Using LiFePO4 Single Cell for On-Board Battery Backup

[Evan.Cornell]

I'm looking at using a single LiFePO4 18650 cylindrical cell for on-board battery backup for a system that typically needs 10-20W power.

Something like this: https://www.batteryspace.com/lifepo4-18650-rechargeable-cell-3-2v-1100-mah-49-5a-rate-3-52wh.aspx

The battery only needs to keep the system running for ~10minutes to ride through any power blips from the main power source.

Obviously a single cell doesn't have any sort of protection circuitry, so I'll be adding that to my board to prevent damage.

However, this application is essentially keeping the battery charged all the time, so does anyone have input on best practices to maintain the cell lifetime?

[Miyuki]

LiFePO4 Can be kept on constant voltage all the time, about 3.6V without any issue
Or are available other chemistries of backup cells, with even simpler charging, designed to be kept at full charge for years, but with just a few hundred cycles
LiFePO4 tend to last really long

[SiliconWizard]

Using LiFePO4 for a back-up solution would have my votes a hundred times compared to Li-ion (there were other threads about this.) They are much safer, can take a lot of beating, and are rated for a much greater number of charge cycles. Downside is, of course, a lower energy density.

The charging process is very similar to that of a Li-ion, only the voltages are different. There are many charging ICs you can use to charge LiFePO4 batteries. But yes, they are a lot more forgiving than Li-ion ones.

I read the datasheet of the one you linked, and it is unclear from it if it does embed a protection circuit or not. But it does list a number of safety tests showing that it can handle pretty severe beating with no adverse effect. I'd actually be surprised if it didn't embed a PCM; Li-ion batteries of the same format do. (It's a small circular PCB right beneath the + pole.) If in doubt, you can ask the manufacturer.

[thm_w]

Here is the previous thread: https://www.eevblog.com/forum/projects/float-charging-lithium-ion-batteries-at-lower-voltages/

3.6V might be ok, depending on ambient, that is around 90-95% SOC. I think you are cutting it close with a single cell though. 20W * 10min = 3.33Whr, and the cell holds 3.5Whr.
Depends on how confident these numbers are.

This paper tested 100% SOC sadly not any intermediate values though: https://uwspace.uwaterloo.ca/bitstream/handle/10012/12177/Catton_John.pdf
But you can see 100% is terrible for life.

[SiliconWizard]

Yes, keeping them at the max voltage (100% SOC) at all times isn't good, even for LiFePO4 batteries.

A battery as a typical back-up solution will sit most of its life not being used. This is just equivalent to "storage" conditions. The datasheet of the cell should give you a hint: "storage performance" is given at 25°C with 50% SOC. It's reasonable to assume those conditions would yield the best figures. 50% SOC on average is what is often advised for Li-ion batteries as well for storage.

Capacity recovery is already "only" 85% at 50% SOC storage after just 6 months. The same with 100% SOC won't look pretty.

Possibly you can stretch that a little. Having read a few papers, 50% to 60% SOC seems the sweet spot. Of course that will shave off a good chunk of capacity in case of power loss. But my guess is that if you settle on LiFePO4, which are significantly bigger and heavier than LiPo batteries for a given capacity, then your application is not critical in terms of weight/real estate. So just pick a battery with more capacity, and ideally do not exceed 60% SOC when the back-up system is "idle".

[Evan.Cornell]

What are advised strategies for keeping cells at a particular % SOC? Use a standard charger IC but set to a lower voltage threshold?