I am designing a product requiring high power and energy storage capability. I have narrowed down the selection to two (most cost effective) systems: a 3s system of protected 18650 lithium ion cells (11.1V), or a 4s system of unprotected 26650 lithium iron phosphate cells (12.8V)

Both systems would be charged with a stand alone battery charging IC, with a thermister safety mechanism. I am also planning on including a fuse that will disable the system with excessive current draw (10A) in the case of the unprotected LiFePO4 cells.

Which system would offer a better safety profile for the use with a consumer product?


2 Answers 2


Let's start with a very important point: With Lithium based batteries a thermistor is a nice last line of defence, but you seem to be making it the centre point of your thoughts.

If your thermistor triggers on a Lithium based battery, purely from charging, you might want to hook that up to an airhorn so you know to start running.

Do not get me wrong: I fully support having a thermistor in everything that's to do with bucket-loads of energy. But with Lithium based batteries, that's not the be-all and end-all of safety it was with NiCd/NiMH.

Hyperbole aside, if a Lithium based battery is hot enough to trigger a charging chip's thermal limits, your batteries are already being destroyed. If you want a safe consumer product, regardless the type of lithium, you need the appropriate charging scheme.

If you want it to be a quality product, just assembling a pack from protected cells is going to get you into trouble. While individual cells will likely not get damaged during its operational life, the operational life will probably be limited.

When an individual cell goes into protection it will open up the chain, where the other batteries are still "active". There's a number of scenarios when a manufacturer uses single-cell-spec'ed MOSFETs in the protection that will not turn out great for your batteries or their service life.

So, if you want to be taking yourself seriously, you need to build the pack properly, with the appropriate protection board. If you want to be actually safe, with a separate pack wired into electronics, this protection needs to be inside the pack to prevent damaged wires from causing short-currents.

This putting either choice on level ground, since protection boards can be had for your voltages for either chemistry, you need to weigh the advantages and disadvantages of each chemistry against your individual needs.

LiFePO4 will be much safer from self-combustion, even when hard-shorted, but not holy! They are still a bear you should not be intentionally poking. Although there's videos out there showing the pneumatic insertion of a nine-inch nail into well built, branded LiFePO4 cells with no fallout.

LiFePO4 also is much more resistant to formation of so-called dendrites on its lithium electrode, causing it to be less sensitive to short term or low-grade over-voltage.

On the other hand, LiFePO4 in most cases has no more than 3C discharge allowance and is usually advised at 0.5C charge currents. Of course, in Lithium-CobaltOxide based batteries ("Li-Ion") the very high discharge types will always still have significantly reduced life at these currents.

LiFePO4 is more expensive (at the moment) and has a lower storage density (for now) and identifying the right manufacturer in the more affordable cells can still be difficult. The fact that destroying such a cell is harder to begin with makes that even harder when samples are sent. That is, the grey area is much thinner, and if you test samples it's that grey area where your data points live.

However, the reason I stock shitloads of quality LiFePO4 and that they are used in high-end automotive, is that they have clear advantages, even if you need a much higher capacity to get the same high currents;

  • They have a lower self-discharge rate
  • They are a little less affected by residing in high or low charge states
  • They can be safely discharged below freezing (a major point for automotive)
  • They contain a somewhat reasonable amount of remaining energy below freezing, 45~80% for LiFePO4, versus 0~35% for most LiIon. Where the exact number depends on the temperature.
  • They have a relatively level discharge curve for the majority of the energy content.
  • Minor: Their voltage levels are compatible with many ARM type controllers.

All that said, I see you mention balancing nowhere. This also is less of a task with LiFePO4, but some plan for keeping the cells balanced should still be considered. Whether you bottom-balance, top-balance, full-balance, or active-balance that's your choice. For either chemistry. Or not balance at all, but again, usable life will be reduced noticeably.

  • \$\begingroup\$ Thanks for your very thorough answer, highly appreciated. One minor annoyance I have (perhaps this is merely ignorance) is that none of the stand alone charger ic's out there do not have integrated cell balancing. In fact, though I've heard of it's importance from many sources, the number of balancing ic resources that exist for 3s packs at all is pretty low. Is cell balancing critical for such a low cell count pack? \$\endgroup\$
    – John Evans
    Commented Dec 29, 2016 at 5:05
  • \$\begingroup\$ My primary concern is the safety profile, so while I do want to design a quality product, I would be willing to trade off charge cycles for a reduction in board complexity and cost. \$\endgroup\$
    – John Evans
    Commented Dec 29, 2016 at 5:06
  • \$\begingroup\$ One final update, since this is in the very MVP stage, I'm planning on buying some off the shelf boards from ebay for the balancing/protection and using protected li ion (LiCoO2) cells. I'll integrate the battery charger ic onto the primary board. Cost wise it makes sense doing it this way for the time being. Still less than 20 units sold, safety is my primary concern at this point. Cost optimization can come later. \$\endgroup\$
    – John Evans
    Commented Dec 29, 2016 at 5:22
  • \$\begingroup\$ @JohnEvans There are balance-charging and balance-protection chips out there. If you want to design everything yourself I'd suggest looking into a balancing&protection board inside the pack. I remember reading a product announcement years ago by Linear about their new stackable active balancer chip, which continuously balances currents through a switching set-up to always have all cells at the same voltage. May be expensive in BOM count though at high currents. I strongly advise you not use protected individual cells in a balancing/protection board setup. \$\endgroup\$
    – Asmyldof
    Commented Dec 29, 2016 at 9:47

I think the lithium iron phosphate system would be safer. These cells are used in transportation systems, and are more robust than some non-specific "lithium ion" (like the kind that blow up in laptops periodically).

However, if your getting into this level of detail, you really need to sit down and talk with application engineers from prospective battery companies. I've talked to apps engineers from A123, for example, that I got some good information from. They sometimes had to go ask engineers at the factory and get back to me. These are the kinds of conversations you need to have.

Batteries are very complex, and everything at this level isn't in the datasheets.

  • \$\begingroup\$ [Please treat this information only as an anecdote.] I've talked to A123 this summer about LiFePO4 cells. Their suggestion was that LiFePO4 are almost safe without a protection circuit. LiFePO4 are intrinsically safer than Li-ion, but still require protection. For a variety of reasons, I ended up using an off-the-shelf protected Li-ion battery pack. \$\endgroup\$ Commented Dec 29, 2016 at 0:07
  • \$\begingroup\$ @NickAlexeev I may end up going with the protected 18650 (Li-Ion LiCoO2) cells at this time. It appears that most of the failures that have happened to these cells happened under invalid charging conditions. So adding the charging IC w/ thermister should improve the safety profile dramatically. \$\endgroup\$
    – John Evans
    Commented Dec 29, 2016 at 0:37

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.