We’ve got to admit to being a bit of a Casio G-Shock watch geek. The big, chunky watches were every day carry items that survived everything we dished out, right up until the smartphone made wearing one seem redundant. But others continue to use and abuse G-Shocks, and some brave souls even hack them.
Replacing the standard quartz crystal with a temperature-compensated MEMS oscillator is one hack that [Alex] tried, and it appears to have worked out well. His project write-up doesn’t specify which MEMS oscillator was used, but we suspect it’s the SiT1552 TCXO. With its extremely small size, stability over a wide range of temperatures, and ultra-low power requirements, the chip is a natural choice to upgrade the stock 32.768-kHz quartz crystal of the watch. Trouble is, the tiny 1.5 mm x 0.8 mm chip-scale package (CSP) device presented some handling problems. After overcooking a few chips in the reflow oven, [Alex] was able to get one mounted to a tiny breakout board, which went into the space formerly occupied by the watch’s quartz crystal. He stole power for the TCXO from a decoupling capacitor, sealed the watch back up, and it’s back in service with better stability and longer battery life to boot. The video below shows the TCXO undergoing tests alongside the original quartz crystal and a comparatively huge DS3231 RTC module, just for fun.
[Alex]’s MEMS transplant seems a long way to go and a lot of fussy work for marginal gains, but who are we to judge? And it does make the watch susceptible to punking with a little helium, which might make things interesting.
Maybe I’m stupid, but I do wonder what he tries to gain from the transplant. In order to meet the accuracies one sees at a every-day quartz watch, the processor would have to apply a temperature compensation scheme anyway. This will interfere with the MEMS that does not require it and will make the precision of the watch worse, not better.
Of course, being able to even perform such a job is no small feat, but apart from that I cannot see much of a benefit.
If I’m mistaken, please set me right, I love to learn.
There’s no temperature compensation in a G-shock. (There are “Atomic-Time” models though!)
I guess it is the age old reason, just for fun and because he can.
other points
– just because he can
– learn something new (knowledge hoarding :P)
– extend the lifespan of the watch by few more years and less batteries
– gain geek points with girlfriend ;) (if applicable)
Also i think that, since this is a watch, the temperature it is exposed should be fairly constant. Provided the wearer does not have fever or does not wear it continuously.
Oh well…..
The temperature that actually matters is the *external* temperature, i.e. environment. It makes one heck of a difference whether you are indoors in 22 degrees C or outdoors where there is -2 degrees C, for ex!
The temperature that ACTUALLY matters is the oscillator temperature. That’s why lab grade time references have small heaters for crystals which keep them at stable temperature. Environment temperature does affect oscillator temperature, but it’s a complicated effect and what really matters is oscillator temperature.
I had a timex with a thermometer in it back in grade school (and a compass on the band)
On that watch the temperature gauge was on the front of a fairly bulky watch…
Yet it never read correctly when it was on my wrist… it’d read high when it was cold, and low when it was hot, always pulled towards the temperature of my skin. Most of the time it read in the high 70’s at room temperature and because of coats and such would often read higher when I was outside in freezing temperatures.
I’d imagine that a oscillator inside a watch, likely in the back near the metal casing touching the skin would fairly stable compared to the environment outside.
The point is the watch has a large conductive metal plate permanently in contact with a 37C heat source on one side, a layered plastic and air structure on the other side and in low temperature situations tends to be covered by a supplementary layer of fabric insulation which varies according to the ambient temperature. It’s internal temperature range shouldn’t be that extreme.
Wristwatches are already strapped to a natural thermostat at 37 degrees Celsius, so TC is mostly not necessary.
Interesting! Please post your test results. Mine are different
Three questions.
1) Do you sleep with your watch on?
2) Do you expect to have to correct the time every morning?
3) Do you keep your home at body temperature?
2. Yess godammit! I’m sick of my G-shock being 0.0502 seconds further off every morning. After 20 days it’s perceptibly different to my atomic clock. I might scrap my body temperature bedside drawer project and build this.
Actually on second thought the drawer project can keep some other stuff body temperature as well.
Interesting case study: does wearing the watch on your wrist versus keeping it in a box make any noticeable difference to accuracy? Your body heat would raise the internal temperature somewhat, but would it be enough to make a difference?
Back in December, with gift cards in hand and Chridtmas sales, I thought about buying a G-Shock. But I’ve been spoiled. About 2006, I got a new Casio Waveceptor for $20, and I’ve never had to set the time. Occassionally it doesn’t sync up overnight, but one night is not enough for the time to change noticeably. And I’ve never changed the battery, this one has sone level of solar charging that keeps the battery running.
So I could spend a lot more and get s nice bright red watch, but it would never be as good as the Waveceptor.
Michael
Here’s a question: Why do atomic-sync watches always seem to go the direct route through WWV? I mean, if you’re going to go to the problem of putting a radio in the watch to receive potentially weak HF signals, why not count on the far more pervasive coverage of WiFi and just sync up to an NTP server over the web? You wouldn’t have to turn the WiFi radio on but once a day to keep the clock in sync, so you wouldn’t be killing battery life.
If I ran the world, every digital clock from the microwave to the nightstand to the car dashboard would have WiFi and an NTP client to keep them all in sync. Nothing cheeses me off more than a microwave clock that’s a minute out of sync with the coffee maker clock – it’s like I have three time zones in my kitchen. Unacceptable!
I think it’s because wifi would require configuration. Entering a wifi ssid and password would be painful on a typicical watch interface. Wifi also requires transmitting, which increases the power budget.
I’ve often thought it would be nice to have a little NTP-synced box that would transmit a time signal in WWVB format at micro power, for use in places where WWVB can’t be reliably received. There have been a few such devices described here on HaD.
As an aside, all radio controlled watches I’m aware of receive WWVB on LF, not WWV on HF.
WiFi is pretty power hungry. BTLE would be better, but you’d still have to setup some way for the watch to get the time. WWV is pervasive, and apparently a receiver doesn’t take much current after all..
Now, if you wanted to standardize on something, you could develop a standard BTLE beacon format for time syncing. You could sell one device that syncs to time via WiFi/NTP, or GPS, or WWV, or whatever, then broadcasts a BTLE beacon with the correct time. Then your microwave, toaster, wall clocks, etc. could all sync to that.
https://xkcd.com/927/
Good point. How about an NTP time broad cast via a pirate WWV Faraday cage watch box, so the watch resets every night?
WiFi is a notorious powerhog and requires space for an antenna…also, anything capable of using NTP needs power as well…
A simple matter of coverage. A lot of us lives outside wlan coverage. Your puny wlan covers a few tens of square meters. The DCF77 signal covers thousands of square kilometers.
It’s going to happen. I bought the amazon smart microwave for this very reason.
The time is always perfect.
This person probably just made his watch a lot worse. In almost all cases MEMS based oscillators are less accurate than crystal oscillators. The crystal used in the watch to begin with was quite good. Plus unless the watch is dirt cheap, there is probably a way to adjust the crystal’s frequency inside the watch, either through a variable capacitor or through firmware accessible by an authorized service center. Case in point: I have a Casio MTP-1240 quartz watch. This high quality watch costs around $26 USD if you shop around. For example:
https://www.amazon.com/Casio-MTP1240D-2A-Silver-Stainless-Steel-Quartz/dp/B003F2WD6E
In a month the watch will be off by less than 10 seconds, that’s roughly four (4) parts per million, and that’s with me taking the watch off when I sleep which causes significant temperature variation. This is how the watch performed out of the box, and it’s still working this well more than five years later. (I have GPS, WWVB, and NTP disciplined clocks for comparison.) I’m never going to get this kind of performance by hacking this watch with a MEMS oscillator.
Just being able to do something doesn’t make it worth doing.
BTW) In his crazy clock project, @nwsayer explains a procedure to get an ATtiny to keep time to within ~.1 ppm. github.com/nsayer/Crazy-Clock