LiFePO4... Why Use A BMS?

[safe]

LiFePO4... Why Use A BMS?

I just don't get it...

What's really great about LiFePO4 is that:

The cells automatically balance themselves much like Lead Acid chemistries when charging.
(so no NiMh cell reversal problems)

You would have difficulty overdischarging them since they can usually handle high discharge rates.

...given these abilities and assuming that you don't do something stupid in designing your bike (like expecting to use 10C all the time) I don't see any reason to use a BMS (Battery Management System) for LiFePO4.

I can understand the manufacturer of the battery packs wanting to build in preventive measures so that the packs become "idiot proof" so that the end user is unable to drain too fast or charge too fast, but if you are educated in LiFePO4 then you should be able to configure things correctly and have no problems.

Can anyone explain why a properly configured bike using something like a 5C maximum discharge rate and a charger system that does not charge too excessively would have any problems going without the BMS protection?

 

[giveahoot]

This is a really good question, when you consider the chemistry. I hope someone can answer this in greater detail. Maybe it's a failsafe for a poorly designed controller? Wouldn't a fuse serve enough of this purpose much more cheaply?

 

[safe]

Wouldn't a fuse serve enough of this purpose much more cheaply?

Can you imagine the uneducated person that buys a battery pack and uses it incorrectly and blows the fuse? They will claim the battery "broke" and want to return it. So it just seems like the manufacturers are trying to make sure that the sale is completed and that it's "idiot proof" once out the door.

I've heard cases where battery packs have been sold without BMS systems and the users did stupid things with the packs that basically destroyed them and then quickly returned them for a refund.

How can a battery pack seller offer a warranty without a BMS?

The acronym for BMS should be changed to "CYA" or "Cover Your Ass" from product returns. :lol:

 

[Ypedal]

Self discharge, charger malfunction, severe imballance, etc..

Cycle Life !!!!

If you expect that expensive battery pack to last 1000 cycles + , you better make sure that every single time you use it.. you do it within spec, otherwise you will never see famed 2000 to 3000 !!!!

 

[giveahoot]

Longevity is probably the best reason. Although, I'd think with a quality controller and proper care you should reasonably expect to come within 20 percent of the BMS.

 

[recumbent]

I'm not getting a "BMS" for my 48 volt Lifepo4 battery pack, but will indeed get a "Cycle analyst" to monitor everything.

I think it's for those that only get 8 AH rating batteries to save money and always run out of juice before getting home I know cause I've done it with my NICAD's often

 

[safe]

Self discharge, charger malfunction, severe imballance, etc..

Self discharge would occur whether you used a BMS system or not, that happens when the battery is just sitting there by itself.

Charging malfunction is possible, but that gets back to the "intelligent user" idea... if the user does not do stupid things or have a broken charger the charging process should be fine. (I've never heard of a charger "breaking")

Severe imbalance should not be a problem if I understand LiFePO4 chemistry correctly. When one cell is full it simply allows the next to suck up what it can't take. Just like with a Lead Acid battery it tends to self balance when charged. Some issues about series verses parallel might exist that I don't see right now (maybe someone could postulate a potential problem) but I just don't see any problems.

It seems to me that the Li Ion batteries gave everyone such a paranoia about Lithium (due to it's "Three Mile Island" type behaviors) that the BMS is still hanging around as a sort of "double security" that we might not really need anymore.

 

[safe]

I think it's for those that only get 8 AH rating batteries to save money and always run out of juice before getting home

Does overdischarge cause excessive wear?

Normally the controller cuts out when the voltage drops and that alone should protect the battery assuming that it's configured correctly. This might be a design issue of the controller (the precise cut off point) as opposed to a need for a BMS. Lead Acid batteries have the same issue... the deeper you discharge them the faster they wear out.

It seems to me that a well thought out system should work perfectly without a BMS... :?

(it's as though there is a logic being implemented in the battery pack that is separate from the controller. If the controller works correctly then the BMS is redundant)

 

[SniperGaulois]

how about if one of the cells fails in the middle of the series/array so that the circuit goes half way, and then is discharge/recharged with two or three times the power...being a bit stringent for reasons...you would probably notice the voltage change on your bike.

probably don't really need a BMS then...as long as you have regulated input and output fuses? does it protect each cell in case there is a short circuit somewhere in between them?

 

[GGoodrum]

You are wrong in your assumption that LiFePO4 cells do any sort of "self-balancing". That is simply not the case. Also false is the notion that you can't hurt these cells. I'm just starting to do long-term testing of LiFeBatt-based packs, but I have done extensive testing of a123 cells over the last two years. While these are extremely robust, it is actually quite easy to permanently damage cells by over-discharging them. If an a123 cell is allowed to get too much below 2V, it can be recovered, but it will lose 10-15% of its capacity. You can tell if a cell is damaged because right off the charger, a healthy cell will hold the voltage close to whatever it was charged to (usually 3.65V...). A damaged cell will quickly drop to somewhere around 3.30-3.37V.

Not only will "stressed" cells lose capacity, they also will have much higher voltage drops under load. This means their "C" rating has been reduced significantly. Even putting stressed cells in parallel with healthy cells won't help much. I haven't checked this, but an a123Sstems engineer told be that stressed cells also only end up with 25-30% of the longevity of normal cells.

There are lots of ways that even carefully "matched" cells can drift out-of-balance with each other. Internal resistances change over time, and at different rates, due to a number of factors like temperature diferences, for instance. In any case, I've seen packs with healthy cells still drift apart. This can poltentially be a problem if the pack is run out close to the end of the capacity. One characteristic of these sorts of cells is that they tend to have fairly flat discharge curves/ What this means is that the power is very consistant, all the way up to the end. The power feels the same 10 seconds from the end, as it does in the beginning. Then, the power drops quick, like a switch was turned off. If you do this with a pack with unbalanced cells, you could end up with cells that get damaged because they were over-discharged.

I'm not a big fan of the RC-styled balancer circuits used in most of the Chinese BMS designs, because all it does is bring down the level of strong cells, to the level of the weakest. What I do feel is needed, however, is low voltage potection for each cell. The cell junctions should also be brought out of the pack so that at the very least, you can individually charge each cell to 100% once in awhile.

-- Gary

 

[Doctorbass]

I would agree with GGoodrum.

Also, I really think that the best way to use LiFePo4 cell would be:

-Building a pack with matched capacity by measuring them before to do the matching serie/parallel.
(By this way, You are sure that each parallel group will discharge at the same time.)

-Then, to place each cells to allow a good sharing of the heat between each others (close temperature)
(by this way you preserve the original discharge curve of each cells and help them to discharge at the same time, and prolong their cycle life AND will get the max Ah your pack could have) Remember: the weakest cell or parallel group REPRESENT THE MAX CAPACITY YOUR ENTIRE PACK HAVE

-The pack should have a LVC not necesarely a BMS.
(by this way, there would be no cells that goes lower than the min voltage)

-Charging them with an individual charger (DC-DC, simple charger(like the 2A), etc
( this is the faster way to fully charge a pack. no cell balance needed or loosing energy by the dissipaton loss of the balance process, and each cell will top to the max recommanded voltage faster and will not depend on a decreasse of the current to allow a balancing process)

Doc

 

[Deepkimchi]

Just had an interesting experience. My Flintstone charger was about 2 hours into charging, all but 2 cells were already at 3.7V, two cells were still at 3.4V.

Came back an hour later, the last 2 were at 3.7 V..

Hmmm - cell balancing or a good charger?

DK

 

[safe]

If an a123 cell is allowed to get too much below 2V, it can be recovered, but it will lose 10-15% of its capacity.
...
One characteristic of these sorts of cells is that they tend to have fairly flat discharge curves/ What this means is that the power is very consistant, all the way up to the end. The power feels the same 10 seconds from the end, as it does in the beginning. Then, the power drops quick, like a switch was turned off. If you do this with a pack with unbalanced cells, you could end up with cells that get damaged because they were over-discharged.

Okay, so the group voltage might still be high, but an individual cell reaches it's endpoint of capacity and when it's the first to drop off in voltage this is going to cause it to get damaged. So you would expect one cell to get weaker and weaker while the rest holds strong.

:? Hmmmmm...

Ideally you would not run your batteries into the ground by running them empty, but unlike Lead Acid batteries that sag and sag and sag so that you just don't want to empty them any more the LiFePO4's flat discharge curve gives you no way to know by feel that you are nearly empty.

So is it fair to say that charging does NOT need balancing, but discharging needs to protect against individual cell overdischarge?

 

[Jeremy Harris]

Sounds a reasonable plan, to me. The RC chaps have been using LiPo battery packs with no discharge balancing for years, at very high C rates. They just balance cells when charging.

Having blown up my NiMH pack rather spectacularly (see here: http://endless-sphere.com/forums/viewtopic.php?f=14&t=3241&start=0#p47328 I'm now looking at getting another battery. My preference would be to charge each cell individually, via a multipole connector (I wonder why??? !!!..) and just build an LVC circuit (like the one Bob McRee has posted) into the bike.

I may try and buy some cheap LiFePo4 cells to make a new custom pack up. Anyone know the size of the 10Ah and 15Ah cells that Ping uses please?

Jeremy

 

[GGoodrum]

Just had an interesting experience. My Flintstone charger was about 2 hours into charging, all but 2 cells were already at 3.7V, two cells were still at 3.4V.

Came back an hour later, the last 2 were at 3.7 V..

Hmmm - cell balancing or a good charger?

DK

Actually, "good charger". Your so-called "Flintstone" unit is made up of a bunch of individual cell chargers, connected in series. This allows each cell to absolutely reach a 100% charge in whatever time it needs, everytime. This is what you want.

The fact that it took two cells so much longer shows that they are getting out-of-balance with the rest of the cells, so further proof that it not only can happen, but that in this case, it is happening everytime the pack is used.

-- Gary

 

[safe]

Sounds a reasonable plan, to me. The RC chaps have been using LiPo battery packs with no discharge balancing for years, at very high C rates. They just balance cells when charging.

Wait a second you just said the OPPOSITE of what I said!

I was saying that discharge is where you could potentially do damage because if the voltage drops too low in an individual cell (not the group as a whole) it's that specific cell that gets all the damage. When charged that damaged cell takes as much charge as it can and then just sort of goes inert while the "healthy" cells continue charging. On the next discharge cycle that same damaged cell is the first to run empty and the pattern repeats until one cell is basically completely gone.

I'm not sure how at this point the rest of the cells get damaged...

It seems like you could almost have a "sacrificial cell" that is always weak and it serves as the trigger for the rest, however, this means that now the rest of the pack is barely being used and the weak one will keep dragging the whole system down. Not a great strategy for getting your full capacity, but it does seem to limit your losses once this pattern sets itself up.

 

[Jeremy Harris]

BUT, I also said

and just build an LVC circuit (like the one Bob McRee has posted) into the bike.

Looks like I didn't REALLY say the opposite after all............................

The Bob McRee circuit seems a sensible (and simple) way to prevent cell damage on discharge to me, as it checks every cell and turn the controller off if any single cell drops below 2.7V.

Charging gives me the willies at the moment, but then I'm still a bit sensitive about that topic.

Jeremy

 

[GGoodrum]

Sounds a reasonable plan, to me. The RC chaps have been using LiPo battery packs with no discharge balancing for years, at very high C rates. They just balance cells when charging.

Having blown up my NiMH pack rather spectacularly (see here: http://endless-sphere.com/forums/viewtopic.php?f=14&t=3241&start=0#p47328 I'm now looking at getting another battery. My preference would be to charge each cell individually, via a multipole connector (I wonder why??? !!!..) and just build an LVC circuit (like the one Bob McRee has posted) into the bike.

I may try and buy some cheap LiFePo4 cells to make a new custom pack up. Anyone know the size of the 10Ah and 15Ah cells that Ping uses please?

Jeremy

The cheap Chinese BMS designs typically do discharge-type balancing, but not not necessarily while the pack is being discharged. Basically, what happens is that these RC-styled balancer circuits will drain the level of the higher voltage cells, down to the level of the lowest one, through a circuit that lets it pass up to about 100-150mA. Not very much, really, but any more than that and they'd have to use a heat sink. With a big Ah pack, this can take 10 hours, or more, if you have cells that are not the same. RC-style balancers work the same way, but RC LiPo packs are typically only 2-5Ah, in capacities. Even still, most of the newer balancers will dissipate 300-500mA now.

RC packs are also made up of very carefully matched cells, in capacity and internal resistance, so balancing does allow the cells to get very close in voltage, while still getting close to a 100% charge. You can do this sort of cell matching, as Doc suggests, if you are using lots of smaller Ah cells, like the 2.3Ah a123 cells, but that's not going to happen when each cell is 20Ah. You will be lucky to get cells that were out of the same production lot. From what I've seen using 30C/50C a123 cells is that eventually the cells end up where they just don't stay that balanced. I suspect the same sort of thing will happen with these cheap 2-5C LiFePO4 cells, only sooner than later.

-- Gary

 

[Jeremy Harris]

Thanks for that, Gary. I hadn't realised that the RC pack charge balancers were really discharge balancers.

Overall, what are your thoughts on charging each cell individually from a properly controlled charger? My guess is that this is both safe and will tend to keep all the cells balanced.

Coupled with simple cell over-discharge protection, via an LVC to the controller cut-off, would seem to be about as safe a system as is reasonable.

Jeremy

 

[GGoodrum]

BUT, I also said

and just build an LVC circuit (like the one Bob McRee has posted) into the bike.

Looks like I didn't REALLY say the opposite after all............................

The Bob McRee circuit seems a sensible (and simple) way to prevent cell damage on discharge to me, as it checks every cell and turn the controller off if any single cell drops below 2.7V.

Charging gives me the willies at the moment, but then I'm still a bit sensitive about that topic.

Jeremy

I just did a new 16-channel version of the "Bob Mcree" LVC board, mainly for use with the LiFeBatt packs we are doing, but it will also work for any LiFePO4-based pack, with up to 16-cells. What is different in the new version is that we added the option of being able to actively cut power, if one of the circuits trip. This is done by using two high-power FETs in series with the negative power lead. The reason for this is so that these can be used in setups that don't have the Crystalyte "brake inhibit" line that can be simply grounded to have the controller cut power. The board layout provides for both cases, with the ones that will be used for C-lyte setups just not being populated with the FET parts. The other change is that we are using a lower cutoff voltage, 2.1V instead of 2.7V, because testing has shown that works best with all but the extreme high-C a123 cells. Here's what the new board looks like:

With the LiFeBatt packs, I'm using a set of 18-pin AMP 4.2mm PE Series connectors to plug the this LVC board and the pack together. One connector mounts on this board, and a matching plug is wired to each of the cell junctions. From this board, I wire another 18-pin plug that goes outside the pack, that serves as a connection point for the new "Charger Management System", or CMS, that Bob and I are developing in order to do higher power individual cell charging, at rates up to 8A per channel. This requires that you use 16-18-gauge wire for the plugs and wires going to the indiidual cell junctions. If you are just using the LVC by itself, though, and not to also pass through charge currents, you could use much thinner wires. The LVC circuits draw next to nothing.

Anyway, I just got the boards for these back yesterday, so I can and will start offering them for use on other packs. PM or email (ggoodrum@tppacks.com) if yo are interested.

-- Gary

 

[safe]

Solid State Relays?

Humor my ignorance, but I always like to do the "Columbo Routine" with these topics and try to steer things through a narrow and simplistic path so as to understand why something more complex is really necessary.

Couldn't you set up solid state relays so that any time a cell dropped below it's lower voltage cutoff it would simply remove itself from the battery pack?

The nice thing about this is that you could go without the controller cutoff completely because the battery would turn itself off by removing one cell at a time. Towards the end of the ride you might go from 16 cells, to 15 cells, 14, 13, 12, 11, 10, 9 and by this point you are going to realize the bike is pretty much done. This would guarantee that 100% of the cells are being used and you can ignore their degree of balance because when using such a system you basically don't care if they are balanced or not... they just deliver as much as they can then stop when they have to. With these bigger sized cells the smaller number of them required to make a battery pack means that an array of solid state relays seems more practical.

What's wrong with this?

 

[GGoodrum]

Thanks for that, Gary. I hadn't realised that the RC pack charge balancers were really discharge balancers.

Overall, what are your thoughts on charging each cell individually from a properly controlled charger? My guess is that this is both safe and will tend to keep all the cells balanced.

Coupled with simple cell over-discharge protection, via an LVC to the controller cut-off, would seem to be about as safe a system as is reasonable.

Jeremy

The best way to ensure that each cell gets to its own 100% level, in whatever time it needs, is to use individual 3.65V CC/CV chargers on each cell. I have a bunch of the Voltphreak CC/CV chargers, and have used them on my a123-based packs. They work great, but only output 2A each, so it takes a long time. If I’m in a hurry, I just use a bulk charger, but in any case, about every 2nd or 3rd charge, I use these to make sure the cells get a 100% charge at least every few cycles.

Individual cell chargers that can handle 8-10A would be too complex and way too costly, although a few of the more ambitious DIY’ers, like Doc, are building such setups. What most “big boy” EV BMS designs do is use shunt-type voltage regulators on each cell, or more accurately, each block of paralleled cells, in order to control the CV portion of the charge process. Just like a big zener diode, functionally. Bulk chargers are still used, but what the shunts do is bypass whatever current the recharging cell doesn’t need, so the next cell (or block of cells…) can have as much of the max current it needs. If the cell is nearly full, most of the current goes through the shunt, so it needs to be able to dissipate the heat generated by whatever current the shunt bypasses. In a larger EV application, there is enough real estate to handle this, but not so in an ebike setup. That’s why we decided to do this as a separate CMS unit. Functionally, what we have is exactly the same as on the larger EV setups, but we just move the shunts out of the pack, and into a separate unit (the CMS…). It means we have to bring out the cell junction wires, but this is much easier to do than having to deal with getting rid of a ton of heat in the packs.

Of the three major functions a "real" EV BMS handles, low voltage protection, over-current protection and shunt-type voltage regulation during charging, only the LVC really needs to be done at the cell level during use, so it is the only function that needs to be pack-resident. All the current used has to go through all the cells so having the controller do the current limit function is fine. You just need to add a big fuse as a "last resort" should the controller fail. the CMS we are trying to get finished up this week (Bob has been testing the prototypes...), has a big heatsink on it so that we can safely dissipate the heat generated when most of the cells are in, or close to full bypass mode.

-- Gary

 

[Doctorbass]

OK.. I already did some test for you last days:

I used my LBC-10 balancer (but i used it ONLY to monitor each cells curve voltage ... and i BYPASSED THE BALANCING CIRCUIT ON THE BALANCER.. so it is only used to monitor and to send the data via my charger and my pc.

Here is some test with deffective cells from the dewalt center... the A123 deffective pack are a great exemple of pack with weak cell noh? 8)

First, i left an empty pack on a balancer with the balancing mode activated to fully discharge each cells to the same SOC (2.5V limit)

-here is the circuit i used (found on RCgroup)
http://www.rcgroups.com/forums/showthread.php?p=7632035
this circuit allow to monitor a multi cell discharge or charge curve without being affected by the balancing circuit

the linear green curve is the mAh and the rest are the A123 cells (in the pictures they are lipo Z.......)

The cell 9(in light purple) is weak and also, the best cell in this pack is the cell 3 (in green)

I hope that can give you an idea... it is rare that graph of multi cells are shown simultanously during charge or discharge...

 

[GGoodrum]

Solid State Relays?

Humor my ignorance, but I always like to do the "Columbo Routine" with these topics and try to steer things through a narrow and simplistic path so as to understand why something more complex is really necessary.

Couldn't you set up solid state relays so that any time a cell dropped below it's lower voltage cutoff it would simply remove itself from the battery pack?

The nice thing about this is that you could go without the controller cutoff completely because the battery would turn itself off by removing one cell at a time. Towards the end of the ride you might go from 16 cells, to 15 cells, 14, 13, 12, 11, 10, 9 and by this point you are going to realize the bike is pretty much done. This would guarantee that 100% of the cells are being used and you can ignore their degree of balance because when using such a system you basically don't care if they are balanced or not... they just deliver as much as they can then stop when they have to. With these bigger sized cells the smaller number of them required to make a battery pack means that an array of solid state relays seems more practical.

What's wrong with this?

And how is this simpler? :? SS relays are bulky and expensive, plus you still need the detection logic to drive them. The beauty and elegance of Bob's design is that it uses just three parts per channel, a voltage detector chip, a current limiting resistor and an optocoupler so that all the outputs can be ganged together. These circuits can be left connected permanently to the pack because the current draw is fraction of what the cells will self-discharge, over time.

If any LVC circuit trips, the load is removed. Can't get much simpler than that. If you have a case where one cell is causeing the LVC to trip too early, that is an indication that this cell has serious problems, and the pack should probably not be used until this is resolved.

-- Gary

 

[safe]

The best way to ensure that each cell gets to its own 100% level, in whatever time it needs, is to use individual 3.65V CC/CV chargers on each cell.

I feel like we're going backwards here...

Haven't we agreed that charging behaves like Lead Acid in that when a cell is full it simply becomes inert and allows other cells to absorb more charge?

There is no need for charge balancing... the only critical thing to worry about is when a cell drops below it's low voltage level while the rest of the pack is still going strong. It's this first cell that drops off that will tend to age faster. So discharge balancing matters while charge balancing does not.

You guys need to learn how to "teach" better.

Slow down the ideas... try to think as if you are a beginner and you are making the path from no knowledge to the final conclusion. Thrown in some fun along the way to make it easier for a novice to enjoy the journey.