I have an application for a backup power supply, which requires guaranteed energy available, but very infrequently, think burglar alarm. I would naturally gravitate towards lead acid as it has a well defined float charge protocol. However my client wants to use lithium, probably for cost in the small size that's required.

My first port of call was batteryuniversity.com, and the various battery manufacturers. The BU papers 409, 808 and 808b cover charging and longevity of lithium cells.

The classic cyclic charge protocol is to charge to 4.2V, terminating when the current is around 0.04C, then let the cell decline to a lower voltage, before repeating. Keeping the cell at 4.2v would accelerate the cathode oxidation wear mechanism.

This protocol makes the cell's full capacity available at the time of the charge, but when it drops to a lower voltage, less capacity is available. The guarranteed capacity is therefore with respect to the lower voltage.

Improved lifetime can be obtained by using lower end-point voltages, and various sources mention voltages of (end_point/recharge_point) of 4.20/4.05, 4.10/3.90, 4.05/4.00. BU808 plots the highest lifetime as a cycle between 75% and 65% SoC, which equates roughly to 4.00/3.90. I'm not sure that's the highest possible, only the highest of the options they considered.

Where hi-rel is required, satellites for instance, NASA apparently charge to 3.92V, which is a compromise between a higher voltage which induces cathode oxidation, and a lower voltage, which loses capacity via SEI (solid electrolyte interface). However, this is still cyclic use. Dr Dahn (who's he?) suggests that below 4.1v, the oxidation mechanism is not significant.

Getting near to the question.

Nowhere have I seen (apart from a few questions on AllAboutCricuits or Instructables, which were not answered satisfactorily) a discussion about whether float charging to a low voltage is ever permissible.

Consider a 4.05/4.00 top-up protocol. Now compare and contrast to a 4.00 float charge.

  • the minimum voltage, so the guaranteed energy, is the same
  • the cell reaches a higher voltage with the top-up protocol
  • the cell is cycling with a top-up, but not with a float

That seems to me to be a draw, then two wins for floating, IF floating is permissible.

In one case, I'm supplying just enough charge over a long period in bursts to keep the voltage at 4v or above, in the other case I'm supplying just enough continuous current to keep the cell at 4v or above. Given that cell leakage is higher at higher voltages, and charge efficiency is less than 1, I would expect to be supplying less total charge over time when floating than with topping up.

Question - When you are well below the maximum voltage, say in the 3.9V to 4.0V range, is there still the prohibition about applying a small continuous current that there quite rightly is at 4.2v?

Question - different wording - When at a low voltage, does applying a lower charge continuously actually damage the cell more than bursting the charge, which involves a higher peak voltage and cycling?

  • \$\begingroup\$ Did you ever find a solution to this issue, or any off-the-shelf parts which help with it? \$\endgroup\$
    – Matthew
    Commented Aug 17, 2023 at 12:41
  • \$\begingroup\$ @Matthew Not really. It's good to see a 'big name' like LT saying the 4 V float prolongs life, which I've kind of assumed all the time. The application I asked about didn't materialise for other reasons, but I've floated LiPos at 4 V without any fires (yet). \$\endgroup\$
    – Neil_UK
    Commented Aug 17, 2023 at 14:39
  • \$\begingroup\$ @Neil_UK - Hi, When you say you've "floated LiPos at 4 V" do you mean it in the sense one would float lead-acid batteries i.e. that voltage (perhaps current-limited) is applied at all times? I'm asking because despite LT's repeated use of the word "float" in the LT4064 datasheet linked in this answer, that's not what the LT4064 does, AFAICT. It's doing a normal CC CV charge, but with a 4.0 V termination voltage and a 3.9 V threshold to do a top-up charge (see "Recharge" on DS page 10). Thoughts? \$\endgroup\$
    – SamGibson
    Commented Aug 17, 2023 at 14:54
  • \$\begingroup\$ @SamGibson For better or worse, yes, I float as in 24/7 constant 4.00 V applied to cells. This is for a mains standby lighting application. I don't recommend that others should do this, but it works for me. I don;t recall what the long term current drops to, but it's less than the total average current of a charge to 4.1 or 4.2 V, then a top up when the voltage has dropped to 4.0 V \$\endgroup\$
    – Neil_UK
    Commented Aug 18, 2023 at 5:53

2 Answers 2


The kind of details you are asking about are not appropriate to get from general information like here or even from Battery University. You need to talk to technical people from the company you will actually get the batteries from.

I was involved with the design of a battery backup system that could spend a few years just sitting there, then would have to cough up about 700 W on a few 10s of µs notice. We were in constant contact with engineers at the battery company, which in this case was A123. It turns out that there are no hard limits and everything is a tradeoff. However, the tradeoffs vary from product to product and certainly between manufacturers at the level of detail you are asking about.


I've found the MCP73841/3-4.1 which will charge to 4.1V instead of 4.2V, which might be a partial solution for projects with small power requirements. However from what you've said, 4.0V would seem a better fit for greater longevity.

LTC4064 seems designed for this use-case and charges to 4.0V.

  • \$\begingroup\$ Matthew - Welcome. As you're new here, please see the site tour & help center as Stack Exchange (SE) rules differ from a typical forum. || As might happen on a forum, you asked the OP a question, as well as providing an answer. Those are two different things on SE. Something written as an answer should (attempt to) answer the original question. Nothing more. \$\endgroup\$
    – SamGibson
    Commented Aug 17, 2023 at 13:46
  • 1
    \$\begingroup\$ (continued) If you want to ask the OP for an update, that should be posted as a comment on their question. You can write comments when you have 50 points (see "Why do I need 50 reputation to comment? What can I do instead?"). As an exception, rather than just deleting it, I converted that part of your post into a comment on the question. Thanks. \$\endgroup\$
    – SamGibson
    Commented Aug 17, 2023 at 13:46

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.