to 11.5 minutes over a 4-week period, and amounts to almost two and a half hours over the course of a year.
    Perfectly acceptable; as long as you remember to fix the time every few months.

  11. Mechanical watches have an adjustment screw in them to correct for errors, although of course you still have to have good frictionless bearings to make them really accurate, and that’s where the term x amount of ‘jewel’ comes from since if you make the bearings from a precious stone it has the lowest friction they found out long ago, and that still remains the optimal way for mechanical I hear.

    And it’s odd since for digital you use quartz, so it seems it’s the stones that help us keep time :)

    Best is using radio-controlled or GPS ones though.

  12. The issue with Rolex needs to be clarified a little. A Rolex is actually a very accurate and precise timepiece, BUT it is also an automatic (offset weight winds it from your movement) so it needs to be tuned to the person wearing it. The escapement attempts to correct and equalize the spring tension into the movement, Bhutan the amount of tension in the spring will still affect the speed at which it runs a small amount. A very active person may, for example take it off at night with 100% charge, where a sedentary person may only have a 50%. Both watches will continue to run trough the night, but at slslghtly different rates

  13. Its a simple “Garbage in, garbage out” scenario. If it comes tuned for 60% and you run 65% it’ll be a little off, and you will want to meet your local watchmaker.
    That being said, my usual lead in to that is “That’s because if you own a Rolex, you aren’t so worried about what time it is.”

      1. Ummm… they are using cheap Chinese movements made from cheap materials by people who make like a dollar a day. (If you could have slaves you could make’em even cheaper, but oh well.) Is that what you’re admiring? The gap between the 1st world and 3rd world?

        Real fine Swiss timepieces are assembled by hand, by people do just that – make the finest watches in the world. And they have been doing it for centuries. Not to mention that some of the historical models were a break-through in Mechanics and Engineering at the time.

        I think you should change your major to Business.

  14. That’s a nice, simple way to check the accuracy of many watches at once, guess I’ll try that one day.
    But how about using a stethoscope-attached microphone to know the accuracy in just few seconds? Just lay your watch on it and check if the tick frequency is fine. With 44kHz soundcard sample rate, you should be able to get the readings accurate up to 2s/day. Or you could get something like NI myDAQ for 200kS/s, that would give less than 0.4 s/day accuracy, all within just a few seconds.
    Downside: you need to know the proper escapement frequency first and it may be pretty hard for those chinese watches :(

  15. Why would anybody buy such cheap sh1t? Either afford a real one or settle for a good looking, moderately priced watch that doesn’t attempt to be something else. Like the Traser Watches or so…

  16. my rolex loses about 1 second a day. my previous one gained about 2.

    the ‘about’ is because i’ve not measured it, but do tend to set the time accurately at the end of any month that contains less than 31 days.

  17. Android phone and an application from the market.

    Set the timer for 1 minute. Put the watch into focus, and hit the record button. The application will also output the resulting time lapse as flv, mp4, and some other options.

    The free versions of the application should do well for this. Limitations include only being able to record in 240p… but, I don’t see that as a problem really.

    Actually spent the last 6 hours testing this with a digital laying next to my trusty Seiko chronograph. The Seiko lost 1 second every 2 hours as compared to the Timex Ironman digital that didn’t show any change over the 6 hour long recording.

    the loss of 12 seconds per day seems like a lot to me… but I am told that is actually quite good for a mechanical automatic watch that is 6 years old.

  18. Mechanical watches are adjustable, and with time and a little record keeping, you can get them really close.

    My x (was my wife at the time) gave me a pretty cool, $3 mechanical watch. Kept lousy time. So I popped the back off and quickly realized this was not meant to be a $3 watch.

    Inside, you’ll find a balance wheel. You’ll see it rocking back and forth with each tick. Right there, is a tiny lever, often marked “Fast” and “Slow.” Often there are markings. Move this lever just a tiny bit, and notice in a couple days if your watch is fast or slow. Then move it just a tiny bit again, and notice in a couple days if it’s fast or slow. You’re hunting for the right spot. It usually takes just a few tries to get to within a few seconds a day.

    Careful! This balance wheel has a very tiny, delicate spring, and you’re adjusting the tension of this spring. If your tool slips and goes inside the balance wheel, you’ll break your spring and your watch will stop. Permanently.

    I gave the $3 watch away to a kid who thought it was cool. Mom has a photo of me as a toddler holding Dad’s wrist to my ear to hear the watch tick. It’s a strangely soothing sound.

  19. The timebase in the camera is IMO not accurate enough to use as a callibration “source” (like in this hack), he should have used some kind of dcf77 (signal based on atomic clock, sent from frankfurt in germany) based timing generator…

  20. The best way to keep a crystal oscillator stable is to keep it in a controlled temperature. The human body is actually quite good for that. So, assuming they are also calibrated at body temperature, many watches will be potentially far more accurate worn 24/7 than sitting on a shelf or even in a drawer. Otherwise, expect the frequency to drift quite a bit as they warm and cool and the crystals age.

  21. As the [revious poster said by far the most accurate timepiece available to any of us is the rising edge of the 1PPS signal available in almost any GPS module. The reason the 1PPS signal is important because otherwise, the GPS NMEA output is prone to varying by as much as half a second (worst case scenario) Most GPS modules with 1PPS output can be used to synchronize a NTP server and that server can deliver extremely accurate time to an entire organization. I am sure that very soon we will start seeing watches that automatically set themselves by means of an internal GPS standard. The price is low enough.

  22. Most mechanical watches drift. Big deal. They’re fashion accessories, not be be confused with timepieces used to evaluate world records. If his fauxlex drifted 24 seconds that’s pretty good. My Seiko drifts about 20-25 seconds per day according to the manual, but again who cares, it will wind down anyway if I don’t wear it and I’ll have to readjust.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.