How far can LiIon be discharged?

[size=+3]How far can LiIon be discharged?[/size]



I do often see a question about how far LiIon can be discharged and if it is safe to charge a discharged LiIon. These questions can be fairly difficult to answer precisely, LiIon manufactures do not publish much about it and LiIon is not just LiIon, there are many different chemistries with similar but not equal specifications.



[size=+2]How to measure the discharge cut-off voltage?[/size]

Some people do not wish to discharge below 3.6 volt (or something), but why stop there when the specifications say 3.0 or lower?
There is some sort of reason for this and it has to do with how you measure voltage: Usual voltage is measured with the battery out of the equipment, i.e. the battery is unloaded. Here is an example:



Here I discharge a cell to 2.5 volt with 2A, when I stop the voltage jumps to 3.3 volt and it jumps fairly fast.



Zooming in on the stop, the curve shows 3 minutes. It takes 12 seconds for the voltage to raise from 2.5 volt to 3.1 volt. This means that if you take the battery out and measures it you will never see the 2.5 volt.



Lets try with another cell.



This is also a 3 minute view, the voltage raises from 2.5 volt to 3.1 volt in 3 seconds and in 7 seconds it is about 3.2 volt. Again it is not possible to measure the voltage where the equipment turns off.

How far the voltage jumps depends on the discharge current, a lower current will give a smaller jump and on the battery chemistry, different chemistries gives different voltage jumps. This means that measuring voltage with the battery out of the equipment will not give a correct indication of the minimum voltage the equipment discharged the battery to.

Due to the above jump in voltage it is rather difficult to know what the actual battery voltage is inside the flashlight when it is on and the safe way is to assume it is empty when the unloaded voltage is below some value. The 3.6 volt is fine for some batteries, but it is possible to go lower with many modern cells as seen above. The table at the end of the remaning capacity article can also be used.



[size=+2]How far to discharge batteries during normal use?[/size]

The first answer to this is easy, check the datasheet:



Here are some values from 4 different datasheets. The most common values are 2.75V and 2.5V today.

These values are the lowest voltage that is allowed on the battery if the full lifetime must be reached*. This means that for any normal usage this is the lowest value to discharge the battery to. This voltage do not include any current, i.e. the limit is not one that can only be discharged to at high currents, it is allowed to discharge with a very low current to this specification.
It is not within specifications to discharge below this voltage, just because the battery will recover to above this voltage when the load is removed.

*Usual, the 2.0 volt specification is not used for lifetime tests on that cell, there the lower limit is 2.5 volt.


[size=+2]But my over-discharge protection do first trip at 2.3 volt, is it faulty?[/size]

Usual no, protection is normally not used, i.e. it is allowed to go outside the range for best lifetime.



Here is some data from Samsung, the over-discharge protection is supposed to be outside the normal usage range.



And if I include a bit more of the table it can be seen that it varies with application. At high current the protection can be at a lower voltage.
From this table it looks like it is acceptable to discharge LCO/LMO (Real ICR/IMR) cells below minimum value, when it is done at high current, but with the more modern cells it is best not discharging too much below rated voltage.



[size=+2]How low voltage is safe to recharge?[/size]

Here are some illustrations from LiIon handbooks:





In both illustrations there is:
Operating range, that is the voltage range from the datasheet.
Protection/safety range, that covers a wider voltage range, but may add extra wear to the battery.
Keep out/unusable range, that is very low or very high voltage and the battery is dangerous.



The Samsung guidelines also list a minimum voltage.

These 3 items gives the minimum voltage as 1.5 volt, somewhere below 2.3 volt and 1 volt. This covers different brands and different chemistries.
Another importance detail is pre-charging, when the voltage is low the battery must not be charged at full current. A reduced current must be used.
When a battery is discharged to this low voltage it may lose some capacity and will lose some lifetime.

The reason to not charge very discharged batteries is because the chemistry breaks down at low voltages, depending on how much the chemistry has degraded the battery may explode when charged!



[size=+3]Conclusion[/size]

With the above as basis I will recommend the following:
1) Do not use over discharge protection as signal for when to charge batteries, it might wear the batteries down faster.
2) Do not charge batteries measuring below 2 volt, except if you have data saying it is safe.
3) When cells are down to 3.3 to 3.6 volt unloaded they are empty according to my tests, when the load is a few ampere.
4) Just because the cell recovers voltage, do not make it advisable to discharge below the limits while current is on.

And remember that if a battery gets hotter than usual during charger and/or will not charge fully (i.e. charger will not terminate), it is dead and maybe dangerous.



[size=+3]Notes[/size]

This article uses excerpts from:
Lithium Ion Rechargeable Batteries Technical Handbook (Sony)
Lithium Ion Batteries Technical Handbook (Panasonic)
Pack Design Guideline (Samsung datasheet, I have done some cut and paste in the table)

More voltage jumps with and without load

Thanks for the info, very interesting!

Lowest i tend to go down to is 3.3v on my ecig mods, flashlights tend to be around the 3.6v-3.8v on average. Usually end up topping them off after set times for convenience and knowing i will be using again soon(work lights for example).

Thanks for the nice info.



EDIT: My cells are between 3.6v and 3.9v after use. A good thing about using 2 or 3 lights a night for my hikes w/ CAPO is that One light is never used more than 30 minutes so the batteries never get drained too much!

15 minutes of use w/ my M2Xvn gives a voltage of 3.89v W/ a 18650 LG HG2 IMR 3000mAh. :thumbsup: I always have 2 w/ me.A fresh one after 15 minutes!

30 minutes of use w/ 4 X 18650 K'power 3400mAh w/ my TK61vn V4 yields a voltage of 3.78v!

Good choice of topic. It would help to explain a bit about the reason for the voltage drop / bounce, i.e. say something about internal resistance, maybe even mention some simple cell models using resistor(s) and capacitor(s) (the motivation being that it is better to teach them how to fish than to serve them fish on a platter, i.e the better conceptual model that the reader has, the easier they can answer such FAQs on their own).

Thanks, HKJ!

The technical information you publish is some of the best CPF has to offer.

Gauss163 said:

It would help to explain a bit about the reason for the voltage bounce, i.e. say something about internal resistance, maybe even mention a simple equivalent cell model using a resistor and capacitor

Click to expand...

The idea of a equivalent circuit is not very good, the resistor changes significantly when the cell is near empty. In the linked article I discharge a cell in 100mAh steps, there it is easy to see.

KeepingItLight said:

Thanks, HKJ! The technical information you publish is some of the best CPF has to offer.

Click to expand...

+1
HKJ is a Bright Star
(Even if I have to study his posts, and still only "get" only about 50% of the content ...)

Can't say enough how valuable all your input is to the forum, HKJ! Fantastic post!

This should be a permanent stickie or tutorial. Definitely required reading material.

A lot of this information is difficult to find.

Thank you very much for all of your educational contributions here, and your extensive and ongoing testing [emoji106]

Agree HKJ is awesome!!!:rock::bow:

HKJ said:

The idea of a equivalent circuit is not very good, the resistor changes significantly when the cell is near empty. In the linked article I discharge a cell in 100mAh steps, there it is easy to see.

Click to expand...

There are many different equivalent models used for pedagogical purposes - some more complex than others. For the purpose at hand, that simple model suffices. Most textbooks on batteries present a variety of such.

For a simple example from a study in 2015 see this later post.

Below is a brief exposition on voltage drop/bounce, excerpted from a paper by one of TI's battery gurus: Battery Selection, Safety, and Monitoring in Mobile Applications, by Yevgen Barsukov, Texas Instruments. See also this later post which includes graphs from a recent study illustrating how highly accurate such simple models my be. Readers interested in more precise battery models can find them in most textbooks on batteries.

Thank you HKJ for this information.

From my own personal use I try to use my single cell batteries down to 3.5 volts under load by trying to check voltage after certain time and load usage. To do this I keep one or two extra batteries in easy reach so I can check voltage at a later time.

After an amount of usage I begin to see a trend of how load and time effect my own battery usage.

With the method I use I discard batteries that do not charge back to at least 4.19 to 4.18 volts. Battery prices for quality 18650 flat top batteries are found at a reasonable price, so the expense for replacement is within my budget.

There is much to learn about lithium ion batteries and every user should be making an effort to increase their understanding.

Very nice explanation of a complex topic.

I always enjoy reading your posts and reviews. They are always informative and educational.

Thank you,

Bob

The NiteCore TM36 and others are able to show voltage under load, though it is with multiple cells.

Interesting and relevant post. A Big Thank You.

Gauss163 said:

There are many different equivalent models used for pedagogical purposes - some more complex than others. For the purpose at hand, that simple model suffices. Most textbooks on batteries present a variety of such.

Click to expand...

As you have already shown with you post of a model circuit it is not that simple and it do not really represent the situation at the end of discharge (As this article is about).

Great information, very helpful! Never know that voltage will bounce back. Thank you HKJ!

Thanks HKJ, great information that's practical.

HKJ said:

As you have already shown with you post of a model circuit it is not that simple and it do not really represent the situation at the end of discharge (As this article is about).

Click to expand...

It is about as simple as one can get. If you think that there is a simpler way to teach these basic ideas then let's see it.

Why do you believe that "it do not really represent the situation at the end of discharge"? Please be more precise.

Gauss163 said:

It is about as simple as one can get. If you think that there is a simpler way to teach these basic ideas then let's see it.

Why do you believe that "it do not really represent the situation at the end of discharge"? Please be more precise.

Click to expand...

This article is not about how batteries reacts to fast load pulses and that model is valid from 1Hz to 1MHz (If you have the correct values to fill in). There is nothing in that model to handle the behaviour of the rising Ri when the battery is near empty. I do cover basics in batteries without complex schematics with a lot of unknown component values.