Practical effects of charging li-ion cells to different target voltages

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
I've done a simple study to demonstrate the practical difference a user would observe if their charger were designed to terminate at a specific voltage. I'd hoped to address a couple things by doing this:

  1. For my own interest, determine what the runtime impact would be for deliberately choosing to dial back charging to 4.10V in order to gain more cycles - lifespan - out of my cells.
  2. Help to put in perspective for new users of li-ion cells what the practical difference of a healthy cell coming off the charger at 4.15 or 4.17 volts as opposed to the 4.2V target many have assumed is the ideal target voltage while everything somewhat lower is somehow a charger failure.

The approach taken for running this test isn't especially rigorous but does look something like what normal user practice for charging cells would be, except that I'm dialling in specific target voltages.



A single AW2900 mAh 18650 cell, which has a Panasonic NCR18650 cell inside, was used for all runs. This particular cell is not new. The cell has been in regular use for about 8 months, and has been occasionally abused (deeply discharged).



Ive not made multiple runs at all voltages and averaged out the result but I have done several runs at some voltages to convince myself that the results Im showing here are typical, or close enough for my purpose. The cells were charged on a multi-chemistry charger capable of aiming for a target voltage and voltages were confirmed with a standalone voltmeter as well as the chargers likely more accurate (than my voltmeter) voltage readout, and were allowed to for 30 minutes after charging and after discharging before being recharged.



What I've chosen to do is mimic the current demands of a ~ 260 lumen flashlight, in this case a Malkoff M61 module. Off a freshly charged cell this drop-in consumes about 800 - 830 mA of current and as the cell is depleted (voltage drops) the current draw will increase. These discharge curves are therefore of a constant power nature to mimic what such a light will draw from a cell. I've chosen a 2.80V cut off for the test even though this cell can tolerate dropping to 2.5V under load, mostly because I wouldn't drain the cell that low in practice anyway.



The first chart shows the capacity extracted:

target-voltage-impact-cp-discharge-850ma.gif




The second chart shows the runtime to the 2.80V cut off in minutes:

target-voltage-impact-cp-discharge-850ma-minutes.gif




At this voltage cut off, or higher up along the curve (which would be a healthier place to terminate using the light) the difference in runtime is approximately 11 minutes. To put that in perspective at the end of the test the cell charged to 4.10V would have run the theoretical flashlight for 2 hours 30 minutes while the cell charged to 4.2V would have run for 2 hours 41 minutes.



Aiming for a lower state of charge target is known to increase cell life, dramatically so depending on where you aim for.



The bottom line: rejoice if your charger doesn't try to hit 4.2V. Your cells are going to last longer, and you aren't missing much.
 

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
NOTE: From here though Post #8 was selectively hand recovered from Google Cache.

This chart includes an additional discharge trace (cell charged with an upper voltage threshold of 4.00V). Vertical dotted lines at 4.00V, 4.10V and 4.2V indicate the non-linear returns of higher and higher upper voltage threshold as it relates to extractable capacity.

target-voltage-impact-cp-discharge-850ma-minutes2.gif


Added a second set of lines showing non-linear additional capacity at various states of charge when cell discharge stopped at 3.4V or 2.8V.

I've read various figures for the gain in cycle life (all other things being equal) obtainable if cells are kept under a certain upper voltage threshold but it isn't clear to me that these figures (as much as 30% if kept < 4.00V from one source) are relevant to today's crop of high quality cells. Does anyone have links to more up to date estimates or research?
 
Last edited:

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
Cycle life observed as cells cycled within state of charge bands

In my quest to understand opportunities for increasing the cycle life of my cells, I've been pondering over what the optimum operating state of charge should be - i.e. upper and lower limits.

We know that li-ion cell cycle life is markedly degraded by overcharging them significantly above 4.2V. We know they can be made unsafe by over-discharging them significantly under < 2.5V under load.

Common wisdom, which might be truth or might be folklore, is that ~ 40 - 50% state of charge (roughly 3.8V depending on cell model) is optimum for long term storage. This may be an optimum balance between cycle life and having a cell that can be put back into service right away or brought up to full charge in a reasonable length of time.

What about discharging cells to, or storing them at, very low states of charge - even approaching 0 - 10%?

The radio control folks avoid keeping their packs below 3.7V for any appreciable amount of time to avoid lost capacity. As some vocal RC proponents of this strategy are well equipped with test equipment and have performed cycle tests I'm not going to argue with them although we should recognize there are chemistry differences between their lipo packs and our cylindrical cells to take into account.

Are our typical cylindrical li-ion cells damaged by regular visits to or near a 0% state of charge? Can we store them near there?

This US Army Study (PDF) I've run across today shows some impact on cell ageing tied to the relative state of charge. Their study modelled four bands of use: 0% to 100%, 50 to 100%, 25% to 75% and 0% to 50%, representing "Full" use, constant "top-off" and constant drain scenarios. The following chart shows the capacity impact over hundreds of cycles.

us-army-soc-capacity-study.gif


Looking at the individual discharge charts in the PDF document -- Figures 4, 5,6 -- as compared to Figure 3, it would appear that the portion of time spent near 100% state of charge might very well account for the additional degradation seen in the 0-100% / 50 - 100% series as compared to the others.

Their observations seem to put lie to the notion that li-ion cells "like to be topped up" rather than deeply discharged. Maybe we should be saying "li-ion cells don't like to be topped up or fully charged".

Any thoughts?
 
Last edited:

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
Re: Cycle life observed as cells cycled within state of charge bands

Originally Posted by jasonck08
Doesn't this other US Military test contradict the test and article that you linked to?

Does it? Maybe they are not mutually exclusive.

The US Army test I linked to is as far as I can remember the only test I've seen that has taken a state of charge -- depth of discharge banding approach to evaluating cycle life at state of charge voltage ranges other than the upper-most band.

The US Army test's upper voltage threshold (presumably only for the 50-100% test band) was 4.2V while the Space Command test's upper voltage threshold was 4.1V, so automatically Space Command's cells accrue some benefit from not being stressed as much near the end of a charge cycle.

I wouldn't want to extrapolate the US Army tests out from 600 cycles to 60,000. That means someone needs to test the scenarios out for longer. When you get to 60,000 cycles let us know how it turned out. :)

SilverFox said:
Hello Tandem,

The key to long life is the amount of time spent a higher voltages.

You will notice that the 25 - 75% range also did very well.

The problem with flashlight use is twofold. On one hand people want to grab a light knowing that they have the most run time available, and on the other hand we want to avoid the damage that comes from over discharge. This has driven the recommendation to frequently top up the cells.

Another problem is that most people don't use chargers that allow flexibility in charging. Hobby chargers allow some flexibility in that they allow for various charging rates, and terminating at 4.1 or 4.2 volts. If you review charge graphs you find that at higher charge rates, more time is spend in the constant voltage phase of the charge. This adds additional stress to the cell.

The idea of keeping a cell at 40% for storage comes from industrial use. This level of charge allows for some parasitic drain over a 1 - 2 year storage time.

An interesting design concept for flashlight use would be to have a hard cut off that eliminates over discharge, then only put 25% into the battery for use. While not practical, this arrangement should give you extended cycle life.

By this do you mean parasitic drain from a circuit with micro ampere draw, or are you referring to self discharge?

Would it be fair to characterize the 40% storage charge level as having been somewhat arbitrarily picked? If so perhaps this was because self-discharge over the long haul at different states of charge wan't well understood? Your own study on self discharge, plus some of the others quoted here recently and in the past, seem to suggest that at anything other than a full to the brim SOC, self-discharge is very low.

In your last thought what would you consider an appropriate hard cut off to eliminate over-discharge? What I'm really trying to get at is whether you'd consider discharge on load to the cell's designed bottom edge (say 2.75 or 2.5V with some cells) to be the appropriate 'too deep' point, or would somewhere within the range of 3.1 - 3.3V on load (current matters of course) springing back .5 - .75V off load to be a good spot to target? Would regular use to the bottom edge result in shortened cycle life as is the case with time spent at the top edge (4.2)?

Approaching this from a different angle -- the 0 - 50% band and 25 - 75% band in the Army Study both result in only gradual declines in available capacity with no overt leader either way. Do you see anything in that study that makes you question the conclusion I'm drawn to - that it almost doesn't matter what you do [1] as long as you leave some distance away from 100% state of charge?

It would have been interesting to see a 0 - 75% band - I'm guessing that it would look not much different than the other two top performing bands of use.

Mike

edit
[1] Avoiding out-sized current draws and avoiding crashing through the design bottom edge assumed
 

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
Mr. Happy said:
When considering cell storage, what matters from an electrochemical perspective is the stable open circuit voltage when the cell is in equilibrium. For a typical lithium cobalt ion cell, this voltage at 0% SOC is about 3.7 V give or take. The voltage you terminate at under load almost doesn't matter as long as the stable unloaded voltage returns to about 3.7 V after you remove the load.

If you also consider the rule of thumb that I think was mentioned above, that each 0.1 V decrease in storage voltage leads to a halving of the deterioration rate, then dropping from 4.2 V to 3.7 V would represent a 2^5 = 30x decrease in deterioration.

Storing at 40% is a kind of arbitrary compromise indeed, since if you drain a cell below 3.7 V the voltage will drop off a cliff and you need some buffer of charge to prevent that. Most lithium ion packs in actual products have electronic circuits attached and these always have a small parasitic drain.

SilverFox said:
Hello Mike,

The discharge comes from the attached circuits. The self discharge rate for the bare cells is very low.

I am not sure it was a totally arbitrary decision... I think some marketing also came into play. When you bring home your new laptop computer, isn't it nice to be able to turn it on and show it off a little before having to recharge it?

The RC people have spend a lot of time and money looking at this. Now they are using Li-Po packs, but there are a lot of similarities with Li-Ion. They seem to have good results limiting the discharge to 3.3 - 3.5 volts expecting the voltage to rebound immediately to a little over 3.7 volts. Now they are using higher discharge rates, but I think their ideas are worth looking at.

On the practical side of things, I have had very good luck charging to 4.1 volts, and recharging often. It is my understanding that in recent years tremendous steps have been made on reducing the oxidation inside the cell. It seems that the electrolyte mix can be adjusted to reduce the effects of oxidation. This may also be an important difference between the bottom of the barrel cheap cells and the premium name brand cells.

Tom
 

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
Hi Tom,

I've been following some of the studies the RC folks have done, partly in preparation for a RC project one of my sons and I are working on together. Aiming for a rebound voltage of ~ 3.7V / cell certainly has become very much a part of the culture there but I've either not seen a definitive cycle test on how they came to arrive at this conclusion or have misplaced or never made a bookmark to it as is my habit. If/when I find something on this I'll link it in here.

They certainly drain their packs at rates flashlight enthusiasts rarely go anywhere near. 15C. 20C. 45C. Amazing. Of course this changes the off-load response too, they get bigger rebounds after draining their packs for a dozen or so minutes than I do draining a cell for 2 hours at .3C even if we end up more or less in the same spot capacity wise.

They also have more fires.

As a result of this thinking process I have also adopted 4.1V for an every-day charging voltage maximum. Picking the end spot will not always be up to a voltmeter of course but the task at hand.

Would you think that as a rule of thumb a best practice would be, at any rate of discharge, to terminate discharge well short of the voltage cliff found at the back end of cell capacity? I'm thinking that one could pick a point say a good 10% in front of the cliff when discharge is still somewhat linear? At flash light rates of drain, stopping there might not see a cell rebound all the way to 3.7V, but in a number of RC oriented discharge (and rebound) studies it appears that is more or less the point along the curve where they are stopping their discharge in order to get the voltage rebound they want. In the process at these higher rates they are also getting the lions share of pack capacity.

At typical general purpose flashlight (~ .3C or less even) rates of current draw a 2900 mAh cell charged with 4.1 max charging voltage might only deliver 1000 - 1300 mAh (seeing this in testing too) if aiming for the 3.7V rebound was the most critical element - only about half-way through the discharge curve. Intuitively it feels like one should be able to take the cell lower at these relatively low rates of drain, but maybe that's bad intuition.

So - is staying well back from the cliff, at any rate of draw, the key thing? Or perhaps just a good rule of thumb?
 

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
jasonck08 said:
The LiPo guys try not to take their cells under 3v under load. The Panasonic 2900 cell which adopts a nickel-based cathode is designed to be discharged down to 2.5v under load. If I discharge a Panasonic 2900 @ 2C to 2.5v, it will rebound to about 3.2 or 3.3v. LiPo guys discharge at much higher C rates (but their cells have lower internal resistance, but the voltage rebound should be about the same, so 3.7v sounds about right).

I will continue to charge my cells to ~4.2v, as I know how many cycles I put on my cells per year and its not much.

I mean seriously, ask yourself how many times you actually put a full cycle on a 18650 cell. I for one probably only put a full cycle on a cell a month. 12 cycles a year. I don't see a point in only charging to 4.1v. In 5-10 years the cells will be old and the capacity will be very outdated. I'll keep them for maybe 3 years until the next generation cells comes out with capacities in the 3.4-3.6Ah range.

What I do wish I could do is tell my laptop to only charge the pack to 4 or 4.1v, as if I am able to double the cycle life on my laptop pack that would be nice.

A fair question. I wouldn't care much about the topic if I mostly had shelf queens or was into flashing a bright light to impress friends once in a while. But this thread isn't just about my usage though. A working person using a light on shift would go through plenty of cycles per year. They could to 4.1 and be done with it if sufficient capacity remains for the runtime they need.

I've got more lights than I need to have so in some respects the most important bit of information for me is what to set max charging voltage at so I'm not unnecessarily cooking cells that are sitting idle in some our lights.

But we have other lights that get used much more frequently. When I'm on call with the fire department my EDC lights get a workout on a more regular basis. Play time however is where we use them the most.

Depending on the time of year, if I'm doing distance training on my bike I often ride at night for a couple of hours at a time and generally drain 1 cells fairly fully, this might grow to 1 1/2 cells if I limit my usage per this thread. In the most recent fall I was riding 3 or 4 nights a week, two hour sessions, in the dark. In the winter I don't ride nearly as much but it is dark more. My wife cycle commutes more or less year round except for the worst of our weather here so there's another set of cells getting a regular work out in the fall and winter. Between the li-ion and NiMH cells there seems always to be something on a charger here.

I've got two different use cases for the biking lights; one is the known duration regular training; if I can get extra cycle life by swapping cells three times during a ride rather than two I don't mind the effort. The other is virtually all night riding that happens during distance riding events and since these I'll only do a few a year, I'm more concerned about max runtime and minimizing weight / bulk than the impact on the cell for a few of these events a year.

Re the laptop, I hear you. It is possible that some of our bike lights at some points in the year see as much cycling as a laptop does -- or more, particularly so if your laptop is connected to a docking station or line cord as often as ours are.
 

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
SilverFox said:
Hello Mike,

Your plan sounds good, but keep in mind that there is a lot about these cells that we may not completely understand yet. The "gray" areas are not well defined, and may vary from manufacturer to manufacturer.

For example, the general consensus is that we should limit the discharge to somewhere in the 2.5 - 3.0 volt range to avoid doing damage to the cell. In order to explore this, I ran 125 cycles on a 2400 mAh premium cell. Charging was done at 2 amps, discharging was done at 5 amps and the low voltage cut off was 1.00 volts.

The cells did show a drop in capacity. They started out at around 2300 mAh and ended up close to 2100 mAh, but there were no other issues with heating or other signs of damage.

I haven't published this study because I really don't know what to make of it. I don't recommend using a cell like this, but the results of this test were definitely interesting.

Tom

That would make for interesting reading Tom.

---------

It isn't clear to me whether it is the higher charging voltage or amount of time spent at higher voltages which puts more stress on the cell but it seems likely that the latter at least has some impact given storage at elevated voltages is known to reduce cycle life. Maybe someone has published a study that answers that question.
 
Last edited:

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
Re: Cycle life observed as cells cycled within state of charge bands
Written by MikeAusC on 02-10-2011 05:39 PM GMT

I used to believe the literature which consistently says that if you don't want to throw away your LiIon cells after two years, you MUST -

- store them at less than 40% state of charge

- store them at a low temperature.

My Toshiba laptop has been permanently switched on and connected to mains power for 99.9% of the last FIVE years - so the cells have been charged to 100% - and stored at high temeperature.

The battery pack still provides a great runtime - so now I'm storing LiIon cells at 100% charge at room temperature.

Re: Cycle life observed as cells cycled within state of charge bands
Written by Mr Happy on 02-10-2011 08:17 PM GMT

Remember that the battery packs provided with brand name laptops contain very high quality cells and also contain an intelligent charge controller. Your loose Li-ion cellsmay not match the performance of your laptop--but they might of course, depending on where you obtained them.

Re: Cycle life observed as cells cycled within state of charge bands
Written by on 02-10-2011 09:35 PM GMT

Although there haven't been as many fatalities in the last couple of years in my office, I've killed enough laptop battery packs single handedly over the years that I'm fully in the camp of believing full charge all the time == bad juju. Dell, NEC, IBM/Lenovo - didn't seem to matter. Maybe Toshiba is using a better charging/battery management circuit than was in my laptops.


Electric autos apparently do not push cells to the extremes in order to gain on cycle life. NASA does the same. It seems logical that both the aerospace and automotive industries would have done rigorous testing given the stakes of failure. Replacing a dead cell in a pack in space might be very costly or impossible; in an auto very costly; in a six cell almost consumable laptop battery pack - neither that costly nor difficult.
 

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
Re: Cycle life observed as cells cycled within state of charge bands
Written by shadowjk on 02-11-2011 12:06 PM GMT

malow said:
this week, i found that every nokia phone i have charge until 4.15v... so im starting to charge until 4.15v too ;)
I've studied some Nokia phones' charging circuitry and algorithms in detail. I've studied some Nokia phones' charging circuitry and algorithms in detail.


Schematics leak and can be found on the internet, and for some units the diagnostics software has leaked, provides about 40 rows of data from the battery management software, and for some the datasheets for the charging chip is available and through rooting the software I've been able to study what it does...


In the end I ended up writing my own driver for it, for proper CC/CV :)


Anyway, one design feature on some phones is a 128mOhm resistance between charger and system+battery. The charger is feeding energy both to system and the battery, so it can't actually distinguish between system load and battery charge current. The voltage drop over the 128 mOhm resistance drops the charge voltage somewhat if system load is high. Basically, the more stuff the phone is doing while connected to charger, the lower the charge voltage will be.


This also makes it unable to properly terminate, if system load is high. In that case it ends up trickle charging at a 4.2 minus the drop from the current over .128 ohm.


When not affected by these issues, the charging system terminates at 4.2V. On a fresh battery, with system load of about 20mA from me having data logging software running on the phone to capture what's happening, the voltage drops off to about 4180. WIth older batteries it drops off lower.


On some phone models charge is started again at 4.1V, and terminates at 4.2.

On some phones, the charge is restarted again at something slightly below 4.1V, and then it terminates at slightly above 4.1V, effectively floating the battery around 4.1V through small microcycles.


On all models, once it has during charge plateued near 4.2V once, the user interface will switch from displaying charging to displaying full. This can happen long before the charging is actually terminated (if it terminates). There's no indication in the display of the microcycling, or top-up charges. You basically don't know if it has just terminated at 4.2V, if it has been sapping it down to slightly below 4.1, or if it's microcycling around 4.1..


The reason I wrote my own driver, was that I wanted to compensate for the voltage drop caused by the .128 Ohm resistance after the charger chip output, before the system+battery. I access two hardware chips directly from my program, with one I can measure the voltage at the battery, and the current balance of power going in/out of the battery. The chip is an integrating fuel gauge type chip, so I get pretty accurate current consumption/charge data, as well as voltage to the nearest 4mV. Ohms law, current going into battery over .128 ohm (I ignore any potential consumption by the system load), tell the charging chip to increase it's CV voltage to compensate.

This has made the CC/CV curve look much more proper, with the voltage climbing all the way to about 4188mV (I don't want to get too close to exceeding 4.2) before current starts dropping off. With the default driver, current started decreasing even before voltage reached 4.1, making a graph of it look kinda similar to those of XTAR chargers and similar :)


And as a side effect, I can now also control whether my phone charges to 4.2, 4.1, 4.0 or whatever I want between 3.5 and 4.76 (and thus I somehow made my post vaguely ontopic to this thread).

Re: Cycle life observed as cells cycled within state of charge bands
Written by Mr Happy on 02-11-2011 12:29 PM GMT

That sounds fun. And as a bonus, if you have a bug in your program your phone might burst into flames :) That's not a penalty most programmers face for errors in their code :devil:
Re: Cycle life observed as cells cycled within state of charge bands
Written by on 02-11-2011 12:40 PM GMT

Sounds like an opportunity to createkiller phones. Coming soon to a spook shop near you! ;)
Re: Cycle life observed as cells cycled within state of charge bands
Written by shadowjk on 02-11-2011 03:39 PM GMT

The killer phone argument is why it wont be on any app stores :)


I have safeguards in very simple code (to make it as bugfree as possible) that checks BattV every 5 secs and aborts if voltage exceeds 4.2V, 30 seconds after my software commits suicide for exceeding the holy 4.2Volt, the charging chip resets itself when it has received no instructions, and goes into safe mode, where it will independently without software control or drivers bring the battery voltage to 3.7V, using a USB-safe charging current of 500mA. :)


Also, there's the battery protection circuit. It works pretty well. *cough*
 

tandem

Enlightened
Joined
Aug 5, 2010
Messages
874
Location
Vancouver, BC
Re: Cycle life observed as cells cycled within state of charge bands
Written by Tandem on 02-14-2011 11:11 AM GMT

Here I've tried to summarize what's been discussed in this thread, adding in some basic truths from all corners to round out anoverview on treating your cells nicely. Edits are on-going, constructive criticism and suggestions are welcomed.

Basic Truths

  • don't over discharge cells; don't leave them over-discharged if you do.
  • don't over charge cells; reducing the state of charge from 100% will help prolong cycle life
  • don't discharge at too high a rate
  • don't charge at too high a rate
  • don't charge at very low temperatures, or restrict the charge at least
  • avoid subjecting the cell to high temperatures, either from environmental conditions (like a hot car all summer) or from abusing the cell electrically from a way-too-high charge rate
  • do use a decent charger that follows the constant current/constant voltage charging protocol correctly and doesn't - through design or not - try to fill your cell to the brim

What too high / too fast / too low actually means depends on the cell specifications set out by the manufacturer. The low and high voltage extremes are generally specified as 2.75V on load and 4.2V max on the charger but some cell models may specify a slightly lower discharge limit.

Using a cell always to both ends of the design specification at a gentle 0.2C rate should provide the user with the life span the maker intended, with a cost per cycle measuring just pennies. But few of us do all our discharges at 0.2C rates at 20 degrees Celsius, avoiding all the don't items in this overview and employing all the do's, so the number of useful cycles we see is going to vary widely. Cost per cycle will still be very low mind you, so provided we avoid the worst behaviours (over charging and too much heat) we can just go out and use the cells and still enjoy a low cost per cycle.

Cycle Life

When we talk about cycle life we are really discussing the cell's ability to hold its state of charge above an acceptable minimum. A cell that comes hot off the charger near 4.2V that rapidly drops a tenth of a volt or more is at or near the end of its useful life.

The industry definition of the end of a cell's useful life is when the cell can no longer hold at least 80% of the cell's original nominal total available capacity. As a rough guide that means a cell which fails to hold above 4.0 V when fully charged will have lost 20% of its total available capacity and should for most users be responsibly recycled.

Prolonging Cycle Life

Heat, over-discharge, and over-charging, are the li-ion cell's primary health enemies. Avoid pushing the cell in these three areas and the cell will likely deliver a cycle lifespan approaching the maker's specification or at least something close to it - in the hundreds of cycles. In this discussion I've avoided dwelling on over-discharge because for most of us our flashlights or protection circuits in cells will save our cells from going there.

If you want to take reasonable steps to prolong cell life, whether for the mere sport of it or for some industrial application, there are a number of things you can do. Some may or may not be practical for your use case:

  • Use a charger that utilizes the approved charging algorithm
  • Avoid keeping the cells always topped up
  • Avoid storing cells with a full state of charge for the long term
  • Avoid discharging the cell much below the point where at rest the cell voltage quickly bounces back up to ~ 3.6 - 3.7V
  • Do arrange for long term storage in a cool or even refrigerated location
Maximum Charging Voltage Is A Key Factor

The first three of the above points have something to do with what happens to a cell near the full state of charge and it turns out that this is the single easiest usage parameter to adjust if one wants or needs to extend cycle life dramatically. If practical for one's use case, reducing the maximum voltage a cell is exposed to will extend cycle life, affording the potential to increase cycle life by a factor of 2x or more.

It is well understood that a cell pushed above the design maximum voltage will have a shorter lifespan. If not for the safety issues this would be a test any hobbyist could explore as cell life drops off markedly as you notch the charge maximum voltage up by a 10th of a volt at a time, quickly dropping to 1 or 2 cycles, and reaching unsafe levels soon after that. Given the fire and other hazards involved in doing this sort of abbreviated cycle life testing volunteers are thankfully scarce.

Given the detrimental effect of a high state of charge on cell life, it shouldn't be a surprise then to learn that intentionally reducing the maximum charge voltage, if you have a charger which allows for this, can improve cell cycle life. From the thread linked above a rough rule of thumb given was a doubling of cycle life for every 10th of a volt reduction in the max charge voltage. Published studies bear this out.

Chargers Matter, But We Don't Know How Much It Matters

A decent charger will treat your cells more gently by reducing the charge current appropriately in the final stage of charging.

Unfortunately a great many consumer-oriented cradle style chargers do not use the approved constant current, constant voltage charging algorithm so the pickings for known-good chargers of this type are as of this writing very slim and have been for years.

Depending on the specific flaws in a charger's approach, cell cycle life may be impacted but by how much no one will likely ever quantify. At the very least avoid chargers which are known to raise a cell's voltage above 4.2V, and avoid chargers which do not properly terminate the charge.

Folklore and Topping-Up

There is a certain amount of folklore surrounding the use of any cell/battery, much of it rooted in learned truths. Sometimes we find practices developed from experience with one cell technology being misapplied to another. Visit almost any consumer electronics store and dollars to doughnuts you'll run into at least one sales person who will give you really bad advice on cell/battery maintenance, often basing their advice on past experiences with NiCd batteries. Fortunately a forum such as this works hard to correct myths and improve upon understanding.

Even if we do have basic li-ion usage principles understood and mapped out, as Silverfox said earlier in this thread, there is more to be learned about li-ion cell behaviour under different usage conditions.

One of the more common bits of li-ion cell folklore speaks to a li-ion cell's amenability to frequent "topping up" -- common wisdom is that lithium ion cells prefer frequent topping up. This isn't quite true - in fuller context we would say that they much prefer being topped up than to be very deeply discharged. Some additional emphasis on "topping up" undoubtedly has come about because li-ion cells are so much different in this regard to NiCd cells and the infamous memory effect and even different as compared to NiMH cells.

While there is nothing terribly wrong with frequent topping up if this is the most practical approach for how a flashlight is used, the prescription is often repeated mantra-like without diving into the details. "Li-ion cells like/are amenable/enjoy frequent topping up" is very popular advice here, and I've said it myself. That's the thing with mantras, they are easy to spit out and often it is best with a new user to hit the highlights rather confusing the key messages with a lot of words. Like these words.

100% SOC Is What Is Wrong With Topping Up Frequently

Earlier in this thread a U.S. Army study linked in showed much more significant degradation in cell capacity when the cycles exercised the cell to the 100% state of charge (SOC) boundary. Over the course of many hundreds of cycles both the 0 - 100% case and 50 - 100% case showed marked reduction in available cell capacity as compared to the cases which examined cycles over 0 - 50% and 25 - 75% state of charge.

If you apply what is learned from that study, and what is known generally about the effect of higher voltages on these cells, it becomes clear that frequent topping up might not be in the best interest of cell cycle life.

Realizing this might change our behaviour in how we charge our flashlight cells, our cell phones, portable radios, iPods, and even our laptops if they are advanced enough to give us some control over the battery management system contained therein.

The Lightly Used Light

Imagine you've got a light powered by one or more cells that can drive the LED for about 2 hours before obvious dimming, or a driver flashing a low voltage warning kicks in, or the protection circuit of the cell kicks out.

I'd bet a great many users don't often take their lights down to these very obvious high levels of discharge. One might exercise the light over a full discharge when it is new to get to know what sort of runtime and output to expect. I do this and then adjust my usage behaviour to avoid too-deep discharges where I can.

A user who adopts the "frequent topping is good" mantra might slap their flashlight cell(s) (or cell phone, iPod, two way radio...) on the charger after any amount of use, even if insignificant. I'm guilty of this in the past -- I have an amateur radio transceiver powered by a lithium-polymer pack and I routinely would slap it on the charger even if I've only used it sparingly on a given day. Drop-in chargers are bad if you are trying to break this habit!

If our example user of the two-hour runtime flashlight happens to routinely recharge cells even if they've only been used for 10 minutes here or there, the cells are only lightly discharged and the use-charge pattern will ensure the cell(s) are spending a lot of time at 100% state of charge.

In other words the usage pattern will look like one of the two worst cases depicted in the U.S. Army study. Those observations suggest it best to avoid charging a cell to the max and/or avoiding leaving the cell at a maximum state of charge for any length of time. Coupled with what is also known about reducing the max charge voltage - a rough doubling of cycle life for every 1/10th of a vole reduction in max charge voltage - it seems that recognizing this pattern of use (abuse?) may provide some opportunities to enhance cycle life.

Topping Up Without Filling To The Brim

Noted earlier, not all consumer style cradle chargers employ the right charging algorithm. Some of these non-compliant chargers do treat the cell more harshly, so simply shopping for the right charger will play at least some part in enhancing cycle life and may play more than a minor role in some cases. The Pila IBC is one of the very few that does charging right.

Consumer style cradle chargers being what they are -- slap cells in and walk away devices -- don't give a user the flexibility to adjust the charging voltage. Some appear to avoid taking at least larger 18650 cells right to 4.2V all the time so there may be a little gain there. The key opportunity with consumer style chargers is to avoid one that hits too-high charging voltages or doesn't fully terminate.

If a user's charger offers no flexibility in selecting charge voltage then perhaps the only avenue to avoid storing the cell at / near a 100% state of charge is to pop it in a flashlight and run it briefly -- a minute or two or five depending on light -- to bring the cell's state of charge down several hundredths of a volt. Is this practical? Maybe not for most.

Hobby chargers on the other hand are a battery geek's best friend and usually offer much flexibility for setting lower charge voltages and various termination parameters. Be sure to cost in the purchase of such a charger and amortize it over the number of cells you have to buy... spending 40 or 100 or 200$ to slightly enhance the life of a handful of cells is hardly worth it, and many would argue there is no real financial argument for spending any additional funds in an attempt to extract maximum cycle life. For the determined we can call such pursuits a sport.

Conclusion

Unless your charger provides you with flexibility, your options for avoiding a 100% state of charge are limited. What will be practical for many is to simply avoid topping up the cell when it has seen very little use.

Of course if you need a full state of charge for your job or activity, then you'll charge the cell accordingly and use it. Likewise if adhering to this practice of avoiding full state of charge is going to take a lot of your time (which is valuable) or investment in new equipment, it probably isn't useful to do -- the cost in replacement cells is almost certainly going to be lower than either factor.

The bottom line: If practical for you, avoid too-frequent top-ups. As a general rule of thumb, provided we avoid extremes in use, avoiding charging to or leaving cells at a 100% state of charge appears to offer the best opportunity for improving longevity of in-service cells.
 
Top