Bottom balancing?

morph999

100 kW
Joined
Jan 20, 2009
Messages
1,721
Have you guys read about him? He has some good real world experience with Lifepo4 cells. He says a lot of the BMS that are in current use could actually be destroying the cells. He says top balancing is actually a bad practice to use. The reason is because each lifepo4 cell has a "knee" and once a cell hits that knee, it only takes 5 seconds for a battery to be destroyed to nothing. The "knee" of the discharge curve is somewhere around the 80 - 100 % DOD. It's the point where the battery has had enough. The reason why top balancing is bad is because when you top balance and then run your battery pack, one cell will reach it's knee point before the others and essentially, the rest of the cells gang up on the weak cell and deplete it to nothing. Bottom balancing ensures that this doesn't happen because all cells reach the knee at the same time.

The best BMS would be something where we could determine the "knee" point and then maybe count how many AH of current is stored from the knee point. Then when you use it back, the computer or meter would count back how many AH you have used ....and how far away you are from the knee point.

EDIT: Actually Jack is correct for most of the things he says however, bottom balancing can also be bad because when we charge, we have a potential of overcharging if we charge in bulk. Sorry, I'm just learning this stuff. Maybe you guys already know this. Maybe this can be useful for the average lifepo4 user. The best way to ensure that you don't overcharge is by top balancing, but the best way to ensure that you dont' destroy cells when discharging is bottom balancing. Hope that makes a little sense. The best of both worlds would be top balancing with LVC for each individual cell.

Here is some quote from his blog if you want to read it in his own words
http://jackrickard.blogspot.com/2009_11_01_archive.html

By very carefully charging each cell to precisely the same 4.000 level, I did indeed "balance" the cells - at least at the top of the charge.

But as I discharged the cells, they reached any arbitrary point on their discharge curve at DIFFERENT times. So at the end of the charge, where the knee of the discharge curve turns sharply down, they became more UNBALANCED at the bottom.

The graph below shows the number of seconds a cell has at a 100 amp discharge rate to 3.00 vdc from a full charge with all of the cells balanced at the top of charge.

The problem here of course is that some cells go over the knee first and start down the steep discharge wall at the end before the others. This has a very bad result. The cells still up on the plateau, making current, drive current through this smaller capacity cell and drive it down to zero volts and ultimately to destruction..

So I was repeatedly destroying cells by carefully top balancing the cells, precisely as a current shunt balancing circuit would, and then discharging past the knee of the discharge curve. The other cells turn on the weaker one and eat it like a pack of wolves.

Worse, your overall pack voltage masks all this - remaining up in the supposedly safe area.

The solution appears to be BOTTOM balancing. With all the cells discharged, I replaced the dead one, and balanced all the cells at 2.90 vdc. Then recharged the pack to 87 vdc (3.625 vdc per cell).

Now the cells are very unbalanced at the top - some slightly over 4 volts and some quite under the 3.625 average.

But I don't care about the top. I don't lose cells at the top, and we're charging at 20 amps. During discharge, even the GEM can go over 200 amps of current. That is a 10x more violent event in the life of a battery. And a weak cell can drop from 2.8 to 2.0 to 1.0 to 0.0 in a matter of a dozen seconds or so at 200 amps.

This pretty much explains why I was able to lose cells on the GEM while balancing to the nth degree, but the Speedster, whose cells have never been balanced at all, wheels merrily along without problems.

In fact, we recently completed a 107 mile test drive with the new tires and really did push the little car to the limit. At the end, all of the cells measured between 2.8 and 2.9vdc in quite balanced fashion - at the BOTTOM of the discharge curve. Things were good BECAUSE we had never top balanced.

What I conclude from this is that these simple current shunt balance circuits are not only a needless expense and a fire hazard, they are doing exactly the opposite of what they purport to do. They are UNbalancing the pack at the bottom where it matters, and potentially leading to the untimely death of cells.

So we're still in search of the perfect Battery Management/Monitoring system. But the current shunt balancers are certainly not it. Save your money, and your batteries.

Jack Rickard
 
I think I did learn something by reading Jack's blog. When I get my lifepo4, I think I should check all the voltages first to make sure they are at least close to each other. That way when I use my 12v charger, I won't overcharge them.
 
Hi,

Here is a quote from his blog if you want to read it in his own words
http://jackrickard.blogspot.com/2009_11_01_archive.html

So this week, I took some time, and brought my really pretty nice Agilent 5 1/2 digit multimeter from the back bench, along with the test device I built for the Mini.

It took 2 full days, but I very precisely hand balanced all 24 cells to precisely 4.000 volts.

morph999 said:
Have you guys read about him? He has some good real world experience with Lifepo4 cells.
lifepo4 should be charged to 3.6v or 3.7v so he was either overcharging the cells or using a different chemistry.

morph999 said:
He says a lot of the BMS that are in current use could actually be destroying the cells. He says top balancing is actually a bad practice to use. The reason is because each lifepo4 cell has a "knee" and once a cell hits that knee, it only takes 5 seconds for a battery to be destroyed to nothing. The "knee" of the discharge curve is somewhere around the 80 - 100 % DOD. It's the point where the battery has had enough. The reason why top balancing is bad is because when you top balance and then run your battery pack, one cell will reach it's knee point before the others and essentially, the rest of the cells gang up on the weak cell and deplete it to nothing. Bottom balancing ensures that this doesn't happen because all cells reach the knee at the same time.
Stopping the pack from discharging before cells are damaged by monitoring individual cell voltage is how a good BMS works. (You could also monitor groups of parallel cells).

NOT by "bottom balancing" and then attempting to stop the pack discharge exactly at the point where all the cells are exhausted. That would be much more difficult and much more likely to result in failure (which he corroborates).
 
yeah, I didn't fully understand how lifepo4 works until I read his blog, though. I was half-way there but after reading about his failures, I thoroughly understand how it works.

I was planning on top balancing my cells. I probably wouldn't have ran into any problem since I was only going to use 50 % though but now I realize that top balancing them really doesn't help with the cutoff voltage. I think I'm going to buy two of those 8s cell monitors . That would be about $60. I might not even bother hooking up my Turnigy watt meter. The watt meter is kind of pointless for me now. I'll just monitor the individual cell voltages.
 
If this guy is charging to 4V per cell, then he clearly doesn't understand the chemistry of LiFePO4 and it's not at all surprising that he's had cells fail.

Quite a few of us on here have a lot of experience with LiFePO4 now and we've found that charging to 3.65V per cell is pretty close to the optimum. Charging to anything over about 3.8V per cell is asking for cell failures.

There's nothing mysterious about charging LiFePO4 cells, you just pump current in at an acceptable rate (usually around 1C to 2C for a standard charge) until the cell hits 3.65V. At this point, you just clamp the cell voltage to 3.65V and hold it there.

Charging cells like this is foolproof, they won't blow up or get damaged and will last a long time.

Jeremy
 
morph999 said:
yeah, I didn't fully understand how lifepo4 works until I read his blog, though. I was half-way there but after reading about his failures, I thoroughly understand how it works.


Nope. Sounds like you're just as confused as he is now. Likely more-so.
 
The only thing I got out of that is that he learned a cell-level LVC is essential, since the cells have different capacities.

Could I say... duh?

Maybe there was more to it than that, besides that shunt balancers can overheat and destroy cars. Possibly over-charging might be reducing the capacity, but that doesn't necessarily result in a 0.0 volt cell failure in my experience.
 
I think that in general, for longest cell and pack life, it'd be better to never charge the pack past maybe 90%, or discharge it past 30 or 40%. Unfortunately that means you'd basically only get at best half the capacity of the pack as usable power, so it would take a pack twice the size of what it "should", and cost would be double. AFAIK that's what existing hybrids do like the Prius, which should guarantee longer life, less likelihood of killing a cell.

It's not really practical to do it that way, unfortunately, for most of us.

There is another way of balancing cells, and there is someone working on a BMS to do that (Gonzo over at DIYEC): shuffle charge from stronger cells to weaker ones. It's complicated, not that efficient, and expensive, so it is probably not worth doing that either. But it would be "better" in some ways than the top or bottom balancing.
 
what i get from the quoted blog segment has nothing to do with batteries.
he repeatedly comes to the wrong conclusion the result of faulty deductive logic, for one:

So I was repeatedly destroying cells by carefully top balancing the cells

no, top balancing gots nothing to do with it.
most certainly the problem of cell reversal is simple over discharge.
even most rank nubes know better, it's one of the first things learnd that it should be avoided at all costs.
keep a respectful distance from the knee to prevent the cell from falling off the cliff if ur at all interested in longevity.

i cringe a little every time i read someone post on here or elsewhere that they drained their pack to the bottom "just to see what it can do".
it's like pumping the max amps thru a fuse, just to see what it can do. :roll:
 
Ignorant and uninformed comment removed.
 
amberwolf said:
There is another way of balancing cells, and there is someone working on a BMS to do that (Gonzo over at DIYEC): shuffle charge from stronger cells to weaker ones. It's complicated, not that efficient, and expensive, so it is probably not worth doing that either. But it would be "better" in some ways than the top or bottom balancing.


It doesn't matter how the cells arrive at the same voltage. A transfer type BMS is what you're talking about. These are the most efficient BMS by a longshot, as it's the only type of BMS that doesn't just resistively bleed off the current into heat. It also corrects cell balance at roughly twice the rate for a given balance current value, due to raising the low cell as it bucks the high cell, so things meet somewhere in the middle rather than at the lowest cell only.

All the energy in a regular BMS has to leave the BMS as heat. A transfer type can use high current levels, and still be sealed in a box because it's only got a small fraction of the heat to deal with.
 
i dont think he is a noob...

he is charging the THUNDERSKY-cells to 4,0V...

look at the TS-specs...TS say, that the cells can be charged to 4,25V...

than he says, that the BLUSKY-cells can only be charged to 3,6V...that right...see the specs of SE...
i dont know why he is calling them BLUESKY...they are SKY ENERGY

watch his video about the cells...there you can see how the cells look inside...
 
I wonder why his other vehicle (which also does not have a low volt cutout) is ok while this one kills cells. I agree that it seems like simple over-discharge.. Its not a lot to go on to try to make a conclusion, two batches of cells in two different cars.
 
This is my interpretation:
The reason rick is charging to 3.8V or 4.0V is because he's trying to simulate the action of a shunt balancer where it kicks in around at that voltage. My guess is that's what point the "Rudman Regulator" kicks in that he's showing doesn't work. He makes sure that all the cells are at exactly 4.0V - where he says this is the point at which the shunt balancer would put the pack in a state that's considered "balanced" since any more voltage would just get shunted off. In his defense he does say that the cells should be at 3.625 nominal.

Then I think he's approaching this by an overall pack voltage point of view, driving the voltage down until it would be an ideal voltage if each cell were exactly the same and had the same Ah capacity. Also, I'm pretty sure he's assuming only HVC shunt voltage regulation on the BMS with no LVC control, in which case obviously you'd destroy the packs.

If you watch one of his later shows on evtv.me, you'll see he accepts that he does in fact need a BMS but he feels that there is nothing out on the market that quite meets his needs, and suggests that he's going to try some solutions in future shows, one of them by using an Arduino.

I've watched quite a few of his shows, but unfortunately it doesn't seem like he's doing any as of late. I really appreciate the fact that he's fearless enough to question the accepted opinions on battery management, etc., and spend lots of his own money doing it even if it takes a little bit of discovery for him to find out aspects he hadn't previously considered. He's also always quick to say that "He could be wrong, but.." Regardless, anyone kind enough to take the time to send a full page email reply to me about how i can get started on my first project is cool by me. I'm a newb and that's my two cents.
 
liveforphysics said:
It doesn't matter how the cells arrive at the same voltage. A transfer type BMS is what you're talking about. These are the most efficient BMS by a longshot, as it's the only type of BMS that doesn't just resistively bleed off the current into heat. It also corrects cell balance at roughly twice the rate for a given balance current value, due to raising the low cell as it bucks the high cell, so things meet somewhere in the middle rather than at the lowest cell only.

All the energy in a regular BMS has to leave the BMS as heat. A transfer type can use high current levels, and still be sealed in a box because it's only got a small fraction of the heat to deal with.
I hadn't thought about that--I guess it does make it a lot more efficient, then. I was mostly thinking about the amount of cell interconnections you'd need to have as higher-current (and thus larger) wires, and the potential for future problems if any of that wiring were to fail shorted later (since it would all need to go to a central point to be shunted as needed).
 
RoughRider said:
i dont think he is a noob...

he is charging the THUNDERSKY-cells to 4,0V...

look at the TS-specs...TS say, that the cells can be charged to 4,25V...

than he says, that the BLUSKY-cells can only be charged to 3,6V...that right...see the specs of SE...
i dont know why he is calling them BLUESKY...they are SKY ENERGY

watch his video about the cells...there you can see how the cells look inside...


The spec sheets for the TS cells are typo's or BS to try to get them to meet the rated capacity at the cost of cell life. We've been through this TS spec sheet thing a hundred times... This guy is totally clueless, and apparently trusting his spec sheet... There is no magic LiFePO4 chemistry that behaves in a different voltage range. The voltage range is set dead fixed by the chemistry, and has nothing to do with cell design or construction. He is destroying his cells through ignorance, and proudly sharing it with the world. lol
 
liveforphysics said:
morph999 said:
yeah, I didn't fully understand how lifepo4 works until I read his blog, though. I was half-way there but after reading about his failures, I thoroughly understand how it works.


Nope. Sounds like you're just as confused as he is now. Likely more-so.

huh? Did you read my "EDIT" part ?
 
Hi,

amberwolf said:
I think that in general, for longest cell and pack life, it'd be better to never charge the pack past maybe 90%, or discharge it past 30 or 40%. Unfortunately that means you'd basically only get at best half the capacity of the pack as usable power, so it would take a pack twice the size of what it "should", and cost would be double. AFAIK that's what existing hybrids do like the Prius, which should guarantee longer life, less likelihood of killing a cell.

It's not really practical to do it that way, unfortunately, for most of us.
I'm not sure if they would have used the exact same values for LiFePO because the LG Chem cells they went with use manganese-spinel but I expect the figures are similar (probably closer than NiMh). GM spent millions figuring this out and came up with 85% and 25% (80% and 30% according to some sources):
http://en.wikipedia.org/wiki/Chevrolet_Volt
GM plans to station charge the lithium-ion battery to a state-of-charge (SoC) range of approx 85%. Then once the battery depletes to a precise low set-point (<25%) the on-board ICE powered generator will maintain the state of charge of the battery between the lower setpoint and an upper set-point above the 30% SoC level

http://discovermagazine.com/2009/apr/09-can-smart-tech-keep-chevy-volt.s-battery-running
Enough spare capacity is built into the pack so that it has to be charged to only 80 percent of its theoretical capacity to provide a 40-mile driving range. By not charging the battery pack to the max, its life is prolonged. For similar reasons, the pack is never discharged to less than 30 percent of its capacity. GM claims the result is a battery lifetime of 10 years or 150,000 miles without any noticeable deterioration.
Probably accurate as its covered by warranty.

Technical reasons for manganese-spinel (according to LG Chem):
http://www.designnews.com/article/162063-GM_Selects_LG_Chem_to_Build_Volt_Batteries.php
Engineers at Compact Power cited two technical advantages inherent in their team's battery design. The manganese-spinel chemistry combines with battery separator technology that enhances safety, they said. Known as a Proprietary Safety Reinforced Separator, the semi-permeable membrane is coated with a ceramic material, which is said to make it mechanically and thermally superior to other separators.

Also key to the company's technology was its use of a "stack-and-fold" configuration in a laminated package, which could provide GM with easier manufacturability. The stack-and-fold concept is used as an alternative to the well-known cylindrical design of conventional batteries.
 
^ one of the reasons I plan to buy the TS 40ah cells. I don't think I will be using them down past 20-30% of their capacity. I imagined that the pack would only provide roughly 34-36ah anyways. I based that on Justin's data on the x-ca, wherein he only was getting about 9ah out of a 10ah pack...or something along those lines. I remember it was around 90% of the nominal rating.

Personally I will be investing in a good BMS. I want to be able to monitor as well though, so I will probably go with that 8-cell monitoring device from hobbycity.

This is all way down the road however :)
 
It's theoretically possible. If you get 1000 cycles of 20 miles or it's equivilant such as 2000 cycles of 10 miles that's 20,000 miles of riding. Even for me that's about 6 years of riding. But realisticly, for me at least, I won't have much use for my batteries when range drops by 20%. So I am going to be real happy if I get 10,000 miles out of a 36v 20 ah ping. hafway there after 2 years, so I expect at least 4 years out of the battery. Mabye longer though, if I find I like 48v enough when the new one comes.
 
Balancing your battery pack at the low (drained state) makes perfect sense. I'm going to try that to see how much variance i got at the bottom (drained state) of my pack.

My typical 16s pack is full at 54V nominal, and the drained state is about 51 volts, I will monitor any lower cells (or higher) and manually balance them, then top-up the pack with my charger and see if that holds. It's usually only one or two cells that are out by only a tiny amount.

95% of the time my pack will stay balanced without charging through a BMS.

Using a single, cell phone charger, or an automotive headlite to slightly drain a cell will fix the problem in 5 minutes.
 
Balancing "at the bottom" can be dangerous, as you can lose cells that go "over the cliff" too easily. Jack's problems stem from not using cell-level low voltage protection, plain and simple. As long as you do that you can balance or not balance, at any SOC, and it matters not. You will stop killing cells. Balancing once in awhile will ensure that at least the cells that are close to the same capacity will get back to the same SOC. To do that, it doesn't really matter whether you balance at the top, the middle, or any other point.

I also wholeheartedly agree with Jeremy that it is ridiculous to do what many TS users are doing which is to simply charge until the first cell hits 4V, or 3.8V, or any other value above about 3.60-3.65V. All this does is shorten the life of the cell, regardless of the BS that TS is putting out. Also, if you are using higher charge currents, the high cell will hit that HVC point sooner than it would if the charge current is at a lower level. Shutting off at this point simply leaves the whole pack under charged, with the "high" cell at only about 70% full. Some would argue that's fine, it will mean the cells will last longer, but that's not right either, if you are letting this very same cell hit 3.8V-4.2V on every charge.

What Richard and I decided to do with our new BMS revision, was give the user the option of balancing or not balancing, but even if you don't balance, the supply current is managed at the end so that no cell is allowed to go over a set limit, and what ends up happening is that at least the high cell will receive a full charge. This is because the so-called high cell's voltage is held at the CV point, and the current is allowed to taper down to a level where the shunt is fully on. If the shunt is in full bypass, that means no more current is going into the cell, so it is full. At this point the control logic shuts off the 12V regulator, which causes the FETs to cutoff the supply/charger current. If balancing is enabled, instead of shutting down, the control logic now sets a timer to allow additional time for the rest of the cells to catch up, and also get "full". The amount of balance time is selectable from 15 minutes to 4 hours. Once it times out, the control logic shuts down.

Full is a relative term, as well, because it is really just a voltage set point that is held to, to start the current reduction, which happens because the cell can't accept current at the same rate without the voltage rising. If you limit the voltage, the current will steadily drop, pretty much at the same rate the voltage rose, during the CC phase. Anyway, 3.65V is typically used as the CV point, as it is around there that the voltage suddenly starts to rise at a faster rate. If you hold the cell voltage at this point, and let the current taper off, The cell will end up at the 100% charge level. What many have found is that by picking a slightly lower CV point, like around 3.60V, there can be a significant increase in cell longevity. At 3.60V, the cell is about 97-98% level, which is not hardly noticeable, but it apparently make a big difference in cell life.

A bulk charger has to set the CV point at the pack level, so it is the cell's CV number times the number of cells in series. Because of this, any significant difference in cell capacities and/or states of charge, will cause a cell to hit the 3.65V "knee" in the charge curve a lot earlier than the rest of the cells. Once it does, the voltage for that cell will simply keep rising as the cell tries to absorb the current at the full charge rate. Even stopping the charge process at this point, which is what those proposing to simply "charge to HV cutoff" would do, still lets this high cell go too high, which simply means this cell will eventually get weaker and weaker. In our mind, it is far better to let the lowest capacity cell at least get fully charged to 3.60V, and then periodically let all the cells charge until they are at 3.60V, so that you are then assured that the first cell that does hit the CV point is the one with the lowest capacity. By combining this with cell-level low voltage protection, you are assured of getting the maximum "safe" range out of the pack, without doing anything to decrease the cell's longevity.

-- Gary
 
Gary- Your entire post was packed with wisdom and tips to help folks have long living happy batteries. You can have batteries last a very long time while still using ~80% of the storage capacity in the cells on each cycle.


For newbs out there who might think the advise from this Jack Rickard guy's site is something to listen to, it's not. I don't think he is retarded or something, but I do think he managed to make an unfortunate series of misinterpretations from his observations with batteries, and has come to some extremely incorrect conclusions which ensure premature failure for batteries. For optimal battery performance and lifespan, you can essentially just change everything he says on his site to say the opposite, and then you've got good advise. Or, even better, carefully go through Gary's outstanding post above, and you've got everything you need to know.
 
Yeah. I'll take listening to Gary..... Maybe I'll never understand all, or even much of it. But I'll take it as gospel for sure. Not using cell level lvc protection is the problem. If your understanding only gets that far you are on the right track. Even an ebay vpower hong cong battery has lvc on each group of paralell cells.
 
Back
Top