Easy way to use Li-ion cell (18650) for 2.4-3.6V circuit

[GiantGnome]

Hi people

I am wondering how to hook up a STM32F030 microcontroller (2.4-3.6V) to run off-the-grid, expecting a maximum current of 150-200mA.

When using AVRs (picoPower variant), I just hook up 3 AA-cells where the AVR is happy running from 4.5V all the way down to 1.8 (which should have used almost all of the energy in the cells).

I can't use the same scheme with the STM32F030, where it will have too high a voltage and not enough at the bottom. My first thought was Li-ion batteries. I have tried running an AVR with Li-ion, but that was with a cheap 5V booster module.

The 18650-cells I have lying around has a max voltage of 4.2V, and should be depleted at around 3V. Given the proper protection circuitry, what is the easiest way to use the module and not blow up the microcontroller?

I have considered the following:

• The fancy way: Some sort of buck-boost topology, getting a constant 3.3v. See
• The old-fashioned way: A 2.5V LDO regulator
• The risky way: Putting a diode in series with a forward voltage at around 0.6V

The fancy way is fancy. Expensive in parts, and would probably take up a lot of board space. A middle ground would be a boost circuit to about 4.5V and then a 3.3V LDO regulator.

The old-fashioned way (at this time my favorite). Everything works fine, but can't reach 5V TTL logic levels.

The risky way. Seems too easy, and could create strange situations at the low level, as the forward voltage rises with current.

What do people do?

[GiantGnome]

Ah! This should probably have been in the beginner section. Should remove it and repost, or could someone move it? Of course, it is microcontroller related, but it seems to be beginner stuff 

[Miles Teg]

I would use 2 cells in series to get 7.4-8.4V range. Along with a simple buck converter which, IMHO, take a reasonably small space.
But I don't know enough your constraints yet.

But does only one cell 3.7-4.2V not enough margin to work with a buck converter?

[tszaboo]

What do people do?

Depends on the application^TM
If 99% you are in sleep go with linear. If it runs on battery full power driving a screen and stuff, do switching. Dont do it with the diode, that is just not proper engineering. a voltage rail should jump around like it is a mercan rap star.

[ralphd]

Use LiFePO4.  They're available in AA (14500) size, are reasonably priced, and have a quite flat 3.3V discharge curve.

[JohnSL]

Use LiFePO4.  They're available in AA (14500) size, are reasonably priced, and have a quite flat 3.3V discharge curve.

Oooh, those batteries are exactly what I need for a project I'm working on now. What type of charging circuit is required for LiFePO4 batteries?

[ralphd]

Use LiFePO4.  They're available in AA (14500) size, are reasonably priced, and have a quite flat 3.3V discharge curve.

Oooh, those batteries are exactly what I need for a project I'm working on now. What type of charging circuit is required for LiFePO4 batteries?

You'll need a special charger.  Both DX and Fasttech sell the batteries and chargers.
I suppose you also could make a CV 3.6-3.7V charging circuit.  Or go ghetto with a 500ma USB charger and a couple 1N4007s to drop the voltage down.

[TJ232]

For 18650 Li-Ion cells (beware, unprotected!!) you can use a simple charger with protection module like this one - 18650 Li-ion Charger/Monitor with protection  + a LDO for a quick & quite decent 3.3V solution

[mikerj]

Use LiFePO4.  They're available in AA (14500) size, are reasonably priced, and have a quite flat 3.3V discharge curve.

Oooh, those batteries are exactly what I need for a project I'm working on now. What type of charging circuit is required for LiFePO4 batteries?

You typically use a constant current/constant voltage profile a bit like a standard lithium cobalt oxide (LiCoO2) cell, but the fully charged voltage is lower (about 3.6v) and LiFePO4 is much less fussy in terms of overvoltage.  It's a much safer chemistry all round, but they don't have as high energy density as LiCoO2.

Tye Microchip MCP73123 is a fairly cheap single chip charging solution.

[KL27x]

I would absolutely use a diode to drop a wee bit of voltage.... if it worked for the application. I would probably use a boost converter to 5V + 2.5V linear or buck for the MCU, if I needed it to interface with 5V TTL in someone else's hand halfway around the work.

Depending on what the app, I might even use a diode to power the MCU from the battery and a 5V boost just for the TTL rail. FVD goes down as current increases. But the output voltage of a fresh Li goes down somewhere between 4.2 and 3.6V as current increases, too. So that's a little bit of coutnerbalancing. And depending on what your circuit is doing, this might not be a concern, at all. A tiny filter cap, your normal bypass cap, and your power rail might be none the wiser. For minimal power consumption, this would allow the MCU to cut power to the boost converter when it's not needed and then you're down to w/e the sleep standby is for the MCU.

Depends on how much electrical testing you are willing to do on your circuit to make sure your circuit is going to work If you're making one, just go all out and make it grand. If you're making thousands, it might be worth it to see if you can make it cheaper and still know it is going to function. That will probably take a little more work looking at electrical specs and doing real world testing. You might take 20 of these MCU's and test them with a 4.3V low impedance power supply + diode to make sure it handles max voltage. Then hook up to the actual battery and increase the impedance to make sure they would still work with a weaker/smaller/worn-out battery. For a 18650 li ion cell, putting a 1 ohm resistor in series will increase the ESR well beyond what you might ever normally expect. But you can go as nuts as you want. Stick a 10-20R in there and see what kind of nonsense it will handle. 
 

[Flanbix]

If you don't mind spending a bit on the PSU, have a look at the LT1613 for example.

[Maxlor]

I'd use a PAM2305. It's a switch mode DC-DC converter that works down to (almost) Vin=Vout. Get the 2.8V fixed voltage version then connect any battery that's between 2.8V and 5.5V and you'll be fine. An LDO would work too.

Or if you want to go for 3.3V, you can assume that an lipo battery is pretty much empty at 3.5V@200mA, the energy remaining is neglible. You're still in LDO territory at that point. You really don't need to design down to 3V.