Will Li-Ion cells in a pack equalize?

e. ~3.7V

Yes, there are companies that parallel cells, but it is debatable that this is good. After all, if the manufacturer thought a fatter, high capacity cell was a good idea, they would just make that cell into a product.

I've disassembled several laptop battery packs, and all contained paralleled cells, with several groups of paralleled cells connected in series. They all were made from 18650 size cells. I believe this size cell is by far the most popular, so it probably has the best or nearly the best energy density and lowest cost. It's no surprise to me that manufacturers would use these cells in parallel rather than larger cells which would probably have less energy density and higher cost for the same capacity.

The Tesla Roadster electric car uses about 6800 size 18650 cells in a

Roy Lewallen

I wouldn't confuse battery pack assemblers with battery manufacturers. Two different beasts generally. Now it may be that the yield is poor on a fatter battery, or it can't cool as well, or any number of other reasons why they don't make a fatter, higher capacity cell. But to say paralleling cells is fine because there are packs like that doesn't mean the battery cell manufacturers are approving such construction. If two in parallel is fine, then how about a hundred?

Fine.

To power larger devices, such as electric cars, it is much more efficient to connect many smaller batteries in a parallel circuit rather than using a single large battery

To obtain higher ampere-hour (Ah) ratings, two or more cells are connected in parallel. The alternative to parallel connection is using a larger cell. This option is not always available because of limited cell selection. In addition, bulky cell sizes do not lend themselves to build specialty battery shapes. Most chemistries allows parallel connection and lithium-ion is one of the best suited.

hat

lls

Yes, wiki is the source of all wisdom.

Motors are basically current fed devices and it would not be too hard to come up with a multiple switcher scheme to parallel a few battery packs of series cells by summing current. Such a scheme would also give your electric car a "limp home" mode when one pack fails.

In good engineering, you should be able to predict the state of every component, the range of conditions it will experience, etc. This is why we don't put capacitors in series for instance. When you start to parallel cells, how do you insure wth 100% certainty the power sharing. Or do you just shrug it off and pronounce it consumer electronics and quality be damned?

Every time I see a product recall, I just roll my eyes and wonder what happened to good engineering. And this is from a person that caused a recall once. You can't be too careful out there.

that

gh

o a

in

ze

rly

cells

he

xt -

And almost all laptop makers. When you see a battery pack with length/ thickness more than 4:1, there are parallel cells. Laptop battery packs are mostly 8:1.

le that

high

nto a

ed in

size

early

hat

r cells

the

in a

rs.

er

say

text -

..

Oh yeah, these laptop makers know what they are doing. Let's see. We have the Apple bulging battery fiasco. The Apple battery fire. The Sony battery melt down. Need I elaborate? Most of these laptops are ODMed these days, and who knows what engineering skill the ODM has.

I can just hear the battery pack design review. Engineer: "Paralleling cells may lead to a shortened battery pack life." Marketing: "Great. Just make sure it lasts longer than the warranty."

The operating time improvement for the paralleled pack isn't really the driving factor. Rather, they DC/DC can be made cheaper if the pack voltage is lower. If you need to handle say 6 cell and 9 cell, that is a harder design than normal 6 cell and fat 6 cell. Also the higher voltage components cost more.

Statistics are on your side with parallel connection. Statistical variations between cells will average over the cells in parallel. Also depends on the failure mode you want. With series cells, one weak cell can significantly reduce the pack capacity. With parallel cells, you have x-times the current to start the fire when one shorts.

~3.7V

Also the impedance of parallel cells tends to be lower than a single cell of the same rating.

e. ~3.7V

ty.

Statistical variations average? How many posts are there about cell matching when paralleling. You can't have it both ways.

Now I have to think about the weak cell reducing capacity. How does this differ in the series and parallel case? If the DC/DC has enough operating voltage swing on the input, the reduced voltage of the series pack can be compensated for, though the capacity will be reduced. But it isn't like the weak cell will necessarily make the buck go out of operating range, i.e. a double penalty.

What is the guideline for protector chips when paralleling cells?

.e. ~3.7V

ity.

Lower impedance but at a higher current. Seems like a wash to me. Further, you need better components to take advantage of the lower impedance if efficiency is the goal.

Thus far, other than a slightly more complicated buck converter for the bigger stack, nobody had stated a good case for paralleling cells.

Well, let's see. If the 6800 cells in the Tesla Roadster were hooked in series, you'd have to deal with about 25kV. I'm afraid that would involve a bit more than a "slightly more complicated" buck converter. Or, I suppose they could get some custom made cells with about 68 times the capacity of an 18650, and wait for development, delivery, and debugging. (And after all that they'd probably find 68 paralleled 18650s inside each of the cell cases. . .)

If it were up to me, I'd very likely make the same decision the Tesla engineers did.

Roy Lewallen

In the series mode, if the weak cell fails towards a short circuit, the charger could easily compensate it if the pack voltage is significant. However, if the cell fails to open circuit, the charger can not be compensated for it.

In the parallel case, a cell failure towards open circuit is harmless (except for loss of capacity), however a sudden cell failure towards short failure would be catastrophic when the charge in the parallel cell would be dumped in the failed cell, so some kind of fuse would be require in each cell.

If both parallel and serial connection must be used due to the large number of cells, should a group of cells be connected in parallel and then each group in series or put enough cells into a string to reach the final voltage and then connect the strings in series at the final voltage level ?

At least the parallel connected strings would not suffer from the catastrophic charge dump from other cells if suddenly a single cell goes short circuited.

Paul

I think you mean ". . .then connect the strings in parallel at the final voltage level".

Seems to me you'd still have a problem if one cell went short. Then the series string it's in would have a lower voltage than the other strings, causing the others to dump into it.

I'd also think the alternate scheme would be harder to manage. The few charge controllers I've seen in laptop packs treat each parallel group as a single cell, and monitor and control each of these "cells" individually. In the alternate scheme, each actual cell would have to be treated separately, or risk voltage imbalance among the cells in each string which I believe can be pretty serious with Li-Ion cells.

Roy Lewallen

You missed my post regarding motors being current fed devices. Thus you could sum the output of current mode (hysteretic) converters and not need to make one huge stack. This would also give you a "limp home" mode.

I was assuming you never paralleled two series strings. The possibilities for bad things to happen would be greatly increased.

With a series stack, one cell shorting out isn't an issue in theory, but I sure wouldn't want to mess with it.

At some point, you have to wonder if these notebook peddlers view recalls as the price of doing business?

Cell protection ICs will shut down a battery if one cell in a series string drops below about 2.5V, or goes above about 4.35V.

~3.7V

I use a PTC between each group of 3 parallel cells. This is a cell manufacuterers (Moli) reccomendation.

(snip)

Charge state largely equalises. Re-read my earlier post.

As series strings get longer, the loss of accessible capacity as a result of SoC inequality renders the scheme unusable. And that is without considering the implications re pack protection/monitoring electronics,