Very, Very Low Power Consumption

We’re pretty far away from a world full of wall-warts at this point, and the default power supply for your consumer electronics is either a microUSB cable or lithium batteries. USB ports are ubiquitous enough, and lithium cells hold enough power that these devices can work for a very long time.

USB devices are common, and batteries are good enough for most devices, not all of them. There is still a niche where& extremely long battery lifetimes are needed and tapping into mains power is impractical. Think smoke detectors and security systems here. How do power supplies work for these devices? In one of the most recent TI application notes, TI showed off their extremely low power microcontrollers with a motion detector that runs for ten years with a standard coin cell battery. This is one of those small engineering marvels that comes by every few years, astonishing us for a few minutes, and then becomes par for the course a few years down the road.

The first thing anyone should think about when designing a battery-powered device that lasts for years is battery self-discharge. You’re not going to run a battery-powered device for ten years with a AA cell; the shelf life for an Energizer AA cell is just 10 years. Add in a few nanoAmps of drain, and you’ll be lucky to make it to 2020. The difference here is a CR2032 lithium-ion coin cell. Look at the datasheet for one of these cells, and they can easily sit on a shelf for 10 years, with 90% of the rated capacity remaining.

With the correct battery in the device, you’ll need a microcontroller that runs at a sufficiently low power for it to be useful in the mid-2020s. The product for this is the CC1310, a very, very low power ARM Cortex-M3 and sub 1GHz transmitter in one package.

Once that’s settled, it’s simply a matter of putting a sensor on the board – in this case a PIR sensor – and a few analog bits triggering an interrupt occasionally. Have the microcontroller in sleep mode most of the time, and that’s how you get a low-power device with a battery that will last a decade.

27 thoughts on “Very, Very Low Power Consumption

    1. Size does not matter but you have to be realy careful about using pull-up and pull-down resistors. Avoding them if not absolutly necessary and even then use values in the Mohm-s

    2. P=IV and V=IR. So P=I^2R or P=V^2/R. In every case the V and I would be the same for a certain R, and thus the P would be the same too.

      The power of the resistor is the max power dissipation before it starts letting the smoke out. In that case yes throughole typically can take more power and have a higher power rating than surface mount ones, but ultimately that just depends on resistance.

  1. Throw on a larger lithium-thionyl battery (like an ER34615) and it’ll last a lifetime. I can see some fun hacks coming out of this thing. Possibly a few spooky ones as well.

    1. This guy is an idiot, I stopped reading when I got to ‘2 – On Typical Specifications’ and he references the chart of recommended daily consumption of various categories of food (fat, cholesterol, sodium) which is on just about all processed foods in america. For those not familiar with this chart, it is essentially the FDA saying ‘according to our studies, we recommend that you distribute your diet across the various food groups according to this chart’

      Ganssle however, presents this chart in a different light:

      Here’s another datasheet — it’s from a box of Entenmann’s donuts:
      [image as described]
      These are worst-case specs. They are guarantees. Contractual terms between Entenmann’s and the buyer. You know what you’re getting.

      Yep the recommended daily dietary suggestions proposed by the FDA sure are contractual terms between Entenmann’s and the buyer. Perhaps if he has shown a picture of the actual nutritional values reported for this particular food item his statement would have made sense, because a food manufacture is obligated to report these values to within reasonable measurement errors, but even still the values presented are typical values, not some sort of absolute maximum or minimum values as he suggests.

      The rest of his analysis is not much better…

          1. He says he stopped reading after the ‘2 – On Typical Specifications’ and then says “The rest of his analysis is not much better…”, so do i need to say more.

        1. J,

          I found your research very interesting and well presented. Personal experience has shown that many surprised lurk for those attempting to design a part into a device based on it’s ‘specified ratings’.

          Sometimes you will underestimate what the part can do, but more often it will have particular fail modes, glossed over idealism in datasheet parmeters, and so on.

          Its strange fuckface up there reacted so strongly, and with this strange gibberish about the FDA..

          Good work!

      1. Jack Ganssle’s page actually looked pretty reasonable to me. Even with the food reference (relevant as other industries include typical and maximum values in their data sheets too). He has some good, thoughtful analysis about losses and speed vs. power consumption. Having worked in the semiconductor industry I can tell you that maximum values on a data sheet are real, but often at very seldom found conditions (low temperature limit, worst case IC processing corners, …) and always with a healthy safety margin applied to them. A customer making their own measurements on a set of parts across the temperature, voltage and frequency range they want to use is actually common practice for many designers (and then adding a statistical-based guard band). I’ve actually had to set maximum values for data sheets and the rule of thumb was to make them pessimistic enough that no customer would ever have cause to return any parts.

        1. easier to nullify analysis with example. 4 years into our project, I see no reason why our RF strain gauges will not last 25+ years. They only need 6 for the application and 11 will be considered awesome. Maybe he shouldn’t have picked a battery chemistry that is retarded for a 10+ year applications. Just a thought.

    1. The evaluation board is for testing ultra low powered sensors. There are piezo generators that make enough power for devices like this from the vibrations of people walking by. Typical uses are keypad- or card locks that don’t need batteries, environment sensors cast into the concrete of a building.

  2. The problem with things like this is that in 10 years the battery dies, while you already forgot about the device 5 years ago (or never knew was there in the first place) so the battery never gets replaced. In the end the thing might as well never be there in the first place. Apply this to a device that is life critical, like a smoke detector or security device, and it’s even worse.

  3. If you ask me, the biggest deal of this is the actual low power PIR sensor, as it draws 1.6uA. Compare that to the typical ones drawing much more.
    If I would design this…. well, it is big enough to put a couple of AA on the back, so I would do that.

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.