I'm working on a long-life IoT sensor device which spends probably about 99.99% of its life asleep, waking up every 15 mins to take measurements and send data. I'm sure this is a problem that many people face as IoT becomes more popular.

I've figured out a few things to start: To get the longevity I require from the system the sleep mode current has to be tiny. Combined with this the primary cell has to be a high-capacity, low-self-discharge cell (Lithium Thionyl Chloride 3.6V battery, for example).

For example's sake, lets say the current requirements are typical of my typical wireless IoT device varying as follows, depending on the mode of operation over a period of 15 minutes:

  • Sleep mode: <10uA (99.99% of the time)

  • Measurement mode: 0.5-10mA (~10s)

  • Transmission mode: up to 50mA (2-5s)

Ideally I'd power the device straight from the cell, but a couple parts I'm using have a max input voltage of 3.6V, and a few LiSOCl2 cells I've seen will sit at 3.7V before settling into their nominal voltage of 3.6V.

I'm aware the cell voltage may droop when under heavier load, so to be safe, 3V seems like a reasonable level to operate without being affected by the droop.

What is the best way to go about efficiently regulating this sort of primary cell to provide around 3V? The regulator would also have to consume very little current itself.

  • \$\begingroup\$ There's a lot of microcontrollers that surpass your specs by a solid amount. There's also microcontrollers that include RF / networking functionality. Many of those also have a wide input voltage range, so I'm not convinced your question is overly specific - what MCU did you have in mind? Generally, your application describes what would usually be done with an LDO, but I really don't think one would generally be necessary. \$\endgroup\$ May 8, 2017 at 12:37
  • \$\begingroup\$ What battery lifetime are you wanting? Unless you want 5 - 10 year range you probably do not need LTC cell. \$\endgroup\$
    – Russell McMahon
    May 8, 2017 at 13:00
  • 1
    \$\begingroup\$ is the mcu that sleeps/wakes a 3.7v tolerant part? If so, you can enable an LDO with one of its outputs for measurement/transmission, so as not to waste an LDO's bias current when you don't need to. Look for one with a vanishingly small disabled current. \$\endgroup\$
    – Neil_UK
    May 8, 2017 at 13:16
  • \$\begingroup\$ @Neil_UK, that's a great idea, didn't think of that. The MCU (CC1310) can take up to 3.8V so it should work. Are LDO's similar in efficiency to buck converters? \$\endgroup\$
    – Sensors
    May 8, 2017 at 13:29
  • \$\begingroup\$ @RussellMcMahon I am aiming for that sort of lifespan (>7 years) which is why I selected the LTC cell. \$\endgroup\$
    – Sensors
    May 8, 2017 at 13:29

1 Answer 1


OK, you're worrying about overvoltage.

I've never designed something like you're doing, but I wanted to focus your attention on something you could have forgotten: temperature. Temperature is something that has a tendency to behave like its other name is Murphy.

I know nothing about Lithium Thionyl Chloride, so I googled it, clicked on the first link, then clicked on a random datasheet for a random cell...

And whaddyaknow, the open circuit voltage rises with temperature at an alarming rate.

Will your stuff be exposed to direct sunlight?...

I would advise using a high-current CMOS micropower LDO.

High current, because you want the pass device to be enormous, so its dropout voltage is tiny under load.

CMOS and micropower, because idle current isn't your friend.


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.