Low-cost power converter for low-power Arduino?

Hello, electronics gurus... I need to call on your help again :slight_smile:

I am working toward a small, low-power Arduino node that I can deploy many of. Now I am thinking about how to power it. Roughly I was thinking to use two rechargable AA cells, and a step-up converter to 3.3V. I'd like to be able to leave it there as long as possible, so have the converter squeeze as much power out as possible. Of course, the ideal part would use little of its own power. And I'm sensitive to price.

Right now I am looking at the MC34063EBN (Datasheet) Page 5 contains a schematic for a step-up circuit that looks like the one I'd use.

Questions:

  • Does it seem like this circuit would do the job?
  • Is there a different part I should look at instead that would better achieve my goals?
  • Can a 'step up' circuit really only step up? Does that mean I couldn't use three AA batteries (3.6V) because it would have to step down to 3.3 at first?

Thanks so much!

If the processor is a "pico power" version, it can be powered directly from 2 AA batteries. This is by far the most efficient way to provide power.

However, other devices may not tolerate the low voltage or erratic current delivery of dying batteries. What other hardware will you be using?

Running ATMEGA328P-PU, which is pico power, right?

For one, an nRF24L01 radio. It's rated for 1.9-3.6. So I guess if I went with straight batteries, I could stack 3 as far as that guy is concerned.

Haven't settled on the sensors, but I suppose I can pick them for voltage range. For temp, I had been looking at MCP9700 which can go down only 2.3V. Though I'm guessing those rated voltages have some play in them that could be taken advantage of for "around the house" application?

maniacbug:
Running ATMEGA328P-PU, which is pico power, right?

From the datasheet...

Speed Grade:
0 - 4 MHz@1.8 - 5.5V, 0 - 10 MHz@2.7 - 5.5.V, 0 - 20 MHz @ 4.5 - 5.5V

Yup. Bear in mind that the maximum recommended processor speed at that voltage (2*1.2=2.4V) is <= 4 MHz. I haven't had any problems running ATtiny85V processors (similar specifications) at 8 MHz / just over 1.6 volts.

For one, an nRF24L01 radio. It's rated for 1.9-3.6. So I guess if I went with straight batteries, I could stack 3 as far as that guy is concerned.

1.2 * 3 = 3.6. That gets you nicely into the 10 MHz band allowing you to reliably run the processor at 8 MHz. If you can, running the processor at 1 MHz will significantly increase the battery life.

For temp, I had been looking at MCP9700 which can go down only 2.3V.

I can't provide any help regarding sensors.

Yeah, all is well on the day with fresh new batteries :slight_smile: That's why i was thinking about a step-up, to keep the voltage there the whole time. I suppose the thing to do here is just run a dying batteries test.

Oooh, scary! :open_mouth: I'm just wrapping my head around 8 MHz. Is there even a boards configuration for lower? Or do I have to learn something yet new :wink:

maniacbug:
Yeah, all is well on the day with fresh new batteries

It's my understanding that NiCd batteries provide a consistent 1.2V until almost dead.

I suppose the thing to do here is just run a dying batteries test.

The first step is to determine if the batteries will be damaged or cause problems if they are run-dead. Then you should decide how you want the application to behave. Should it try to "limp along"? Should it stop running and remain stopped until you intervene?

Oooh, scary! :open_mouth: I'm just wrapping my head around 8 MHz. Is there even a boards configuration for lower?

For 8 MHz? Absolutely. The LilyPad, for example, runs at 8 MHz. It's a good example for what you are trying to do.

For 1 MHz? I have a 328P application running at 1 MHz and, as far as I can tell, there are three problems: 1. The PWM frequency is too low. This can be easily adjusted. 2. delayMicroseconds does not work. 3. Some baud rates are not correct (but I can't remember the details so don't ask :zipper_mouth_face:)

How about this gadget I bought from Modern Device? (see attachment)

It provides 3.3V from a single AA battery. And the active components are so small I initially thought they were a speck of dirt (see second photo, parts circled).

According to their site:

The input can be anything from 1.0 to 5.5V, so this can also be used with 2x or 3x AA battery packs (even 4, if NiMh), and it will continue to work when hooked up to a 5V power supply. The regulator will work down to less than 0.5V and can squeeze the very last bit of energy from the battery.

I have no connection with them, I just thought it was interesting. If you want to squeeze energy from your batteries, and you can use 2 or 3 AA batteries, this could be the thing for you.

AA_battery_gadget1.png

AA_battery_gadget2.png

Yeah, that thing rocks. And $16! I would love to find a way to include something like that in my nodes but with way less cost.

ST makes a booster - L6920 - that is about the best I've found in terms of quiescent power consumption (down to some 20uA or less depending on input voltage) and low voltage startup behavior. This will run down to 0.5V or so (actually, 0.4xx V in practice if you're not drawing much current). Not sure how low in cost you want to go; there are definitely cheaper ones out there if you don't mind higher idle current.

Do you have any plans or schematics online? I'm also working on a very low-powered Arduino variant geared toward energy harvesting projects; might be good to compare notes.

Drmn4ea:
ST makes a booster - L6920 - that is about the best I've found in terms of quiescent power consumption (down to some 20uA or less depending on input voltage) and low voltage startup behavior. This will run down to 0.5V or so (actually, 0.4xx V in practice if you're not drawing much current). Not sure how low in cost you want to go; there are definitely cheaper ones out there if you don't mind higher idle current.

That's a pretty cool part, thanks! $2.33 from Mouser. Not too bad, though I probably would give up current to get down to something like $1-1.50. And then, uh, I have to confront the dreaded TSSOP8 package.

Drmn4ea:
Do you have any plans or schematics online? I'm also working on a very low-powered Arduino variant geared toward energy harvesting projects; might be good to compare notes.

I am just starting this journey. I built my first regular-power Arduino Clone, and I'm putting it through its paces (just discovered a bug today!). Next up, I am going to take the 3rd PCB I got in the set and build it out as a 3.6V 8MHz unit with no power regulator. I'll hook that up to 2.4GHz Transceivers, and start running tests to see how long they last, and who fails first. (Radio or uC. The datasheet says the radio goes first.)

Thanks for the pointer to your blog, I am going to dig through it!

I'm working with some low power stuff too, have you ever considered using a DS3231 RTC and sleep mode on the ATMEGA? That RTC battery can last years and it can also generate a wake-up signal to the chip..If you are just using this set up to occasionally transmit data, well, you don't really need the chip awake until you are ready to, right?

We would have to do some research on the RTC to see the idle current and if it would trigger ATMEGA interrupts from a CR2032 :wink:

I want to build a drift buoy to send off to see if it could make it across the pacific, using ocean currents at particular depths as a conveyor and pressure sensor to determine depth. I will have a 'report' routine where it will surface and get a GPS location, charge the batteries with a solar cell, and transmit its data points and location to me using ULF. Obviously I want the thing to be using little-no power when it isn't collecting data or transmitting...

I will probably lose 90% of these buoys- but not before hopefully getting some transmit data!

I want to build a drift buoy to send off to see if it could make it across the pacific, using ocean currents at particular depths as a conveyor and pressure sensor to determine depth. I will have a 'report' routine where it will surface and get a GPS location, charge the batteries with a solar cell, and transmit its data points and location to me using ULF. Obviously I want the thing to be using little-no power when it isn't collecting data or transmitting...

Now that is what I call an ambitious project. The part that would scare me as not likely is the 'call home' using ULF (ultra low frequencies). I came across some ULF systems in the military and they took pretty high power and very long antenna. Certainly it's possible as that's how US nuke subs got their launch orders even while submerged, but it's a pretty expensive power consuming method to 'call home'. I think most applications like yours use satellite uploads to send their data, but that also has expensive and complexity.

Good luck, and keep us posted on progress.

Lefty

Ambitious indeed! That's a great project. I bet vlf won't do the job tho. Transmitter sites are measured in miles, and as noted take a ton of power.

I do have a ds3234 in fact, and it would work to wake. But the board I got cost like $20, so it is completely out. I am counting on watchdog timers to do the job for waking. Or i can use my rf radio which has an irq line too.

I have been doing some experiments on this. Stacking 3 batteries will not do the job for my RF radio. Though the batteries are rated at "1.2V", they deliver close to 1.5V at the start, and then plateau at 1.285V for much of their life. So they are pushing 4.5V in the beginning and 3.85V for a long time, which is outside the rating of my radio (3.6V max, 3.0V typ).

Using a 3.3V MCP1700 regulator, as recommended elsewhere on this forum, seems to do the job. They burn ~10uA, which is effectively zero for my uses. It also reduces the draw on the batteries when the unit is transmitting, which is beyond my electronics understanding to grasp why. They provide 200mA which is WAY more than I need.

The closest graph I found on the net that matches my observations is this one, specifically the 200mA graph. I am actually pulling 13mA when transmitting, so I'd expect something a little higher up the Y axis still.

The good news is that the NiMH batteries fall off quickly at the end, so my original concern of wringing the last mA's out of near-dead batteries isn't a concern. They deliver quality voltage right up until they pass out.

Thank you for the follow-up.