I'm looking for a low power clocking solution and it looks like crystals bottom out at 32 kHz. Are there any low power (nA) solutions (crystals/ceramic resonators/custom transistor oscillator circuitry) that operate on nanoAmps of power in the single digit kHz range or lower?


I wanted to summarize the facts from the answers below as well as my own research:

  • Crystals do not commercially exist lower than the standard 32 kHz flavor, due to size/resonance constraints of the quartz used internally (thanks to Olin Lathrop)
  • For a 32 kHz clock solution in the 100s of nA range, this oscillator IC could be used (thanks to stevenvh)
  • For lower frequencies (but not necessarily nA current consumption) many silicon oscillators, frequency synthesizers, PLLs, or real time clocks include internal clock divider circuitry and can be used to generate clocks as "slow" as 1 Hz.

So there is no solution which satisfies both of the constraints of sub-32 kHz & sub-\$\mu\$A operation, but individual solutions that will satisfy one or the other.

  • \$\begingroup\$ What do you need it for? I guess it's not that you just want a low power oscillator? \$\endgroup\$ – stevenvh Jun 28 '11 at 16:58
  • \$\begingroup\$ not that I am aware of, I would of suggested an RC timer but probably a bit hard to do at nA! \$\endgroup\$ – smashtastic Jun 28 '11 at 17:00
  • 2
    \$\begingroup\$ Through very careful design, a 32kHz crystal oscillator can operate in the nA region. You can also get custom crystals cut at lower frequencies, but it's very likely that they'll be larger than today's common 32kHz crystals. Is the requirement for a slow oscillator, or are you assuming that you need one in order to meet the power requirement? \$\endgroup\$ – Adam Davis Jun 28 '11 at 17:07
  • 1
    \$\begingroup\$ @Adam - Do you expect a few nA to be sufficient to keep the oscillation going? How do you do that? What's the "careful design"? \$\endgroup\$ – stevenvh Jun 28 '11 at 17:16
  • 1
    \$\begingroup\$ @Joel - I see, that's also what I thought of. And you can't spare a few \$\mu\$A for an oscillator? \$\endgroup\$ – stevenvh Jun 28 '11 at 17:50

Some manufacturers refer to 100kHz ~ 1.8432MHz as "very low frequency". Below 100kHz you'll find only a few specific values like 77.5kHz for (DCF77 receivers) and of course the 32.768kHz watch crystal. You won't find standard crystals below 10kHz.
Nothing lost, by the way. They won't help you meeting your low power requirements, anyway. A crystal needs a certain power to maintain its oscillation, and if you look at the datasheets you'll often see drive levels in the order of 1mW.

The lowest power microcontroller application using a crystal I designed was an MSP430 running at 32.768kHz, and that used 3\$\mu\$A, a thousand times your requirements.
This oscillator still uses 300nA, that's 0.3\$\mu\$A.

  • \$\begingroup\$ I believe that a CPU that was clocked by a crystal below 7KHz and had all switching on edges defined by that crystal (without clock multipliers, etc.) would be exempt from FCC testing requirements. One could clock a CPU using an RC oscillator at such frequencies, but not terribly accurately. \$\endgroup\$ – supercat Aug 8 '11 at 22:28

I don't think you're going to find much that can be run at lower power than a 32,768 Hz crystal. That's what wrist watches use, and they need to be small and low power and generally succeeed well at that. If there was something lower power with crystal accuracy, I think it would be used in wristwatches.

Also consider the mechanical aspects. Crystals work on carefully cut quartz vibrating at the desired frequency. The lower the frequency, the physically bigger the crystal needs to be. The 33 KHz watch crystals are already cut like tuning forks to get a low frequency with small size. Again, if there were tricks to be had, I think wristwatches would be using them.

  • \$\begingroup\$ Do you have any data on crystal watch current draw? (RTC chips like Dallas DS1302 claim 200 nA typical at 2 V. I doubt our correspondent will get much lower power than this.) \$\endgroup\$ – markrages Jun 28 '11 at 18:14
  • 1
    \$\begingroup\$ @markrages: neat! I wonder if by 'nA' Joel B means 'less than 10' or 'less than a few hundred'. I don't think the first is possible. This may be the best he can get. \$\endgroup\$ – Federico Russo Jun 28 '11 at 18:24
  • \$\begingroup\$ @markrages - indeed nice (even if the datasheet actually says 300nA, not 200). Not even EMMarin does better, and they're usually good at very low power. \$\endgroup\$ – stevenvh Jun 28 '11 at 18:39
  • \$\begingroup\$ 200 typ, 300 max. \$\endgroup\$ – markrages Jun 28 '11 at 18:43
  • 1
    \$\begingroup\$ @markrages - Did your mom never tell you that typical values are for sales engineers? Design engineers and other Real Men use maximum values. \$\endgroup\$ – stevenvh Jun 28 '11 at 18:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.