New 16-cell Battery Management System (BMS)

[Deepkimchi]

Maxwell - what type of charger you had that fried ?

Fixed mine for about $30 bucks of parts.

Bryan

 

[fechter]

I still can't understand why you put so much effort into something to sell and then post the schematic. Why not just PM it to the few individuals that might have a valid interest (and actually understand).

If someone buys the board, like me then you could include the schematic with a description of how it works, critical voltage points, as an "owners manual".

I really think you should protect your investment and hard work.

Interesting point.
If someone really wanted to copy the design, they could just buy one and reverse engineer it.

What would you rather buy?; a board that has full schematics and a description of the operating principles, or one that has all the numbers sanded off so you can't figure out anything. This is the beauty of open sourcing.

 

[PJD]

Don't know how old anyone is here, but it wasn't that long ago that all consumer electronic goods - from 'transistor radios" to TV's always came with a schematic, and people would have thought "intellectual property" was a phrase from a George Orwell novel.

What happened?

 

[GGoodrum]

What happened?

We started spawning too many lawyers...

 

[docnjoj]

Agreed, with emphasis on the "greed"!!!
otherDoc

 

[maxwell652000]

hi bryan,
the charger is a no name from china , i opened it up 2 caps an mov and a temp sensor have to be replaced,eventually i will fix it,,,as for my cell that registered 2.83v when all the rest were at 3.25v,,i have a vector smart charger @12v with amps adjustable at 2/6/12,,, i isolated 4 cells of my pack and hooked up the charger and i am happy to report that the unbalanced cell at 2.83v came back up to 3.40 and once i unplugged the charger all cells settled at 3.3v the cell that was at 2.83v settled at 3.29v and i hope through more charge discharge cycles it will even out at 3.3v.as for garys concern that the cell came so out of balance in the first place i will only know on my next discharge cycle as to whether the cell is damaged,,,im hoping that the cell came so out of balance because of the balance board function on my defective charger went haywire.

so now i have another question...the vector smart charger im using to charge 4 cells at a time seems to bring the cells to a maximum 3.4v each however if i pump more amps and i go to the maximum 12amps the cell that was originally 2.83v within two minutes jumps to 3.67v while the other 3 cells are holding at 3.55v and 3.50,,i should point out that the cell that was originally 2.83v is at the end of the string using the negative charger pole ,,,the cell at the other end registers 3.55v on the positive charge pole is this an indication that the cell is bad?,,,or is it because its at the end of the string on the negative charger pole that it is registering at 3.67v while the other cells are lower..after the charger is diconnected they settle down to 3.34v while the cell that was originally 2.83 and goes up to 3.67 has now settled to 3.32v also what is the ideal cell voltage of lifepo4's when it is completely charged,,,3.2v/ 3.3v /3.4v?

 

[maxwell652000]

as it is the first time im charging my cells with a sla smart charger i did not complete the charge cycle ,,,when i saw that the cells were at 3.5v i stopped charging,,, but then i decided to see if i let the charger continue would it lower the amps it normally puts out as the cells become charged until the point that it stops on its own and indicates a full charge,,,and it did! all the cells now register 3.3v exactly.

 

[fechter]

also what is the ideal cell voltage of lifepo4's when it is completely charged,,,3.2v/ 3.3v /3.4v?

Best I can figure is 3.65v per cell. It seems they can tolerate going over 3.7v, but I wouldn't push beyond that. (These are the voltages during charge). I have no clue what the resting voltage is, and I believe it's not a useful measure. I think you need to put at least a light load on the cell, then measure to drain off the 'surface charge'.

If you can pump the right number of amp-hours into the cells without any cell going over 3.7v, you should be good.

The design of the BMS circuit has evolved further, and I am starting to feel pretty comfortable with it. More testing is needed. The redesigned circuit should be significantly smaller and have a lower parts count. This is good if you're the one who has to solder in all the parts. I still think it should be able to handle around 20 amps charging with no additonal heat sink. With a heatsink and more FETs in parallel, there should be no practical limit on charging current. The balancing current is still targeted at 150ma due to the total heat dissipation of 16 cells. With upgraded parts and some kind of cooling, the balancing current could be greatly increased, but I don't think this would be necessary unless you were doing 100 amp-hour packs or something.

 

[Deepkimchi]

Funny, one of my cells on the end near the terminals gets higher than the rest.

They mostly top out at 3.7,and settled down to 3.34, 3.33V

DK

 

[DavCbr]

I'm new here, and have no contributuion to make as far as designing this stuff. However, I can solder and am interested in what you are creating here. I do have a couple of questions...
Can this be adapted such that 14 or 15 cells are used? The reason I ask leads to my other question. I have a Crystalyte setup with a 408 motor and their standard 36V - 20A controller. I have been told that I can supply this with 48V, whioch I would like to do. But I would also like to make the switch to to lithium. 16 cells fully charged start off at nearly 60V, soon dropping to about 53V. Dropping one or two cells would put me closer to 48V.
So my other question is, does anybody here have any idea if my contoller COULD handle the 16 cells, which I would prefer anyway. I ride this to work each day, and I'm often running late. Any extra speed would be nice.
dc

 

[maxwell652000]

yes sir, all my cells at resting current are at exactly 3.3v after being off the charger for more than 12 hours,,i think the fact that im using a 12v smart charger for only 4 cells at a time is what is allowing for the good balance,,,so now i did a little experiment and hooked all the cells in series and hooked up my 12volt charger to 4cells and of course the charger only sees those four even though the series connection is not broken, so now im going to mount the cells back on my bike with 4 sets of wires coming from each set of 4cells allowing me to charge one set of 12v at a time ,,,next step is to buy garys bms and charger when its ready,,hallayuyaaa,,,next project= electric car running on 288v/100ah lifepo4's

 

[Ypedal]

I'm new here, ... does anybody here have any idea if my contoller COULD handle the 16 cells, which I would prefer anyway. I ride this to work each day, and I'm often running late. Any extra speed would be nice.
dc

Hello Dave.. yes.. your crystalyte controller will take a 48v pack .. charged up to 56v no problem .. ( well. sometimes they blow up.. but i've done this many..many times over with 16 cells without any issues.. ) :wink:

 

[nomad85]

I'm new here, ... does anybody here have any idea if my contoller COULD handle the 16 cells, which I would prefer anyway. I ride this to work each day, and I'm often running late. Any extra speed would be nice.
dc

Hello Dave.. yes.. your crystalyte controller will take a 48v pack .. charged up to 56v no problem .. ( well. sometimes they blow up.. but i've done this many..many times over with 16 cells without any issues.. ) :wink:

I'll second that, I just tried my 36v controller on a 48v pack that was charged to 60v, it went fine. and the voltage leveled off around 52-53v.

 

[DavCbr]

Ypedal & nomad85

Thanx

Now I just wait for these lead batteries to wear out. There is a future!
dc

 

[Anonymous]

.

 

[GGoodrum]

Progress...

I finally got a stable working version. Here's a shot of my test article, which has half the board populated:

This version has a lot fewrer parts, and some more robust logic to turn the FET on and off, during the pulse mode. We now have a separate orange LED for each channel that comes on as the cell voltage reaches the cutoff (between 3.65-3.70V...). Just befoe the shunt gets swamped, an opto is turned on which causes a gate driver chip to cut off the FET, which removes charger power. this starts an oscillation as cell voltage drops enough to turn off the opto, which turns the FET back on. When a channel is active in this mode, its orange LED dims a bit. There is a "master" red/green LED that comes on red, when charger power is first applied, and the charger is in the normal CC mode. Once one of the optos starts tripping, the master LED turns a yellowish-orange. Basically, all you do is wait for all the LEDs to come on, and then the cells are about as full as they are going to get. When each one comes on, in relation to the others, will give a good indication of the relative balance of the cells.

It has taken a bit of time, but part of the problem is that I had to do most of the testing here, with remote help from Richard. He's been on vacation with his family this week. Now, I'm going on vactaion with my family today, so the tables will turn, and Richard will continue on next week. We are still playing with a couple resistor values, and I think we are going to change one of the transistors to a higher current version, but basically, mine is working. As you can see in the picture, I've tack on a couple of parts, and we eliminated quite a few, so there are some jumpers. Bottomline is that I need to do one more version of the PCB layout. This will let me scrunch things up a bit, which will make the board a bit narrower.

Anyway, I think Rchard can finish up the testing next week, and then I'll submit the new boards to be done, I suspect I will be able to get them back by around Monday or Tuesday of the following week. That's when the kits should be good to go.

-- Gary

 

[Randomly]

Looking good. If you need a higher current relatively cheap TO-92 PNP the KSB1116G or KSB1116L parts are good. They maintain high beta even past 1 amp.
I would like to know what connectors exactly that you are using? thanks.

 

[GGoodrum]

Looking good. If you need a higher current relatively cheap TO-92 PNP the KSB1116G or KSB1116L parts are good. They maintain high beta even past 1 amp.
I would like to know what connectors exactly that you are using? thanks.

I think those are the ones that we are switching to, but I'll check. I think the shunt one is a tall TO-92?

BTW, we are also taking your advice, and are putting separate resistors on each leg of the bicolor LED. The reverse voltage is only 6V.

The connector I'm using is an AMP 4.2mm PE series. The pins are each good for up to 9A.

-- Gary

 

[Deepkimchi]

Hello Gary,

If you are going to change PCBs, could you space the battery terminals a little further apart?

Safer to work on..

DK

 

[Randomly]

I think those are the ones that we are switching to, but I'll check. I think the shunt one is a tall TO-92?
-- Gary

The KSB1116 is in a regular TO-92. I'm sure whatever you're getting will work fine, it's a very noncritical part. As long as it can handle the current, the LM431 can supply plenty of base drive. Using a TO-92L package with it's larger heat dissipation capacity can't hurt either since the transistor is going to be operating in the linear region and dissipating some heat. At the worst case bias point the transistor will need to dissipate 1/4 of what the load resistor will dissipate when it's put across the full cell voltage. Don't pack it in too tight, and any extra PCB copper area you can afford on the Collector pad will help some also. You can always scale back the balancing current if you end up with heat problems.

 

[GGoodrum]

The KSB1116 is in a regular TO-92. I'm sure whatever you're getting will work fine, it's a very noncritical part. As long as it can handle the current, the LM431 can supply plenty of base drive. Using a TO-92L package with it's larger heat dissipation capacity can't hurt either since the transistor is going to be operating in the linear region and dissipating some heat. At the worst case bias point the transistor will need to dissipate 1/4 of what the load resistor will dissipate when it's put across the full cell voltage. Don't pack it in too tight, and any extra PCB copper area you can afford on the Collector pad will help some also. You can always scale back the balancing current if you end up with heat problems.

So far heat hasn 't been a problem at all, but we are using 20 ohm 2W shunt resistors, for now, That's about all the 2N3906 parts I had already can handle.

I'll defintely leave soom room for the transistor, thanks

-- Gary

 

[MagicBob]

Sorry if I've missed this, but how do I order one?

 

[J Barrie Wilkinson]

I am interested in buying this BMS but would like to know if it similar to the original Goodrum/McCree design, which required electronic equipment to set it up after assembly. My knowledge of electronics is close to zero, I have a large and a small soldering iron and a DMM. I am willing to assemble a kit but do not have equipment to " tune " it.

 

[GGoodrum]

Sorry if I've missed this, but how do I order one?

Once the board is ready, I will make kits available here:http://www.tppacks.com/products.asp?cat=26. It'll be another week, or so, though.

This new version doesn't require any calibration or adjustments.

-- Gary

 

[jeffkay]

Gary, Will you be explaining the values needed for different "top charge" voltage detection? In other words, maybe 4.2v rather than the 3.75ish A123?
Jeff K.
P.S. Glad to hear this is almost out of the oven...