'Exact Behavior' of Battery Chemistries?

[safe]

This information is all available "somewhere" and I have been to the Battery University to try to somehow condense it into one spot, but it's hard to do. Let's see if we can come up with a simple list (to be made into a chart possibly) that says what happens to a cell on charge and discharge for the various types. To get started (and to provide a foundation) this is a chart from Battery University that describes the behavior of the different chemistries generally:

 

[safe]

Okay now what I'm really wanting to know is:

"How do the cells BEHAVE during charge and discharge?"

So for example, if you have a Lithium cell and it's in parallel and the cell is "weak" does it tend to drain faster than a "strong" cell also in parallel and eventually risk a meltdown? How about nickel cells, what do they do when they are charged and discharged? Are the traits the same or different? So I'm looking for a chart that would have something like:

Lithium
Charge Behavior in Series? In Parallel?
Discharge Behavior in Series? In Parallel?

NiMh
Charge Behavior in Series? In Parallel?
Discharge Behavior in Series? In Parallel?

NiCd
Charge Behavior in Series? In Parallel?
Discharge Behavior in Series? In Parallel?

 

[fechter]

Lithium and lead-acid: charge and discharge in parallel OK

Nimh and NiCd, discharge in parallel OK
charge NOT OK

The wierd thing about Ni batteries is when they approach full charge, the cell voltage drops slightly. If you have a bunch of cells in parallel, the first one to hit full charge drops its voltage and hogs all the charging current. It will also draw current from its parallel buddies. This results in overheating, separator failure, cell shorts and an unpleasant cascading thermal event.

I had a bunch of NiCd's in a series-parallel pack I built once. The first time I charged it, I came back later and found razor sharp shards of stainless steel battery case stuck into the wall and caustic electrolyte goo all over the place.

Good thing I wasn't around when it happened. Those things are like mini-grenades when they blow up.

 

[safe]

Lithium and lead-acid: charge and discharge in parallel OK

Nimh and NiCd, discharge in parallel OK
charge NOT OK

The wierd thing about Ni batteries is when they approach full charge, the cell voltage drops slightly. If you have a bunch of cells in parallel, the first one to hit full charge drops its voltage and hogs all the charging current.

Okay, well let's go a little deeper here.

The Nickel batteries "behave" in such a way during charging that a cell once full will "hog" what it can. So you have to charge Nickel in series. But does this mean that once the "first" cell is full the charger will likely turn itself off? Could the entire series be left less than full? Or does this mean that in series the cell that is "full" will eagerly pass the charge along and all the cells will be properly charged?

For Lithium and Lead Acid it's "okay" to charge and discharge in parallel, but how do they "behave" in those situations? When a Lithium cell is "weak" and cannot hold much capacity anymore and it's in a parallel configured pack arraingement does it INCREASE it's drain rate as it reaches empty or does it DECREASE it's drain rate as it reaches empty? Does it tend to "balance itself" or do the opposite and try to "run away" from you and go into a meltdown. For charging, what "behavior" do you see?

Eventually I'd like to get a chart built that really "nails" these behavioral charactoristics and make it really easy to understand.

 

[xyster]

So you have to charge Nickel in series. But does this mean that once the "first" cell is full the charger will likely turn itself off?

Depends I think on the charger -- in general their designed not to quit when the voltage drops the amount equal to the expected single cell drop, but when some number more of cells' voltages drop. The entire series is always left at least a little less than totally full, or some cells a little overcharged. That's one reason for nickel batteries, "matched" cells are preferable, as is charging in strings no longer than 10 cells (12v) if possible.

Could the entire series be left less than full?

And either always is, or is always over-charged, damaging the cells some over hundreds of charges. This is why a balancing system for nickel batteries would be even more useful than for SLA or lithium. Cells matched within a certain tolerance limit the end-of-charge difference, and the amount of overcharge.

Or does this mean that in series the cell that is "full" will eagerly pass the charge along and all the cells will be properly charged?

Cells in series unlike cells in parallel, to my knowledge, do not ever "share" their excess charge. The higher voltage cells may drain a little faster though, and so tend towards rebalance by that mechanism.

For Lithium and Lead Acid it's "okay" to charge and discharge in parallel, but how do they "behave" in those situations?

Specify. "Behave" is too general a term for me to cross-ref an answer via my puny biological brain processors.

When a Lithium cell is "weak" and cannot hold much capacity anymore and it's in a parallel configured pack arraingement does it INCREASE it's drain rate as it reaches empty or does it DECREASE it's drain rate as it reaches empty?

Current does not move (isn't current) when voltages are equal. If you have ten cells in a parallel subpack at 12 volts, theoretically no single cell's voltage will go below 12v, because any voltage difference will be quickly equalized. In practice I think this can happen under high-rate situations where minor differences in cell resistance, wire length to/from the cells etc can lead to an imbalance that takes longer to equalize than the time it takes to run the battery down to dangerously low levels.

For charging, what "behavior" do you see?

Again, in this instance "behavior" is too general a term for me to evaluate.

 

[safe]

Current does not move (isn't current) when voltages are equal. If you have ten cells in a parallel subpack at 12 volts, theoretically no single cell's voltage will go below 12v, because any voltage difference will be quickly equalized. In practice I think this can happen under high-rate situations where minor differences in cell resistance, wire length to/from the cells etc can lead to an imbalance that takes longer to equalize than the time it takes to run the battery down to dangerously low levels.

But if you have 10 cells in parallel and one of them cannot tolerate less than say 2.9 Volts while the expectation is to go to 2.5 Volts what happens when the cell that is "defective" sees that 2.9 Volt charge? I would presume that it tends to overheat for some reason and this is how a "meltdown" conditon occurs. All the other cells are happy with voltage levels dropping to 2.5 Volts (or even below) but one threatens to "go nuclear" if it goes to only 2.9 Volts.

How does it "behave" in that situation?

Does it "signal" it's "unhappiness" in some way with some sort of voltage drop or resistance going haywire?... what are the specifics of failure? How could you "know" beforehand that a cell was "hurting"?

 

[xyster]

Does it "signal" it's "unhappiness" in some way with some sort of voltage drop or resistance going haywire?... what are the specifics of failure? How could you "know" beforehand that a cell was "hurting"

Ah, yes there is ample signal! At least in my experience with lithium-ion 18650s. The overdischarge event, where my voltmeter reported each parallel subpack's average voltage right before was about 3.7v, was preceded by a very noticeable 10v or so increase in total-pack voltage sag. Upon opening the pack for inspection, I noted two cells in separate subpacks had ruptured. Here's an image of one of the two subpacks with a ruptured cell:

The voltage for one of those subpack's was 2.35v, and the other 0.7 volts. Most of the other, unruptured cells in these two subpacks were totally dead, their PTC's I presume opened, saving them from the same ruptured-can fate. These cells may have been revivable still, but I elected to toss them.

Presumably, these were the weakest in some aspect.

 

[safe]

Ah, yes there is ample signal! At least in my experience with lithium-ion 18650s. The overdischarge event, where my voltmeter reported each parallel subpack's average voltage right before was about 3.7v, was preceded by a very noticeable 10v or so increase in total-pack voltage sag.

Okay, now let me see if I understand your terminology. By "voltage sag" you mean that the voltage coming from the pack "suddenly dropped". One would presume that a "bad cell" is basically stealing all the energy out of the pack in big hurry and turning that cell into a lightbulb!

If you have a "parallel cells" situation and the voltage is slowly declining as expected for normal use and just one cell is "defective" or "weaker" such that it can't handle the voltage that all the others can, then at some point the "bad cell" will become a vortex into "battery pack hell" and a meltdown begins.

You can avoid this if you don't allow the packs to get used very hard, (you stay clear of any lower limits) but that just means more money for a bigger pack and more weight to get the same job done. (however it does tend to make the packs last longer, so it's probably a good idea)

Going further on this.... I wonder what is happening BEFORE the runaway condition gets started? Are there "signals" that could be exploited so that you know to take the "bad cell" offline or at least balance it so that it doesn't "freak out" on the rest of the pack.

Before a "runaway event" occurs doesn't there have to be a sudden drop in resistance in the cell? All of a sudden the energy finds a "better route" and that's what causes the "runaway".

 

[xyster]

Okay, now let me see if I understand your terminology. By "voltage sag" you mean that the voltage coming from the pack "suddenly dropped".

Yes. Voltage from any source normally sags an amount proportionate to the resistance the voltage encounters, and the current being drawn from the source. In my case, as communicated by the voltmeter and ammeter on my handlebars, my whole pack's normal drop of ~6v at 35 amps suddenly became ~16v at 35amps.

One would presume that a "bad cell" is basically stealing all the energy out of the pack in big hurry and turning that cell into a lightbulb!

It certainly got my attention, but I should have stopped the bike right then and there instead of using electricity to get home. One can presume in such a circumstance that somehow and somewhere resistance is abruptly climbing.

If you have a "parallel cells" situation and the voltage is slowly declining as expected for normal use and just one cell is "defective" or "weaker" such that it can't handle the voltage that all the others can, then at some point the "bad cell" will become a vortex into "battery pack hell" and a meltdown begins.

Yes. But other factors may be the cause instead. For instance, perhaps the PTCs on my ruptured cells did not open as they were supposed to, whereas they did on other cells that did not rupture. As cells in a parallel pack are taken offline, resistance is going to increase because resistors in parallel decrease total resistance, and every battery has a resistance, and so is a resistor. See "Resistors In Parallel Calculator" :
http://www.1728.com/resistrs.htm

You can avoid this if you don't allow the packs to get used very hard, (you stay clear of any lower limits)

That's what I've been doing.

but that just means more money for a bigger pack and more weight to get the same job done. (however it does tend to make the packs last longer, so it's probably a good idea)

Exactly! 33ah is more than I generally need anyway...I've taken to operating the batteries between 3.8 and 4.0 volts when possible (50% - 80% DoD), recharging partially so as to extend pack life by limiting time cells are at high voltage, and on the other end limiting the overall depth of discharge. And if I get lost while exploring, I have 50% left to find my way home And if I'm going out for a long ride of 25 miles or more, I charge up all the way so I still have ~50-60 miles total range (in case I get lost, or decide to explore more on the way).

Going further on this.... I wonder what is happening BEFORE the runaway condition gets started?

In my case, as mentioned, I think the PTC's were opening on some batteries in response to the heat from overdischarge, pulling them offline, creating a cascade effect while resting voltage was around 3.6-3.7 volts and the cells were under 1.3C load.

Are there "signals" that could be exploited so that you know to take the "bad cell" offline or at least balance it so that it doesn't "freak out" on the rest of the pack.

The more expensive 18650's with the full PCB may protect much faster, and more reliably. PTCs are well known to display wide response variance.

 

[safe]

As cells in a parallel pack are taken offline, resistance is going to increase because resistors in parallel decrease total resistance, and every battery has a resistance, and so is a resistor.

So a defective cell gets INCREASED resistance?

That seems strange. If the resistance went down then the battery turns into a "lightbulb" and produces heat. If the resistance goes up then I'd figure that it would tend to avoid a meltdown.

Can you explain why this is?

Are you sure you don't have this backwards?

 

[fechter]

If you had 10 cells in parallel, they would always have the same voltage because they're wired together.

During discharge, if one cell is weak, its voltage will still be the same as the other cells, but it will stop contributing to the output. The opposite is true during charging. The weak cell will stop accepting charge before the others, but its voltage will be the same.

Xyster's failure was more like too high of a discharge current, and is not directly related to letting the cell voltage get out of the allowable range.
The balancer will not help this situation. You still need some form of current limiting so you don't exceed the max discharge rate.

Most motor controllers have a current limiter. As long as it is dialed in right, it should prevent overdischarge current.

 

[Leeps]

Safe increased resistance will make for more power loss. P=RIsquared. P is the power loss, i is the curent squared R is the resistance . Another way to look at it is the voltage droped is E=IR e is the voltage dropped I and r you already know, this voltage drop multiplied by the current gives you your power loss. If you have a shortcircuit then you are right lower resistance is bad, a lightbulb could be looked at as a shorted out resistor i suppose. In this case with a battery lower resistance is a good thing, it means less voltage sag and thus less power loss and less heat. Theres another thing to look at, the difference between ohmic resistance and the impedance. The voltage sag we experience is the result of overall impedance, this is the voltage sag at a certain amperage and frequency, it includes the ohmic resistance, inductance, and the chemical ability of the battery to deliver the current at that moment in time, it changes over the course of discharge. The actual ohmic resistance im not really sure what it is but im under the impression it is very low, not that it is of much importance, a significant amount of heat might or might not come from the chemical reactions in the battery itself.
Joe

 

[safe]

Xyster's failure was more like too high of a discharge current, and is not directly related to letting the cell voltage get out of the allowable range.

But each individual cell is naturally going to have slight variations in the tolerance of overdischarge. I really don't know what the exact conditions were that caused his battery problem, but his pack is really quite huge so for him to be able to draw more current than was possible would not be my first choice as far as an explanation. Could it be possible that one of the cells simply could not tolerate the low voltage state that he got it into? One bad link breaks the chain?

 

[safe]

...a significant amount of heat might or might not come from the chemical reactions in the battery itself.
Joe

Well that's what I'm trying to learn.

Is the meltdown because the resistance breaks down across the battery (it goes from BEING a battery to being a wire) or is the battery never acting like a wire, but simply heating up and melting?

If a "cascade faliure" takes place then that "seems" like something where the battery loses it's resistance and turns into a wire which then "cascades" that energy on to the next cell and on and on. Otherwise you could prevent "cascade failure" by simply thermally isolating the cells. (which might work for all I know) If this were the case you might be able to embed the entire pack in some fire retardant material to completely eliminate "cascade failures".

 

[xyster]

Xyster's failure was more like too high of a discharge current,

I don't think so, not directly anyway. The discharge rate was within spec, and much less than Patrick Mahoney has run his same cells at.

At the time of the event, the subpacks were 12, 2.2ah cells in parallel for 26.4ah. Current was limited by the controller (and verified by my onboard ammeter) at 35 amps. So discharge rate was 1.33C.

The cells are spec'd for 1.5C:
http://www.all-battery.com/index.asp?PageAction=VIEWPROD&ProdID=1643
"Max. Discharging current 1.5C ma (for continuous discharge) "

Lithium-ion are supposed to be good for about 2C no problems.

Patrick runs 8 of these cells in parallel, with the 20 amp controller for 1.13C, and has written that he's had no problem pulling 1.8C from them.

 

[safe]

Battery becomes Wire?

or

Battery becomes Paper Weight?

Which of these two descriptions is more accurate when a cell has a "meltdown"?

 

[Leeps]

Im under the impression that xysters problem was due to overcurrent discharge. Overall the pack was within spec but id bet there was enough of a variation in impedance that the load was not spread evenly.

And safe what if it became a paper weight made of wire?
Personally i wouldnt use a meltded down battery neither as a paper weight nor as wire, i would throw it away and call it garbage. But seriously i think your asking if it goes short circuit or open circuit, honestly i dont know but i would assume it could do either depends on what happened on the inside. Li-ion cells that end up flaming are usually reduced to something that no longer resembles a battery. I remember reading threads from lead acid batterys that had the internal connections melt and thus go open circuit. Ni-cds i believe go short circuit quite often from dendrite formation piercing the insulating material. It all depends on the mechanism of failure, also i dont really think a simple going below threshold voltage will cause a catastrophic failure of a li-ion battery. If i took a 400 ohm resistor and drained the battery down to 0 volts it would be death to the battery but it wont blow up. However if i shorted it and brought it down to 0 volts instantly it definetely would get angry and bite back.
If i had to pick between the two i would say paper weight.
Joe

 

[xyster]

Im under the impression that xysters problem was due to overcurrent discharge. Overall the pack was within spec but id bet there was enough of a variation in impedance that the load was not spread evenly.

Could be. The bike goes over 20mph with only 5 amps, and 26mph with 10 amps. So I'm planning to run a test by running the batteries down to around 3.7v again, but this time at a tame, 5-10amp rate. I was purposely testing the pack's limits when the accident happened. I wanted to be sure I could run up steep hills full-bore for miles at a time. I got the answer :-( .... In their 15p arrangement, these seem to be great cells for 0.5C-1C continuous; bursts over 1C for a couple minutes seem to be A-OK. As these age, I'll replace subpacks one-by-one with a more robust, power-dense lithium chemistry. And that's another advantage of separate chargers for each parallel subpack.

 

[fechter]

Maybe Xyster ran the whole pack down so low that one of the sub-packs went into reversal. Bet lithiums aren't happy with that.

My experience is lithiums usually fail in the "paperweight" mode and go open. If you smash it, it could short.

 

[xyster]

Maybe Xyster ran the whole pack down so low that one of the sub-packs went into reversal.

That's what I was thinking. Before I set out for the fateful ride, the resting subpack voltages were about 3.7v. I figured from graphs I saw at BatteryUniversity and other places that the cells still had usable charge remaining down to 3.0 - 3.3v resting. This does not seem to be the case! Later, I found this, which fits my experience much better:
http://www.rcgroups.com/forums/showpost.php?p=3272601&postcount=11
"This is a test I did for capacity vs. Resting voltage.

72 degree pack and room temp.
The (999mAh) pack was discharged each time at 1C, to a 3v per cell cutoff.
Then I added a fixed percentage of charge, and measured the resting voltage after 12 hours.
Next I discharged the pack and verified the percentage of capacity.
At the end of all testing I charged the pack fully and did a accuracy / repeatability test.
The repeatability of the test was over 98%.

4.20v = 100%
4.03v = 76%
3.86v = 52%
3.83v = 42%
3.79v = 30%
3.70v = 11%
3.6?v = 0%

Capacity below 3.7v "resting" is not usable for flying, it's where the battery voltage dump begins."

As you can see from the 3s graph below (from the same link), the 3.6-3.7 volts commonly listed as "nominal" is actually at the 0% point. So in battery-world jargon, I guess "nominal" really means "zero", and not "within the range of acceptable".

Now, it seems to me quite likely that some cells (like the two that ruptured) went into reversal while others in the same subpacks did not (like the ones that were hot, but OK, and that I'm still using today).
Damage to these cells I'm still using could explain my present underwhelming performance when at >1C for > a few minutes.

 

[safe]

My experience is lithiums usually fail in the "paperweight" mode and go open. If you smash it, it could short.

From a design perspective it's much easier to deal with the "open" paper weight dead cell than the "closed" short circuit cell. Well unless you have a pure series string in which case the "open" cell disables the pack, but that just means you have to "stop" which given the fact your pack is melting is probably a good idea anyway!

If you build your "subpacks" in parallel and the most likely failure (which should be rare) is an "open" circuit and then you charge each "subpack" individually the odds are that you can keep it all alive for a long time without really doing much in the way of control logic.

Knowing this, the idea of a "naked" pack seems a little more acceptable than I first would have thought.

Anyone ever thought about encasing the cells in a lightweight fire retardant material to contain a "meltdown"?

Some sort of fire retardant foam maybe?

 

[safe]

http://www.beepscom.com/product_p/ba-xl-205f.htm

XenoEnergy XL-205F D Size 3.6 Volt 19Ah Lithium Battery

$18.95

What about this battery? Too good to be true? (is this one of those non-rechargable ones?)

 

[xyster]

Yes, Safe, thionyl chloride lithiums are primary (non-rechargeable).

 

[safe]

It's funny, they have them listed as "rechargable" on that website, but I just got done downloading the literature from the company to find out "yep" they are not rechargable.

Would make for a great bike that you only rode once!

 

[Lowell]

The discharge rating on those types of cells are pretty low.