How to / DIY > Everything else

Lithium batteries, the care and feeding of

<< < (2/7) > >>

mikenash:
Good advice & my thoughts too.  Cheers

BruceM:
A most interesting report, Starfire, thanks.  The need for a BMS on lithium cells is well documented engineering, though with only 12V you only have 4 series banks of cells to worry about so it's less of an issue while cells are newish. (Parallel cells are regulated as one.) 0.3 volts variation is significant for a single cell and I would expect it to get worse over time.  BMS is often ignored for low voltage strings for portable equipment where service life can be sacrificed for lower initial cost.

I agree it's interesting that just 100 ma of balancing shunt current is the typical unaided current rating of most BMS IC's.  Many BMS IC's allow external power transistors to have much greater balancing current.  A true BMS system will limit charge current based on IC feedback so a newly replaced, fully charged cell would cause the charge for the entire series string of cells to be limited to 100ma once that cell was full.  By balancing on every charge, small balancing currents can keep cells from diverging...so perhaps with matching age and use cells, 100ma is actually useful; or, designers are cutting corners and don't want to add additional hardware for increased shunt current and the heat sinking required. If the BMS system doesn't limit total charge when the cell shunt regulators say they are maxed out, it isn't a true BMS system, and cells will drift out of sync until cell murder is achieved.

I use my own design BMS for wet lead-calcium batteries (10 batteries in a nominal 120V series string) and have up to 3 amps of balancing current available per battery before charge current is throttled.  Current is then limited to whatever provides just under max shunt capability of the fastest charging battery.  In actual service, charge is never limited by my regulator shunt current, as the batteries stay closely matched and while a few individual cells sag a bit over a month as seen via hygrometer, monthly equalization for 5 hrs brings them back to matching.  If I could manage to the individual cell level, equalization could be eliminated. 

Historically,  BMS systems didn't come into play until higher voltage series strings  of AGM batteries (96 volts and higher) were used for early electric cars.  Equalization can't be done regularly with AGMs...so batteries were dying fast.  By adding a BMS of shunt regulators with charge current limited based on regulator feedback and a shunt regulator on each 12V AGM battery in series, batteries were kept in sync, and battery life was greatly extended.

I modeled my system on these AGM designs since I was planning on a 120V series string using AGM batteries.  I used Lead-Calcium batteries initially due to low cost for new hardware checkout, and found that thanks to the BMS, service life was nearly 5 years.  AGM prices never came down enough to justify their use. 

BruceM:
Mikenash, if you haven't done it already, I suggest a careful equalization, 15.1V charge on your AGM battery, 3-4 hrs.  That will recover a sulphated cell.   AGM's are murdered by sulphation and also by drying out from regular overcharge.  Natural death in small DOD applications is around 8+ years. 

starfire:
Hi Bruce.
Yes, my legacy  system is "stuck" at 12 volts. This is what I started with back in the 80s and it probably will stay that way unfortunately. All the appliances and ancillaries are 12 volt, quite a mission to change to even a 24 volt system.
The way I have understood this cell balancing is that any cell drifting high compared to its neighbours will reach the 3.65 fully charged voltage first, thereby leaving the other at a partial charge if charging stops at that point, which it must if cell damage is not to occur. Looking at the charging curve, the top most fractions of a volt represent such a very small part of the total; capacity of the cells, the loss is minimal. I have wired each battery in parallel, then ran 3 wires between cell junctions, this effectively places all cells in all batteries as one giant series parallel configuration.  As you say, this limits the problem to just 3 junctions.
These are also connected to 4 of those small Chinese digital voltage displays that are taped to the battery giving a constant display of cell condition.
I have since constructed a automatic cell monitor using 2 LM339 quad comparators. One  input is referenced to a seriesed and tapped  resistor string across the battery, the other input referenced to each cell junction. The tapped resistor voltages rise and fall with battery voltage, but remained in a fixed relationship. The cell voltages rise and fall, but are not in a fixed relationship with each other.
This allows the comparators to monitor not the absolute voltage of each cell, but the difference between what they shuld be from the fixed resistor string, and what drift has occurred in the cells,.
 By alternating between  inverting and non inverting inputs, we can have a window comparator with just a few millivolt window. The open collectors are simply paralleled through one LED to indicate a fault. The remaining two comparators detect 14.6v and 10.7v respectively.
Ill post this diagram too if its useful.
To me, this has been a better way to do this.
The other BMS issue that caused me  problems was its ability to turn the seriesed BMS FETs off as the battery voltage got near 14.5v, it would simply totally disconnectthe battery  from the "grid", giving an open circuit to the alternator, high voltage into the house wiring and killing anything that happened to be turned on at the time.  The overvoltage switch to the field works very well, and in my case, much more elegant.
The one strange thing I have noticed, the high cell is never the same one each day, it varies but so far has never exceeded 3/10ths.
To understand the balancing part of these BMS units, I removed the 100 ohm smd resistors from the dump FETs, ran wires to outboard resistor and LED combo to monitor the action. These turn on at cell voltage 3.55. Now, this is at the top of the charging curve where, at a charge rate of even a few amps, this would have little to no effect given that charging at this point is almost finished.
Still, this is my knowledge of Lithium so far.

starfire:
And Bruce, yout knowledge appears greater on this topic. I have just run the Lister for several hours and put in around 80 amps constant. The cell voltages every 30 minutes read:.
3.37, 3.34, 3.35, 3.37
3.42, 3.46, 3.41, 3.51
3.43, 3,48, 3.36. 3.57
3.52, 3.44,. 3.54, 5.55

after sitting for 10 minutes

3. 36. 3,38, 3,37, 3,38

This suggests there is some kind of self balancing happening sans BMS,?
Under charge, the chemistry is all over the place and voltage readings are invalid?
This bit I dont understand.... maybe you already know this?

Navigation

[0] Message Index

[#] Next page

[*] Previous page