Lithium-ion charging

[jawa_master]

Hi guys.

so, recently my Nintendo DS lite broke down, so i decided to re-use the Li-Ion battery in in, since it have 1000 mAh.

I am not very experienced with Li-ion battery's so i'm asking you: what is important to remember when you charge a Li-ion battery?

(some info: 3,7 volt, rechargeable (of course),
and it says "do not modify". I say "cant tell me what to do.

But seriously though, what is important when charging it?

[Siwastaja]

1-cell battery is extremely easy - you just need a lab power supply with a working current limit and working voltage setting. Set the voltage to 4.2V and current limit to at most 500 mA. If you are not sure about the supply, short it with a current multimeter to verify it gives 500 mA. Measure that it gives 4.2V without load. Connect the battery, and Bob's your uncle!

Once the voltage reaches 4.2V, the PSU starts dropping the current. Once the current is below about 50 mA, you can consider the battery full. You can stop right away when it hits 4.20V, but the battery will be at about 90-93% at that point.

Before charging, measure the battery voltage, and if it is below about 2.5V, discard the battery - it's dead. If it's between 2.5V and 3.0V, you can safely try to revive it by charging first with a very low current, like 20mA, until it's at 3.0V. These are approximate numbers.

And don't charge it if the ambient temperature is below freezing...

[jawa_master]

Ok, but my dad said that the battery might need some special charger to make it "stop charging when it's fully charged". Does he have a point, or is he a complete tech noob?

[jawa_master]

and you probably look at me as a complete tech noob to, saying this but: when you say "dead" do you mean that it cant be saved? like "dead forever?"

[eneuro]

1-cell battery is extremely easy - you just need a lab power supply with a working current limit and working voltage setting.

Why do we need a lab power supply? Probably much easier write a few lines of code and program small MPU with good stable external voltage reference.
The only thing-what about ripple current? What are safe ripple levels for such small Lithium-ion during charging if we wanted do it on daily basis-not only emergency charging when no chance to make better charger?

[Howardlong]

Why do we need a lab power supply? Probably much easier write a few lines of code and program small MPU with good stable external voltage reference.

From experience, I can tell you that is not true I'm afraid. Try it yourself!

Edit: in a production unit, if you are not using a pre-baked li-ion charging solution and rolling your own instead, you need to ask yourself why, the risks are usually too great to be thinking you can do better. (Remember the exploding laptop batteries of a few years back?)

[jawa_master]

Does it exist some sort of IC, that makes an indicator LED blink when it's about to run out of power?

[Siwastaja]

Ok, but my dad said that the battery might need some special charger to make it "stop charging when it's fully charged". Does he have a point, or is he a complete tech noob?

No, it doesn't hurt even if you leave it for some extra time. If you are unsure, just reduce the voltage to 4.15V and you can leave it there for practically forever. If you leave it at 4.20V for too long, it is technically overcharging, but to something like 101% or 102% which doesn't matter too much.

Manual stopping is just fine. Stop a bit early and it's at 98% or so, no big deal. Forget it at 4.20V too long and you have very slight overcharge. It's not gonna blow up.

But it's best to verify that the power supply indeed is a proper CC-CV charger with both working current and voltage limits. A good lab supply is exactly this.

[Siwastaja]

1-cell battery is extremely easy - you just need a lab power supply with a working current limit and working voltage setting.

Why do we need a lab power supply?

You need a supply which is both current and voltage limited (CC - CV modes). Lab supplies typically provide this, but beware, some might be of low quality, so check the voltage and current limits by measuring open-circuit voltage and short-circuit current.

Probably much easier write a few lines of code and program small MPU with good stable external voltage reference.

  Microcontroller is not a power supply. It's a microcontroller. You can use it as a component in a power supply. You still need the power supply. It can be a linear IC (integrated li-ion charging IC, even LM317), or a DC/DC converter. It can even be just a voltage regulator (like LM317) plus a beefy series resistor, although it slows down near the end. Anything which limits both current and voltage will do.

There is no magic going on in standard li-ion, it's just like an ideal battery. No need to sense special things like NiCd/NiMH "full" dV/dt detection. Nice voltage curve going up. 100% charge efficiency. No Peukert effect. Same amount of electrons go in and out. Voltage shows the state-of-charge nicely. All kind of "special stuff" happens with lead acid, NiCd and NiMH due to side reactions. Li-ion just moves ions around and the voltage shows how full the electrodes are.

The only thing-what about ripple current? What are safe ripple levels for such small Lithium-ion during charging if we wanted do it on daily basis-not only emergency charging when no chance to make better charger?

AFAIK, ripple current shouldn't matter, but to be really on the safe side, I would design it around the peak current, in which case, you get less average current out when there is a lot of ripple.

TLDR; Just make sure that the charging current never exceeds 500 mA and voltage never exceeds 4.20V, and it doesn't matter how you do it. And, using a bit lower voltage will be a great thing for cycle and calendar life, so if you don't exactly need 100%, do it at 4.10V, get 92% energy and enjoy your doubled lifetime.

[eas]

Before charging, measure the battery voltage, and if it is below about 2.5V, discard the battery - it's dead. If it's between 2.5V and 3.0V, you can safely try to revive it by charging first with a very low current, like 20mA, until it's at 3.0V. These are approximate numbers.

First, a protected cell, which the OP likely has, may cut off the output if the cell voltage drops below a threshold value, so it might read ~0v. Throwing out such a cell because the voltage is "too low" is a waste.

Second, for unprotected cells, that have discharged slowly due to a small load or self discharge, I think a 2v threshold is reasonable.

In either case, a low charging current of 20-50mA or so should bring the cell to 3v in 10-15 minutes. If it does, the cell can be charged at a higher rate, if it doesn't, the cell is dead.

Look for "TP4056" charging modules on ebay. They are available for a dollar or two. They automatically handle the low-rate charging below 3v. The constant current charging from 3-4.2v, the constant voltage charging, and the charge termination at ~1/10th the original charge rate. TP4056 boards are usually set to a 1A charging current. You'll want to modify the board by changing the charge rate resistor to something that will give you a 500mA charging current for that cell.

[Siwastaja]

Most packaged single cells I have seen do not have active protection switch. Instead, the product itself that uses the battery provides the low-voltage cutoff (it's easier that way); the pack only has a thermal overload fuse. But yeah, some do have the active protection circuitry, in which case the cell indeed could show 0V plus some leakage through the switch - if this is the case, very carefully use current like 20-30 mA for a few seconds to see if anything happens.

[Howardlong]

Does it exist some sort of IC, that makes an indicator LED blink when it's about to run out of power?

Yes, more than you can shake a stick at in fact! They usually do much more.

If you are using protected li-ion cells, which are common these days and physically look identical to non-protected units, they protect themselves with integrated mosfet switches against under voltage and over voltage. That does not absolve the charger itself of taking responsibility for that, the integrated protection is there as a backup rather than to be used as the primary method. They can also protect against excessive charge and discharge current. The mosfets usually operate as simple switches, they are not in the linear region, so are not current limiters: they're either on or off, and an integrated timer and/or power off latch resets the state.

[jawa_master]

Ok. thanks Howardlong. could you at least shake that stick of yours at one of them, so i could go ahead and buy one on http://dx.com ?

(ergo: say the name of it?)

[eneuro]

Microcontroller is not a power supply. It's a microcontroller. You can use it as a component in a power supply. You still need the power supply.

I've made many chargers based on 8dip AVR ATTiny85 for lead-acid batteries with Hall current sensors and active mains monitoring and ideal bridge rectifiers, so I DO NOT NEED lab power supply to charge Lithium-ion batteries, but talking about 4.2Vmax without voltage ripple tolerance in power supply doesn't make any sense, unless we assume we have good lab power supply, but not such a big deal provide 0.5A low ripple voltage.

I know that those batteries can be flammable and... my cell phone is powered from... 12V car starter battery-doesn't need recharging for many weeks- http://www.fairchildsemi.com/datasheets/LM/LM350.pdf 3A simulates 4.1V Lion and bloody crappy phone is happy without oryginal dead Lion battery

[Howardlong]

Ok. thanks Howardlong. could you at least shake that stick of yours at one of them, so i could go ahead and buy one on http://dx.com ?

(ergo: say the name of it?)

The Microchip MCP73871 among many others has a low battery output. I'm afraid I don't know about specific end user consumer grade units themselves, the ones I design have the charging and power distribution system integrated into the unit itself, it is not a standalone charger. Sorry, I thought you were looking for a chip solution rather than an end user solution.

[Howardlong]

Microcontroller is not a power supply. It's a microcontroller. You can use it as a component in a power supply. You still need the power supply.

I've made many chargers based on 8dip AVR ATTiny85 for lead-acid batteries with Hall current sensors and active mains monitoring and ideal bridge rectifiers, so I DO NOT NEED lab power supply to charge Lithium-ion batteries, but talking about 4.2Vmax without voltage ripple tolerance in power supply doesn't make any sense, unless we assume we have good lab power supply, but not such a big deal provide 0.5A low ripple voltage.

I know that those batteries can be flammable and... my cell phone is powered from... 12V car starter battery-doesn't need recharging for many weeks- http://www.fairchildsemi.com/datasheets/LM/LM350.pdf 3A simulates 4.1V Lion and bloody crappy phone is happy without oryginal dead Lion battery

Yes, but you stated that it's easier to make an MCU solution than it is to set up a lab PSU to charge a Li Ion cell, when it simply isn't. You're already taling about messing with Hall sensors and bridge rectifiers, at some stage you'll be needing some form of power transistor no doubt and need to consider the thermal engineering. In the time it took me to write that last sentence, I could've set up a lab PSU to charge a Li Ion cell. In the meantime your soldering iron's barely got up to temperature in that time!

[amyk]

The only thing I'll add is to mind the voltage tolerance: 4.2V is the safe limit but the absolute max is 4.25V (a little over +1%), so you'll need to make sure your equipment can achieve that accuracy otherwise you could be overcharging and not know it. As mentioned above, there's always the option of going slightly lower. Lions do not suffer from memory effect and are perfectly fine to stay below full charge - in fact they degrade less if they're kept around 40% full (which is usually in the 3.7xxV range.)

[Seekonk]

There is a something 4056 chip that they sell on ebay for under $2 as a complete assembly with status LEDs that does all the thinking for you.  That board requires 5V, many old cell phone wall warts are suitable and they are everywhere.

[Siwastaja]

I've made many chargers based on 8dip AVR ATTiny85 for lead-acid batteries with Hall current sensors and active mains monitoring and ideal bridge rectifiers, so I DO NOT NEED lab power supply to charge Lithium-ion batteries, but talking about 4.2Vmax without voltage ripple tolerance in power supply doesn't make any sense, unless we assume we have good lab power supply, but not such a big deal provide 0.5A low ripple voltage.

It's a bit unclear to me what you are trying to express.

Lead-acid is kind of easy because they can be charged with very high currents, as long as you watch the voltage. So any voltage source with some kind of short circuit protection will do. Like a three-pin linear regulator. Lead acid batteries are also usually very large.

Most li-ion, however (expect for Lithium Titanate Oxide, but that's not reality for most people anyway), has seriously unbalanced discharge/charge limits; You don't have to stress about discharge current limits so much, but too high a charge current will result in lithium plating (intercalation to graphite lattice fails), which is both a lifetime and possibly a safety risk. This is especially the case when the cell is nearly full.

So you absolutely need to limit the current to a known value! If you don't have the datasheet of the particular cell, the safe maximum assumption is 0.5C (500 mA with a 1000 mAh cell, i.e., two-hour charge). This assumption does not have any extra leeway! Now, the way you do the current limiting is irrelevant and it is up to you (as long at is works), but it's hard to see how a "microcontroller" is an answer to that. It's the trivial part. Potentiometer is just fine too. You still need the actual current limiter. Either linear (constant current regulator chip) or, for example, a current-controlled buck converter.

Any battery is like an LED, almost a "short circuit" with a voltage drop that is only slightly related to the current. If you connect a small 1000 mAh li-ion cell that sits at 3.6V to a 4.2V voltage supply, it probably takes about 10A of current and self-destroys, hopefully any thermal fusing preventing fire.

Again, about voltage ripple, it is not critical. You want to have a regulated voltage. Any proper regulation works. The battery also smooths out the voltage ripple, working like a capacitor. Just watch for the maximum voltage.

Why I suggested lab power supply? That's because I think that many of the forum members here have access to one, so it's the easiest solution for one-off or random charging, without building or buying anything! And, people seem to dismiss that option because a lab supply is not sold as a dedicated "li-ion charger", but in fact a proper CCCV lab supply is a proper li-ion charger (when used correctly). Of course, if OP doesn't have access to a lab supply, then the easiest thing is to just go and buy an off-the-shelf small li-ion charger. The e-cigarette folks know a lot more about these chargers. I have one that has switchable 0.5A/1.0A charge current and runs off a 12V DC wall wart. These things do everything automagically.

But I suppose the OP wanted to know about the li-ion charging and how to do it. Lab supply is a good first step, because that shows exactly the correct procedure with CC and CV phases.

[Siwastaja]

Lions do not suffer from memory effect and are perfectly fine to stay below full charge

+1 for this, absolutely!

Typically, you can expect something like this:

Charge end voltage -- Usable capacity -- expected cycle life
4.20V -- 100% -- 500 cycles
4.10V -- 92% -- 1000 cycles
4.00V -- 84% -- 2000 cycles
.... or something like that. There is some research on this, and every product is different, YMMV, yada yada, so these figures are for the general idea. But it's always a good idea to "undercharge" if you know you won't need the full capacity. IMO, all modern devices should have an option to configure the charge level to increase battery life.

In my own studies, I have found that charging current near the top voltage has extreme effect on the cycle life. So really respect that 0.5C limit. If you absolutely need to be fast and want to "break rules", do it below 4.0V and then slow down.

[cs.dk]

Sorry to break in, but this topic seems to cover my question, rather then starting a new one.

I have some Canon batteries, but my original Canon charger will for some reason not charge them more.
Batt1 have 5,734V, Batt2 have 6,542V. The batteries are Li-Ion 7,4V - 1500 mAh. My Canon charger states 8,4V and 1,2 A on the label.

Would it be safe to "wake" them, by charge them slowly, maybe at 0,2-0,3C? Or is it better to charge them with higher current, maybe 1C? My charger is a Hyperion EOS SuperDuo3 - http://www.hyperion-eu.com/products/product/HP-EOS0720iSDUO3
The manual says this:

Lithium Polymer (3.7v) vs. Lithium Ion (3.6v)
Please make absolutely sure that you properly identify the type of lithium battery you are charging, the capacity in mAh, and the number of cells wired in series and the pack voltage!
Lithium Ion(3.6v nominal) should never exceed 1C charge rate. Never charge Li Ion 3.6V types by LIPO 3.7V charger mode!

On the batteries there is stamped Li-Ion, which should be 3,6V/C, but the 7,4V rating indicates a LIPO? Should i choose 3,7V/C and charge in the LIPO-setting?

Thanks.
EDIT: I've got a lab supply to, if it should be better then the charger.

[Siwastaja]

I have some Canon batteries, but my original Canon charger will for some reason not charge them more.
Batt1 have 5,734V, Batt2 have 6,542V. The batteries are Li-Ion 7,4V - 1500 mAh. My Canon charger states 8,4V and 1,2 A on the label.

5.734V would work out 2.867V per cell if they are in balance, but it's probable that one cell is even more empty, which mean the chances for survival are slim. Is there a center tap brought out you can measure to get individual cell voltages?

The only safe assumption is that those are dead, but you can experiment with a small risk if you do it right - very slow charge, and stop it to verify if they start self-discharging.

Would it be safe to "wake" them, by charge them slowly, maybe at 0,2-0,3C?

Waking current is a lot less, 0.01C to 0.02C to be safe. So something like 30 mA max. Voltage should start rising slowly - if it doesn't rise after a few minutes at all, the chances are not good.

Or is it better to charge them with higher current, maybe 1C?

Never ever charge random li-ion cells you are not sure about at 1C. Healthy power cells may be good at 1C charging or even more, but you need the specifications first.

My charger is a Hyperion EOS SuperDuo3 - http://www.hyperion-eu.com/products/product/HP-EOS0720iSDUO3
The manual says this:
Lithium Polymer (3.7v) vs. Lithium Ion (3.6v)

This is so total and utter bullshit that it's not even funny anymore. I wouldn't trust a charger if the manufacturer has no freaking clue what they are doing. So what this basically is, they throw out random words and numbers which happen to sound nice, but this process has nothing to do with engineering. Take anything they say as a grain of salt. It's probably specifically engineered to charge high power RC pouch cells and probably works fine for that purpose - it's probably designed by trial and error, and then they come up with an alchemy-like "theory" which explains their limited experience, using "science words" randomly. I have seen this design pattern in RC world many times.

I also wouldn't trust a cheap high-power device to be reliable at low power levels, and it's 100% sure it won't go down to less than 50 mA that you need in conditioning these overdischarged cells.

These kind of cheap Chinese kludges that are overmarketed can be really dangerous (I have horrible experience with POWERLAB 8 ). They are really only good for what most people use them for - just charging the typical large, high-power and healthy RC battery.

On the batteries there is stamped Li-Ion, which should be 3,6V/C, but the 7,4V rating indicates a LIPO? Should i choose 3,7V/C and charge in the LIPO-setting?

First, there is no such a thing as "lipo". It's a marketing term that means absolutely nothing. It's all lithium ion. The gimmick is that the term "lithium polymer" was supposed to refer to a new type of groundbreaking solid electrolyte which was a huge thing in the 1990's, but it was never actually realized. But they still wanted to start using the name as it had marketing expectations, so they just refer to the polymer package - the plastic bag the cell is in! In professional world, we call it a pouch cell, but this is not important. Those who want to get the best of the two worlds call it "polymer pouch" - they can get the word "polymer" in so that ignorant people will be "oooh! lipo!!", but professional people don't feel as insulted as they would if the term "lipo" was used.

Second, the voltage doesn't really mean anything - li-ion products of even the same or similar chemistry have very slightly different nominal voltages, usually around 3.63V..3.65V, so you can see where this is going, it rounds to either 3.6V or 3.7V. No big deal, it doesn't tell you anything. It's a nominal value, some kind of average, and the difference is negligible. The exception would be 3.8V in which case it is probably a higher-voltage LCO cell that can be charged to 4.30 or 4.35V.

7.4V rating means there are two cells in series. 7.4V/2 = 3.7V.

EDIT: I've got a lab supply to, if it should be better then the charger.

It definitely is, if you can adjust it to small currents. Verify the current with a multimeter. Small current is really essential in safe revival of overdischarged cells.

Disclaimer: Reviving overdischarged cells is not guaranteed to be 100% safe.

[MrAl]

Hi,

You have to be more careful with Li-ion cells than any other type of battery.  That's because they can explode if not charged properly, and most of the problems in the past have occurred during charge.

First, the voltage limit is 4.20 and max is 4.25 but because of volt meter inaccuracy it is best to set the voltage limit to 4.15 volts and then you get extra life if it really is 4.15v.

Second, the current must be limited.  For a small cell 300ma is a good setting, for larger cells you can go up to 1 amp but you really should consult the data sheet for the individual cell to be sure.

Third, charging two cells in series is a dangerous thing to do.  One cell could charge to a higher voltage than the other and thus explode.  If the cell has built in protection you could get lucky, but it's not really a good idea to rely on that protection, it is better to design the charger properly in the first place.  This means either charging each cell individually or monitoring the charge of each cell individually and bypassing once cell when it gets charged while you let the other one still charge.

The lower voltage limit is really 2.5v, and anything lower has to be charged at lower than normal current until it gets up to 2.5 volts or so.  Ideally you shouldnt even use a cell that goes too low because things happen to the cell internally that make it more subject to internal shorts, so a little extra vibration one day could set it off.

You can build a charger with a wall wart and an LM317, which is probably the simplest charger.

So to list the precautions:
1.  Charge limit to 4.2 volts.
2.  Charge limit to 300ma for unknown cells, higher for known cells after checking the data sheet.  This sort of goes with the capacity of the cell.
3.  Charge at low current with voltages starting less than 2.5v until the voltage ocmes up.
4.  If the cell goes too low discard it, or be very careful with it in the future as to vibration or pressures applied to the outside package of the cell.  The recommended procedure it to throw it out.

The LM317 charger is just the wall wart, series resistor to limit current, and LM317 set to put out about 4.15v, with the cell connected to the output to ground.  You can get more fancy on the current limit with a NPN transistor and a couple resistors.

[cs.dk]

Thanks a lot for a really nice detailed answer.

It's possible that the cells are uneven, there is no centertab, so it is impossible to measure, without breaking them apart. There is +, - and B, D. From + to both B and D i measure the batteryvoltage. Don't know what those are for.
I actually have a third battery that is at 0V. I will try to knack it open, and measure the individual cells.

The batteries are cheap, so this is just an experiment, inspired by this thread. I have got a lot of lead acid car batteries up running again, with a labsupply and some patience. The reason i've got this "cheap chinese" charger is that it can charge two batteries at time. LA batteries at those sizes can better handle some "failures" in accuracy. Yes, the charger is build for the R/C people.

One of my supplys should be able to go down in that ballpark. I'll double the short-circuit current with a meter.

So, if I understand you correct, I program the supply to 8,4V and ~30 mA, and see if the voltage raises within 15 minuttes? No raise = dead battery..

very slow charge, and stop it to verify if they start self-discharging.

By selfdischarge you mean the voltage would drop again, as soon the PSU is off?

Thanks.

[Siwastaja]

If there is no center tap, then there is no cell-level monitoring / balancing at all. Surprisingly, this is just fine with 2s packs when done right, and it's quite normal to do it this way. However, this won't give you a chance of reviving the battery - the balance that once was there is probably lost.

I would discard especially the one with the lower voltage for safety. It may revive, but if the cells are not in balance anymore, it's going to be a huge problem when fully charged.

So you really need the center tap for this.

Now taking it apart and reviving and charging the cells separately would make a lot of sense. Then you can manually balance them by charging them separately to 4.15V, for example, leave them for a few days or weeks, and if they don't lose voltage and stay in balance, build the pack again.

Yes, by checking the self-discharge, I mean that stop the revival process every now and then and check if the voltage stays. If it starts dropping back, the cells should be considered toast.

It would be very important to be able to separate the two cells for this reason, too: if one of them revives, it may mask the one that regresses as the voltages are summed by the series connection.