Hi.
I'm looking for a very accurate home clock, and many search results point to DIY projects using DS3231, they say it's the most accurate chip you can get without cost a leg.
I just want to know, is there any commercially available clock that have DS3231 as it's gut, or something else with equal or even greater accuracy?
I know a Wifi connected clock would be always accurate, but I just want a stand alone time source.
Thanks.
I think even more accurate would be a clock that receives the time signals from WWVB in Colorado. They are called "self-setting", or sometimes "atomic" clocks. They need to be able to receive the very low frequency WWVB signal, but that frequency goes through walls pretty well. In theory you never have to set them, and will never be more than one second off if you have good reception. But even if you get a good signal only at night, it's never going to be off enough for you to notice. Look at the WWVB entry on Wikipedia.
I suspect most normal clocks just use a crystal, which is a lot cheaper than a DS3231.
Thank you, I'm aware there are sync methods other than NTP, like WWVB and GPS, but their signal reception is a big problem.
DS3231SN does cost an arm and leg USD $12 ea. so there are clones 1/10 the price at large. Some are good.
I have a few
DS3231 Aliexpress clock kits one is very accurate and another (plexie) runs fast, performs on par with a DS1307- not so accurate. I haven't tried replacing that RTC with a reputable part, or it could be the clock's crappy firmware.
I think it would be possible to hack a better reference into any cheap quartz clock? You can buy SMD 32kHz TXCO modules cheap. I think those salvaged 10MHz oven oscillators are probably the next step up in cheap precision.
Around here you better spec exactly how much drift you can stand or someone will suggest using an atomic clock.
I used one of this chip's "cousins" in a project before (DS139X). The accuracy was based on the quality of the crystal, and keeping noise away from the crystal. The thing drifted a few seconds a day. If I redid the project, I'd switch to what you have here in a heartbeat. It may seem a little expensive, but not having to worry about the crystal is easily worth $10. I'd use a 4 layer board. Keep this chip as far away from other things as you can. And possibly try to use WiFi or something periodically do corrections.
The big advantage I see in the older chip is that it had a battery charger that would work with super caps. A 2 farad super cap would power the chip for at least 2 months when the main supply was disconnected. The chip could only keep track of time. It could not actually be accessed without main power. But one of the main goals of the project was to deal with week long power outages.
I think it would be possible to hack a better reference into any cheap quartz clock? You can buy SMD 32kHz TXCO modules cheap. I think those salvaged 10MHz oven oscillators are probably the next step up in cheap precision.
Around here you better spec exactly how much drift you can stand or someone will suggest using an atomic clock.
Yes I thought that earlier, and here's my found:
https://electronics.stackexchange.com/a/629657https://dpwe.github.io/arduinoclocks/2022-07-21-bedside-tcxo.htmlIn this case it maybe better to just use a TCXO instead of a full RTC like DS3231.
I don't have a hard limit since it's for personal use anyway, but I think 1 minute drift per year would be good enough.
How stable is your ambient temperature?
To my huge surprise, a firmware compensated RTC on an ATtiny provides my office wall clock with a time that is accurate to about 1s/month.
Clock source is a 2^15Hz watch crystal, no TXCO. My office temperature is between 18..30°C max, usually 20..22°C.
Edit: I guess I misunderstood, sorry. You are not looking into DIY, but off the shelf?
Edited: The DS3231 has a surprisingly high supply current when the serial interface is up (100µA). This is what made me bring up the my ATtiny(414), which uses <4µA for time keeping. A more thorough look at the datasheet will show that the DS3231 should require about 1µA for pure time keeping.
A more thorough look at the datasheet will show that the DS3231 should require about 1µA for pure time keeping.
But I think that's only if you power it through the Vbat pin. If you power through Vcc, it's going to be 100uA even if nothing is going on with I2C.
How stable is your ambient temperature?
To my huge surprise, a firmware compensated RTC on an ATtiny provides my office wall clock with a time that is accurate to about 1s/month.
Clock source is a 2^15Hz watch crystal, no TXCO. My office temperature is between 18..30°C max, usually 20..22°C.
Edit: I guess I misunderstood, sorry. You are not looking into DIY, but off the shelf?
Normally it will be as stable as a living room, but I would like it to be more robust (temperature irrelevant).
Yes initially I was looking for a off the shelf, but I should have realised I'm going to DIY since I'm asking in this forum.