What is the most energy-efficient component for measuring long durations?

[clauswobbe]

Which electronic component (or other means) would you use to measure duration, like a timer, if energy must be saves to the maximum?

This component should use as little power as possible, while being able to measure duration up till 5 years with a resolution of, say 100 hours.

In the old days one would use an integrated circuit called 555, but today I'm sure better ways exist :-)

 

[Simon Bridge]

Welcome to PF;
the old 555 timer is still good.
you can also use an RC circuit.

A big tub of sand that runs out and a capacitor weight?

 

[CWatters]

If you need battery life in years you need to look at things like battery self discharge rates. Some battery types will self discharge in months.

I think even low power versions of the 555 burn 300-400uA. So for 5 year life the battery would need to be somthing like 20Ah.

 

[Danger]

A big tub of sand that runs out and a capacitor weight?

Curse you, Red Baron!
I fully expected to be the only person here non-techy enough to suggest a sand clock. I still think, for the requirements given, that it's the best choice.

 

[clauswobbe]

Hehe - funny guys! :-)

I was thinking of powering my "gadget" through induction from the (weak) EM field surrounding an electric bulb. Hence the need for ultra low power consumption. I'm afraid a sand clock will be a bit too clumpsy, however, I'm really, really into keeping it simple :-)

 

[NascentOxygen]

I was thinking of powering my "gadget" through induction from the (weak) EM field surrounding an electric bulb.

You have a light source! So, what's wrong with using a solar cell, perhaps charging a NiCd cell to allow operation even in the dark?

 

[clauswobbe]

That's a possibility. My thought was, though, to avoid being dependant on having actual line of sight with the bulb. And further, the gadget must only work when the light is on, so no need to store any juice :-)

 

[Danger]

Hehe - funny guys! :-)

While I appreciate the applause, I was actually very serious about a sand clock. If you're worried about a large dune not being reliable enough, you could also go with a water clock deriving its input from a local stream. After all, the lifetime is so-so and your accuracy requirement is almost non-existent. (I could throw a dart at a calendar and be within 100 hours of when I wanted.)
I agree with NO that you should have mentioned a light bulb right off the bat. Why not just buy one of those bulbs with a 5 year guarantee and tie your circuit into it? As with everything else, it's a sure bet that the thing will go tits-up the day after the guarantee expires, which you can use as the trigger for your "alarm" circuit.
Okay, that time I was kidding.

edit: This would really be a lot easier (but less fun) if we knew just what it is that you want to accomplish.

the gadget must only work when the light is on

In that case, lose the battery and just power whatever it is directly from the photovoltaic supply.

 

[clauswobbe]

Oh, I forgot to mention another requirement, I'm looking to meet: this thing must be small - ideally dime size...

 

[Danger]

Oh, I forgot to mention another requirement, I'm looking to meet: this thing must be small - ideally dime size...

Am I going to have to crawl through this monitor and slap you?

 

[CWatters]

How about one of the Real Time Clock chips from companies like Intersil. ...

http://www.intersil.com/content/dam/Intersil/documents/fn63/fn6314.pdf

Designed to run from small batteries.
Pretty sure some will wake up at a predetermined time and date and generate an interrupt.
Will even allow for the leap year.
I^2C so easy to program it and then disconnect the programmer.

 

[NascentOxygen]

Oh, I forgot to mention another requirement, I'm looking to meet: this thing must be small - ideally dime size...

A $15 lady's digital wristwatch (it keeps track of months & years). Have a jeweller install a 5 year battery and all your problems are solved!

Or maybe I have too loosely interpreted "measure duration"?

 

[Carl Pugh]

clauswobbe, is what you are looking for a running time meter or something similar?
There used to be a device that measured time by plating a copper wire. The thicker the plating, the longer the time.

 

[Danger]

There used to be a device that measured time by plating a copper wire. The thicker the plating, the longer the time.

That sounds weird enough to be really cool. How on Earth do you measure it, though, without sticking it under a microscope? Some kind of slip-fit gauge? Electroplating would take a lot of power, so I'm assuming that this is some sort of natural accumulation like a patina...?

 

[Simon Bridge]

I've heard of these accumulation things - the conductivity changes.

Note: a US dime is about 17mm across right? Will a 555 timer circuit fit inside that? The IC case will but you need a resistor, two capacitors, and a power supply. The ideal electricity-efficient timer draws no power at all until you need it to go off.

The sand clock still works - but you need nanoparticles and a very small aperture to get the size down.
Maybe buckyballs? The details are engineering.

You could also use a radio-isotope clock.
Those things can be tiny.

I friend of mine once built a timer out of an RC circuit - timed two years to a couple of hours. When the timer ran out, it dumped the contents of another capacitor through a piezo buzzer. He'd hid it in the headmasters office at his school and needed it to go off at a particular time on a particular day but well after he'd left.

Lastly - I canvasses the people I know who are still into electronics and the 555 timer IC is still used for breadboard rigs etc. Otherwise you use crystals.

 

[clauswobbe]

Wow, great input! All of it! Sorry for being so secret about my end goal here - it's for a proof of concept for an invention of a new product, so I'm just protecting the ip :-)

Anyway, your suggestions are really powerful - thanks a lot. I have input for my project :-)

 

[vk6kro]

You can get CD4521 chips on EBay.
These let you divide by up to 16 777 216 or 2^24.

So you can work out the number of seconds in 5 years (3600*24*365*5) and drive the 4521 at an appropriate frequency to produce a square wave with a 5 year period.

I get a frequency of 0.1064 Hz or a period of 9.3984 seconds.

This is for a complete square wave, but the first "high" output would be after 2.5 years. So if this "high" after 5 years was required, you would drive at a period of 18.79 seconds.

You could start with an oscillator at 435.814 Hz. Divide by 8192 then by 2^24 to get a period of 10 years with a "high" out after 5 years.
Note that an error of 1 Hz in the oscillator frequency results in an 8 day error in the output, so a crystal oscillator source would be a good idea.

The quiescent current drain of this chip is about 0.05 uA and the storage life of alkaline batteries is about 10 years.