LiPo charging -end of cycle current rates

[Siwastaja]

https://hal.archives-ouvertes.fr/hal-00876555/document is one good paper I've found on the topic, although there are plenty of others. I'd encourage anyone making general claims such as above to provide their sources, which shouldn't be difficult if it's so well documented.

That paper is about LFP chemistry which is practically unused expect for some niche applications. It's very different due to lower voltage.

If we look at the modern world (and foreseeable near future), NCA and NMC are the relevant chemistries. LFP was a commercial failure, and LCO is really getting obsolete now.

On NCA and NMC, there is a correlation between SoC and calendar fading, although it's different from what people generally think it is - storing at 90% or 80% is not the answer, if you want to really increase keeping properties, you need to go as low as 40-60% (source: http://www.a3ps.at/site/sites/default/files/downloads/evs28/papers/A3-01.pdf Figure 6 for example); my own research shows some improvement at about 80% on some cells, but it really starts getting better at 50%. This kinda sucks. However, more damage is done during charging at 100%, so still charging only to 90% or 80% makes sense, but mostly due to preventing cycle damage, not due to calendar fading.

You are right that tempereture is definitely important, agreeably perhaps more important than SoC, at least on some cells.

If the cells are really going to be stored, i.e., not used during storage, then it's very easy to just discharge them to low SoCs; temperatures may not be as easy (or economical) to control. You are pretty much stuck at room temperature if you don't want to invest money in refrigeration. And room temp may be rather high in certain parts of the world.

For example, in my study, the worst performer of calendar fading, a power cell from LG, was capacity fading at 4.9%/year at 23 degC when fully charged. Stored at 3.60V, the fade was down to 1.9%/year in the same conditions. On the other hand, in a refrigerator, at +4degC, the fully charged cell faded at 1.3%/year, so the temperature is very important but so is the SoC.

But there are huge differences between the cells. The best sample was a cheap Samsung LCO cell that didn't fade measurably when stored at room temp fully charged for 500 days, and even when stored at elevated temperatures, fully charged, only faded 1.1%/year during this time. The LG power cell faded 8.4%/year in the exact same test case.

[Skimask]

Excellent reading linked above.

Gonna awhile before I can get around to putting the 18650 cells I've recovered thus far to good use.  So far, all 36 cells, after a charge and sitting for at least 24 hours, are within 30mV of each other (1 at 4.161v, 1 at 4.139v, the rest at 4.144v +/- .002).
Running each of them thru a couple of light charge/discharge cycle to make sure they all have a decent amount of capacity left and are useful,
Then one final charge, and discharge at about .2C down to 3.7v, and throwing them in the fridge.  Close enough for me...

[Siwastaja]

discharge at about .2C down to 3.7v, and throwing them in the fridge.  Close enough for me...

Nothing to add to your other points, sounds good, but if you are going to do this anyway, then go a bit lower. 0.2C discharge down to 3.7V would leave you at about 60% SoC, and I'm sure that if you do it to 3.6V or even 3.55V instead, you can get some real calendar life benefit. Nothing ground breaking, but why not? 

It may actually happen that the cells stored as you describe may live very well for a decade or two! (Or may not, but it's possible.)

[Skimask]

Following up on this a bit...

I'm working on a dual rail converter circuit, basically beating up on the chips by working them past the datasheet limits for current, etc.  Getting them hot, watching them shut down, cool off, and restart themselves, etc.  At the same time, I'm also beating up a few older 18650's that are powering the PCB.  I'm using one of those ebay TP4056 PCBs for 'protection'.  Working out well enough so far.  It's cutting off the battery at around 2.6v or so.  And shuts off the charge at about 4.2v.  A bit low and a bit high for each for my tastes, but it works.

Question is...in general, what conditions cause the cells to self-destruct, eg. vent, explode, melt-down, etc.  I'm not talking about just "not taking a charge anymore" or "not holding a charge anymore"...  I'm talking about the causes of violent self-destruction.
I'm surfing the misinformation intarwebs right now looking for something conclusive and factual, rather than the normal "well, this happened to me one time, so it must happen to everybody all the time" anecdotal information.

[Zeranin]

Q about proper charging of 18650 type cells...

For the purposes of discussion, we'll just assume that the cell being used is a standard 18650 type rated at 2000 mAH, rated at 1C charge/discharge, max of 4.2v, low cutoff at 2.75v...

Normal LiPo charging starts of with CC, in this case 1C, 2000mA, until the cell reaches 4.2v, at which point it switches over to CV, until the current drops to 1/10 C, in this case 200mA.

So, maybe one guy is going to push that and charge with 5C.  Charging would stop when the CC level dropped to 1/10C of the charger, which would be 1000mA rather than 200mA.
Likely not going to do the cell any favors by pushing it harder and probably going to shorten the lifespan of the cell significantly.  'nuff said.

Now, maybe another guy is charging those same style LiPo's with a cheap USB charger which is only going to get you 500mA at most.  Assume that the charger is 'properly' designed and charging stops when CC level drops to 1/10C of the charger, which in this case is 50mA.

Does this present an issue in future?  eg. charging longer at a lower current, making the LiPo sit at the same absolute maximum voltage for a longer period?

You raise good points, and I have had similar thoughts in the past. One point that I believe you are getting at, is that the rule ‘terminate charge when current has dropped to 1/10th of inititial charging current’ will result in a different final state of charge, depending on the charging current. Specifically, a lower current charger using this algorithm will give the cell a higher final state-of-charge, compared to a higher current charger. Presumably all chargers aim to provide the same final state-of-charge, therefore this charging algorithm cannot in general be exactly correct, and will need to varied slightly depending on the charge rate relative to the capacity of the battery. Agreed.

The final state-of-charge is judged from the cell voltage, after the cell has been removed from the charger for a few hours. That is the ‘bottom line’ in determining whether any charging algorithm has correctly charged the battery to a full state of charge.

As much as anything, the common charging rule as described is predicated upon reaching full state of charge within a reasonable time. In principle (within reason) one could use a lower charging voltage, but wait longer until the current had dropped to 5% or 1% or whatever, but in practice the consumer won’t stand for waiting that long, so the charging voltage is set to a voltage that would overcharge the battery is left on for a long period of time, and disconnected while the current is still falling, to compensate, at least that is my understanding.

[Siwastaja]

Question is...in general, what conditions cause the cells to self-destruct, eg. vent, explode, melt-down, etc.

The most prominent reason, by far, is poor cell-level design and/or manufacturing quality, from the lack of cell safety features (shutdown separator etc.), to outright manufacturing defects such as contaminants, electrode alignment problems...

Slight electrical abuse may increase the risk of catastrophic failure on high quality cells, too, but the risk is still very small, because it's miniscule to begin with.

For non-quality cells, I don't like speculating, because you never know. Some may go off without any abuse at all, some may be surprisingly good, until you buy more from another batch. This is why if you are concerned about safety at all, you absolutely must buy Samsung, Panasonic, etc. Or take the very real risk.

Now, to the abuse side:

Thermal abuse is the worst. Thermal runaway onset temperature depends on the chemistry, but all modern high-capacity li-ion chemistries are at about 160 deg C. AFAIK, they can't lower this no matter what safety features they design in. This (about 150-170 degC) is the hard limit for any li-ion brand, if it uses LCO or NCA chemistry. Of course, real damage happens way before this limit. Competeting chemistries with higher onset temperatures (LMO and LFP) are unsatisfactory for other reasons. Thermal abuse point is really important, because if you really fail your electronics side, it can heat up even the best quality, safe cell over it's onset temperature! Design proper fusing, and don't design in high-power resistors dissipating several watts for balancing.

Mechanical abuse; it depends. Low-quality cells have high chances of blowing up if you penetrate the cell with a nail or crush it. Higher quality cells are not immune, either, but the risk is smaller. 18650 is relatively safe due to the strong steel casing.

Overcharging; depends a lot. High-quality cells are unlikely to blow up even if you severely overcharge them (like connecting to 12V for hours!), but it can happen. Going to about 4.30V should be safe (while shortening the life), expect for the poorest cells that are on the brink of exploding anyway.

Too high charging current: Comparable to overcharging, but partially different mechanism. High-quality cells are unlikely to really blow up. Same thing for charging at temperatures too low.

Charging after severe overdischarge; again, depends. Severe overdischarge causes copper dissolution, and copper ions can build up dendrites which cause internal shorting, similar to lithium plating. Any cell below about 1.5V-2.0V should be discarded. High-quality cells are rather immune, because they are designed to be immune against internal shorts.

Having said all this, you should still avoid abusing even the highest quality cells. No cell is completely immune to abuse, there always are increased risks.

[Skimask]

1 - Design/quality/etc - The only cells I've got at the moment are used 18650's out of old laptop packs, all unprotected, only a few that have had <1V on them when I took the packs apart, only a couple that wouldn't come up to any usable voltage under a light charge.

2 - Non-quality cells - I'll play games with used cells, but from what I've seen on the intarwebs, I'm not even thinking about buying or using any of the cheap "insert garbage name here"-Fire cells.

3 - Thermals - 160C!!!  And here I was beginning to worry when the outer wrap was getting up above 50C.

4 - Mechanical - Don't see that happening in my situation, but, ya never know.

5 - Overcharging (voltage) - Usage those TP4056 boards.  They seem to be cutting off right around 4.2V.  Good there.

6 - Overcharging (current) - Those same TP4056 are limited to 1 amp (by a programming resistor) depending on the USB charger connected, and I'm using an old 5V @ 500mA part.  Good there as well.  Near the end of the charge, I'm not seeing much above 100mA before cutoff.  Good there.

7 - Overdischarge - TP4056 cutting off at about 2.5V under light load, about 2.8V under heavy load.

Maybe I'm playing it too safe.  Who knows.  Suffice to say I'll probably NOT have any unintended fires going on around here.

[Siwastaja]

3 - Thermals - 160C!!!  And here I was beginning to worry when the outer wrap was getting up above 50C.

Don't get me wrong, 160 deg C is kind of "will burst in flames despite all manufacturer safety efforts" limit. Note that hotspotting can happen inside the cell, and case temperature is not the same as internal temperature, even though in 18650 it's close (which is great in 18650, and one of the reasons Tesla states they use them for!). In any case, a lot of damage will happen way below 160 deg C. The shutdown separator will melt, practically disabling the cell (permanently) somewhere near 120-130 deg C; given that there is a working shutdown separator (poorest Chinese cells may not have it).

160 degC thermal runaway onset temperature is kind of low, because when leaving appropriate leeway for shutdown separator to work, the "absolute maximum rating" cannot be much over 90 degC, which leaves little margin for internal heating in hot environments. Just leaving a black laptop battery pack in a car on a sunny day is enough to destroy the cells, in extreme cases.

You need to worry if the 18650 steel case is getting over 60 deg C for extended periods of time, or over about 70-80 deg C momentarily. A lot of damage will happen anyway beyond let's say 90 deg C, even though it's unlikely that they go in flames until you are getting nearer to that 160 deg C region.

Maybe I'm playing it too safe.  Who knows.

Nah, you are implementing the basic minimum protection, as you should. At least it means you are not destroying the cells. Even though it would be very unlikely that any small abuse beyond normal limits would cause catastrophic failure such as fire, don't risk it. And destroying working cells makes no sense anyway.

[Siwastaja]

7 - TP4056 cutting off at about 2.5V under light load, about 2.8V under heavy load.

Did you get this right? It sounds to me that those voltages should be swapped for maximum capacity under heavy load, and good protection under light load.

[Skimask]

Thermals - Ya, I figure 50C on the outside means (value thrown to the wall here) 70C on the inside.  Trying to take images all around the cell to see what's happening.

7 - TP4056 cutting off at about 2.5V under light load, about 2.8V under heavy load.

No, those numbers are right.  Thermal images don't show anything getting hot.  Could be a discharge current limit on the DW01 which are present on the TP4056 PCBs I've got. ( http://www.ebay.com/itm/10PCS-5V-Micro-USB-1A-18650-Lithium-Battery-Charging-Board-Charger-Module-/201039030988?hash=item2ecedc2acc:g:zHgAAOSw1vlUxZ0Q )

See at 20:57 below...


Had the battery loaded down pretty good in that test trying to find the "boiling point" of the circuit, and it cut out at 2.984V, which is measured AFTER the TP4056 board, along the "rails" of the protoboard I was using for power distribution.  So, yes, maybe as the voltage gets lower and the current gets higher, I'm hitting that limit on the DW01 and it trips.
On other "tests" with a much lighter load, the battery would die at about 2.5V, which is the limit for the TP4056.

I'll try it again with slightly less load this weekend and see what happens.

EDIT: Aha!  I was just re-reading the datasheet for the DW01.  Says that it uses the voltage drop across the external control MOSFETs to determine current.  So, too much current or too high resistance across the FETs, overcurrent detection trips, DW01 engages the shutdown on the FETs.  Next time thru, I'll try to put a meter across the FETs and see what that reads.

[BravoV]

EDIT: Aha!  I was just re-reading the datasheet for the DW01.  Says that it uses the voltage drop across the external control MOSFETs to determine current.  So, too much current or too high resistance across the FETs, overcurrent detection trips, DW01 engages the shutdown on the FETs.  Next time thru, I'll try to put a meter across the FETs and see what that reads.

I've been using these cheap TP4056 boards for a while, beware of these with DW01 mosfet + 8205A protection ic, the RDSON of DW01 vary so much.

Nowdays I used those without these protection IC + Mosfet as its more stable and efficient among boards since its just the single IC that does all the magic. I'm using separate PSU rails soldered on board that is adjusted exactly to minimize the drop out, not thru the USB port.

Attached photo of two versions that are common in the market, left boards that are with protection, and the right are just pure bare minimum with single TP4056 ic only and with it's supporting discrete components.

If you watch closely, the ones without protection have the shunt resistor 0.4 Ohm as the TP4056 design specification, that alone is enough to measure the current with adequate accuracy if needed. Yeah, the IC's quiescent current is too small that I think can be safely ignored.

[Skimask]

The two types are basically the same price on ebay, figured having the DW01 was better than nothing at all.
I understand that process variations in the 8205A and DW01 make the over-current protection iffy at best with potential for a wide range of cutout currents.

Are you implying NOT to use the type with the DW01 due to the V(drop) across the FET's?
Surely these things wouldn't be used in any kind of consumer product, but good enough for the home lab, yes?

[Skimask]

And yet another slightly confusing...thing...

Still working on loading down the 18650 cells (old cells, rated at 2200mAh on the package, who knows what their true capacity is at the moment, semi-irrelevant at this point)

With about a 1.8amp load on one cell, about .81C, the load circuit ran for 50 minutes before the low voltage limit kicked in on the DW01.
I paralleled up 2 fully charged cells, cells that I knew would hold X volts after charging and an overnight rest, and ran the load test again.  Same load, about 1.8amps.  It ran for 103 minutes before dumping.
In theory, the load on each battery went from .81C, to about .4C.  I would've figured, based on looking at various discharge graphs, I would've gotten a fair bit more run time on the circuit.
Maybe the one battery has significantly less capacity than the other.  Don't know.  I don't have a proper way to test that.  But I did run both batteries, individually, on the same load circuit, and got the same results for each one, 50-51 minutes before dumping.
Maybe I have to cut the discharge rate significantly before seeing a measurable increase in run time.
Maybe I should just be happy with what I've got and run with it...

[Siwastaja]

If you have 50 minutes of runtime to a constant current load with both cells separately, that should definitely double to 100 minutes when connected in parallel to the same load, or maybe just a little bit longer due to smaller voltage drop at the end of discharge.

If you expect significantly larger capacity at lower load, this is a wrong expectation maybe based on lead-acid, where the capacity is radically reduced at high discharge rates. Great thing in li-ion is that capacity is not (much) dependent on load current, you always get (very close to) theoretical full capacity out of it.

Capacity may appear to increase considerably on small loads when the higher reference load is too much for the cell (considering the aging-induced increase of DC resistance and discharging temperature), or when the cutoff voltage is higher than necessary.

CC-CV discharge always gets the full capacity out of the cell.

But your test procedure appears way too indefinite to do precise conclusions about the cells.

You can easily measure both coulombic and energy capacity (Ah and Wh) with any resistive load and two multimeters, graphing V & A every few minutes. Yes, you need to concentrate on this for an hour or so, so you want to have a tester if you do it more than a few times.

Typical discharge cutoff for LCO laptop cells is approximately 2.8V @ 0.2C discharge.

[Skimask]

50 minutes of runtime to a constant current load with both cells separately

It's almost a constant current load.  The load is my TPS65131/TPS63001/TPS63002 buck/boost converter board.  But with that being said, the conditions are the same between runs...no changes in load resistors on the outputs of the converters, I don't change the wiring around, etc.

If you expect significantly larger capacity at lower load, this is a wrong expectation maybe based on lead-acid, where the capacity is radically reduced at high discharge rates. Great thing in li-ion is that capacity is not (much) dependent on load current, you always get (very close to) theoretical full capacity out of it.

Could be.  Another look at the discharge curve for capacity vs. discharge rate shows the mAh drawn off the battery converging once it gets down a really low voltage, where if you stop the drawdown higher up, the capacities vary a bit more.

But your test procedure appears way too indefinite to do precise conclusions about the cells.

The only thing definite is that the "procedure" is the same between the runs.

You can easily measure both coulombic and energy capacity (Ah and Wh) with any resistive load and two multimeters, graphing V & A every few minutes. Yes, you need to concentrate on this for an hour or so, so you want to have a tester if you do it more than a few times.

I've been doing it the hard way...3 meters, a stopwatch, and a spreadsheet

Typical discharge cutoff for LCO laptop cells is approximately 2.8V @ 0.2C discharge.

I've been running them down until the TP4056/DW01 cuts out, usually about 2.5v, which in the last run was only 6 minutes after the batteries hit 3.000v, for a total of 103 minutes.  If I wreck those particular cells, no biggie.  They were used in the first place, and I've got to find my own limitations.

No real problems or issues overall.  Probably digging into it wayyy too deep anyway.  As I said earlier, I should just be happy and go with it.

[amyk]

And yet another slightly confusing...thing...

Still working on loading down the 18650 cells (old cells, rated at 2200mAh on the package, who knows what their true capacity is at the moment, semi-irrelevant at this point)

With about a 1.8amp load on one cell, about .81C, the load circuit ran for 50 minutes before the low voltage limit kicked in on the DW01.

That's 1500mAh, so the cells are around ~68% of their rated capacity. Very well-used but there's still plenty of capacity left.

[Skimask]

That's 1500mAh, so the cells are around ~68% of their rated capacity. Very well-used but there's still plenty of capacity left.

Oh ya, they're "vintage" cells, circa 2002 or so.
The price was right...zero