I'm working on a project called The Fifty Year Beep. The goal is to produce a battery-powered circuit that will beep once and only once fifty years from now. It is an art project of a kind, inspired to some degree by the 10000 year clock. I love the idea of me and my friends gathering together as old men to listen to a single derpy beep.

Ideally, the clock is self contained and powered. Our current thinking is that lithium cell batteries are our best bet, given that there are examples of computers from the 90s where the button cells are still good.

We started looking at off the shelf RTC module like the DS3231. The datasheet lists a timekeeping current of ~1uA at ~3V. A CR2477 coin cell has ~1000mAH at ~3V. Naively, this gives a lifetime of about 100 years. Naively.

My main question is about the choice of RTC module. The DS3231 has an alarm feature that seemed promising at first. However, as far as I can tell, the alarm can’t be set more than one month in the future.

We don’t have experience selecting or searching for microchips. Is there a single chip that can do this job? Or should we be building a system that combines an RTC chip with a low power microcontroller?

Any suggestions or advice would be appreciated!

  • 2
    \$\begingroup\$ A 32 bit counter will hit 100 yrs on 1 cycle, with a slow clock but you might be anxious that it won’t work. You better add a solar cell. \$\endgroup\$ Commented Jun 29, 2021 at 2:47
  • 2
    \$\begingroup\$ 50 years +/- 1 second, or +/- 1 year? What accuracy timekeepig? A low drain CMOS oscillator could take less than 1 uA, CD4060 says typical quiescent is 40 nA, but that goes up radically with temperature, what's your environmental specification? \$\endgroup\$
    – Neil_UK
    Commented Jun 29, 2021 at 4:22
  • 3
    \$\begingroup\$ You should use a Lithium thionyl chloride battery instead. These are used in long lasting remote applications such as power-meters for homes. I also recommend that you find a temperature compensated RTC module, because uncompensated crystal oscillators can drift significantly with temperature, unless you are ok with several 10's of hours of time drift. \$\endgroup\$ Commented Jun 29, 2021 at 5:06
  • 2
    \$\begingroup\$ The smaller MSP430 CPUs can sleep with the RTC clock running (to wake up some programmable time in the future) on less than 1 uA. Then the programming (and maybe a 1 second 1mA LED flash every year) is entirely under your control. Remember to account for 20ppm or so crystal accuracy and drift ... what is 20ppm of 50 years (1.5 Gigaseconds)? \$\endgroup\$
    – user16324
    Commented Jun 29, 2021 at 12:04
  • \$\begingroup\$ @TonyStewartEE75 I think a solar cell makes a lot of sense, thanks! \$\endgroup\$
    – ishmandoo
    Commented Jun 29, 2021 at 23:13

3 Answers 3


Given the application, 50 years with chemical battery is quite unlikely to happen as they are not designed for such lifetime.

Self discharge comes first in mind, CR2477, depending the chemistry is 2% per year, Lithium thionyl chloride is 1%, wiping 50% off.

Another issue is the ageing effects on the electrolyte, the degradation of the casing and so forth, and you don't really have a way to test it, it would be bad to wait 50 years and it doesn't work.

You could have a look at tritium based photoelectric generator, but usually those offers a current in the range of tens of nA, and getting tritium, depending where you live might be complex.

You could eventually contact this company if you are based in the US.

Besides nuclear based solution, your best bet is to have a photovoltaic panel and a super-capacitor.

If your system will be indoor, you can use a thin film panel that has good efficiency in low light, and scale the system so that it can run for a few days without light.

You can use an ultra-low power RTC like MAX31341B (180nA time keeping) and ultra-low power MCU like ATSAML10E14A 500nA sleep or 100nA in off mode.

  • 4
    \$\begingroup\$ For the MCU, I've had an STM32L053 run at 400 nA with the built in RTC working (tested by toggling a dangling GPIO once a second). \$\endgroup\$
    – jaskij
    Commented Jun 29, 2021 at 10:30
  • 1
    \$\begingroup\$ @JanDorniak that is a good start \$\endgroup\$
    – Damien
    Commented Jun 29, 2021 at 10:32
  • 1
    \$\begingroup\$ Yes. It is relatively easy to get a low-drain clock. The hard part will be dealing with self-discharge and shelf-life issues of the battery. The other difficulty will be finding a location that can be relied upon for 50 years. But that is not an electronics issue. \$\endgroup\$
    – user57037
    Commented Jun 29, 2021 at 16:28
  • \$\begingroup\$ These are really useful comments, thank you! I wasn't aware of lithium thinyl chloride, I'll definitely look into that. I think the PV idea makes a lot of sense too. \$\endgroup\$
    – ishmandoo
    Commented Jun 29, 2021 at 23:10

The 10,000 year clock project uses the daily (dinural) thermal change to power it. This seems like a more viable approach than relying on a battery alone. A solar collector and solar cell could also work.

If your piece is intended to be viewed by people, using a piezoelectric mat nearby could harvest some energy just by people standing in front of it.

If there isn't some thermal gradient, solar radiation or... humans nearby, getting power from radioisotope decay is possible, especially so if you’re JPL or otherwise have connections that give you access to this technology. Might not be practical for the rest of us.

That said, here's a tritium decay-based battery you can buy, for a price: https://citylabs.net/products/

Combine one of these energy harvesting modes with a super capacitor and it could provide the needed current for a low-power RTC that has an 'alarm' output.

As far as how to make the 'beep', lots of options for that. It would only need to be powered on when the alarm fires off. Could be a micro with low leakage or some low-power discrete solution.

  • \$\begingroup\$ The nuclear diamond batteries are a scam. Citylabs stuff costs north of $5k, but at least it's an actual product you can buy. I've edited the question to remove the mentions of NDB, since there's no need to give scammers any exposure. \$\endgroup\$ Commented Feb 23 at 3:13

If one wants to build a build a ludicrously low power rtc, i highly recommend AMBIQs line of RTCs (Artasie line). I2C or SPI, ~256 bytes of user RAM, two alarms (as far as I know), built in oscillators (two, if I recall correctly) and a current consumption of ~ 70 nW with autocalibration active / 50 without it. Yep. Nanowatts. VDD 1.5V - 3.6V.

[The following are really basic calculations, not considering voltage curves, etc. but you get the gist]

A typical primary lithium battery 3V CR2032 holds ~ 225 mAh worth of juice. It has a self discharge rate of roughly 1% per annum. The RTC burns through 70×24×365,24 nW per year. So you could easily power such a device for more than 100 years. I mean you could: solar, kinetic, thermal, rf harvesting etc. But what would be the point? :)


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.