SparkFun will be closed on Monday, May 30th, for Memorial Day. We will resume normal business hours at 9am (MT) on Tuesday, May 31st. Orders placed after 2pm (MT) on Friday May 27th will process and ship out on Tuesday, May 31st. From all of us at SparkFun, we wish you a safe and happy Memorial Day weekend!

Rock Bottom Power

How low can we run a microcontroller?

Favorited Favorite 0

I feel like I'm always posting news on businessy topics. To balance this affliction against my inner geek, I sat down last night and wrote a tutorial on low power management on the ATmega328. Check out the Adventures in Low Power Land:


I got an ATmega328 to run down to 1uA!! I think I can generate more energy rubbing my fingers together than what this 8MIPS power house sips for power. Incredible! Checkout some of the lessons learned.

Comments 22 comments

  • TeslaFan / about 11 years ago / 2

    Can you even legally measure down that far with that set up? I think those jumpers alone need more than that to stay up!

    • sgrace / about 11 years ago / 2

      If you've watched EEVBlog, you might know that the creator David Jones made a project to measure that low with good accuracy.
      It's called the uCurrent. I would check it out.
      The DMMs that Nate is using, not sure on the accuracy, but the uCurrent would be a good device to use to check.

    • mkings01 / about 11 years ago / 1

      Yes you can. KCL.

    • Kilo / about 11 years ago / 1

      Imagine the measurement uncertainty at that lo range on a hand meter without kelvins..+/- 30%?

      • mkings01 / about 11 years ago / 1

        Thanks to KCL, the only thing you need to worry about in this case is current leakage through other paths, and that will be negligible. 1uA is a good measurement.


  • 48X24X48X / about 11 years ago / 2

    You guys might want to try an Arduino compatible low power library that we wrote: https://github.com/rocketscream/Low-Power
    if you don't want to mess around with the fuses.

    • Sparko4Marko / about 11 years ago / 1

      A shameless plug...
      You can build your own ISP, for this project and other Atmel chips. Tutorial link below.

  • joe555 / about 11 years ago / 1

    Idle current is one of the important factors in low power design. However, there are more factors including:
    1. How long does it take to change from idle to active in order to do something useful? It will impact the overall budget and if it is too slow, it may cause you to miss the event.
    2. What is the active mode current? Even for low duty cycle, it has an overall impact.
    3. Can it do something useful at low power mode? e.g. DMA capture.
    TI MSP430 and Microchip PIC XLP series are very good and metioned. There is a newcomer, EnergyMicro EFM, worths taking a look. 0.9uA draw, with 32KHZ osc on, brown out detector enable, and memory data retention.

  • Promptcritical / about 11 years ago / 1

    For $6 you can get an energy harvesting chip, hook it ti a TEG or a pezio and not worry and about micro amp sleep modes.

  • noworries / about 11 years ago / 1

    Yes, air is a better insulator than a PCB, but it is seldom that one needs to resort to such extreme measures as the cost of a production run with such techniques used would likely be prohibitive.
    I have used the technique Bob suggested on both DIP and SOIC parts in the past for very high impedance / low leakage circuits which works well for small production runs.
    It is sad that Bob is no longer with us, he will be missed.

  • Is it just my browser, or is the article date really saying it's the 18th? Last I checked, today is the 17th.

  • mkings01 / about 11 years ago / 1

    what this 8MIPS power house sips for power... while running at 0 IPS.
    1uA sounds low, but any modern micro-controller worth its salt can do it.

  • noworries / about 11 years ago * / 1

    Another note regarding low power microcontrollers is to avoid the temptation of setting unused GPIO lines as inputs without tying them either high or low. While it may seem to allow for lower current consumption on the bench, later on some leakage current may cause a GPIO line to float to a voltage that is halfway between a high and a low, where the current that will result internal to the microcontroller will be several milliamps.
    Setting an unused pin as an input with an internal weak pullup is probably your best bet, but don't be tempted to leave them floating without a pull up or pull down resistor be it internal or external.

  • noworries / about 11 years ago / 1

    You may even be able to get slightly less current consumption if you do away with the breadboard and are careful to keep the components clean and dry. If you do the math, 1uA at 3.3 volts is an equivalent resistance of 3.3 Megohms. Breadboards that have been used a while can be pretty leaky, a clean epoxy PCB is better. Do clean the flux off, however. Also, you may want to consider a low leakage conformal coating for the components and printed circuit board if you desire the current consumption to be kept low for long durations (months - years). This is because leakage that may occur due to humidity and PCB board contamination over time may become significant sources of power loss for your battery.

    • TheRegnirps / about 11 years ago * / 1

      As long as it is a DIP part, you can do even better with no PCB and something like Kynar wire-wrap wire soldered to the pins of various parts. Ball it up when finished for super light weight and lowest possible leakage. Here is a Bob Pease circuit.
      Ooooh! Linear 1 and Linear 2! Along with National's "Linear Applications", they are an education in themselves. If you can find an old copy of the Linear Applications data book, get it by hook or by crook.

  • SomeGuy123 / about 11 years ago / 1

    Is that a Sparkfun mouse pad I see on the far right of the picture?

  • Hmm. This article isn't showing up in Google Reader OR iGoogle.

Related Posts

Recent Posts


All Tags