Even Newer 4 to 24-cell Battery Management System (BMS)

[GGoodrum]

When you quote, please try abbreviating like this:

Jack said "... blah blah ..."
(not the "..." in place of 5 paragraphs of information)

Otherwise your post is 3 pages long and becomes hard to read.

I couldn't agree more. If I could edit the post myself, I'd take out all the repetitive crap.

*** LIPO ***

BTW: I am about to order the parts for my 24 cell Lipo version.

TC-54 is now available in the TO-92-3 packaging @ 3.0V in quantities of (1) (in stock at Digikey) for those doing LiPo. You will have to use the Open Collector version but it should behave just the same. (right?)

Yes, either version works fine.

Only two changes that need to be made to do Lipo on the current boards:

For HVC: Swap the 75k resistor at R101 with a 110k resistor to get a 4.2V cut off. (Could be done other ways)
(swap part: 270-75K-RC|24 for part: 270-110K-RC|24) - Mouser BOM

For LVC: Swap the 2.1V part with a 3.0V part (TC54VN3002EZB-ND) - Digikey

I have not actually tested this yet so not fact until tested

-methods

I haven't run the math, but that sounds about right for the resistor change. Ideally, you would want the number to be just under 4.2V, like around 4.18-4.19V. That is so that the shunts will all kick in without having to "tweak" the output voltage of the charger. It is usually not possible to adjust an RC LiPo charger up over exactly 4.2V per cell.

 

[GGoodrum]

space is not so much of the issue. mostly it is that it would be easier to have a separate lvc so the diode method for paralleling on the cell level would work better because if 1 cell goes bad it would not take down the other cells and would not affect the lvc or charger.

if i put the diodes to prevent the dying cells from draining good cells then it means that the battery will not charge but the vlc would work.

Why on earth would you want to put a diode on each cell? Look, these things don't just wear out, and die off.The only way to kill one is to over discharge them. You can protect a parallel block of cells with a single circuit. I routinely marry up cells that Ive damaged by s"slightly" over-discharging them, with healthy cells in parallel. The strong ones will "help" the weaker cell. As long as you always use LVC protection, from that point forward, the weaker cell will be fine. I've got over 1000 a123 cells, in various packs, and I've never seen a single cell fail, on its own, and pull down other cells paralleled with it. That just doesn't happen, as long as you have LVC protection.

 

[ejonesss]

i have seen other people use the diodes to prevent the packs/cells from being drained by a weak pack/cell.

looking at http://www.neodymics.com/Images/V24ProtoSwitch070818E.pdf shows diodes used to keep the power from going back into the battery by another battery.

so i can be able to mix various sizes of cells like 18650, a123 and 40138(bmi lifebatt) and there will not be any problems?


also in a future bms/lvc is it possible to use bypass shunts on the lvc do if a pack gets out of balance for example.

if a cell goes low then that cell is bypassed as if it was removed and then the load will be carried across the rest of the pack and so on until the controller's lvc kicks in?

it would allow a little more run time in theory.

space is not so much of the issue. mostly it is that it would be easier to have a separate lvc so the diode method for paralleling on the cell level would work better because if 1 cell goes bad it would not take down the other cells and would not affect the lvc or charger.

if i put the diodes to prevent the dying cells from draining good cells then it means that the battery will not charge but the vlc would work.

Why on earth would you want to put a diode on each cell? Look, these things don't just wear out, and die off.The only way to kill one is to over discharge them. You can protect a parallel block of cells with a single circuit. I routinely marry up cells that Ive damaged by s"slightly" over-discharging them, with healthy cells in parallel. The strong ones will "help" the weaker cell. As long as you always use LVC protection, from that point forward, the weaker cell will be fine. I've got over 1000 a123 cells, in various packs, and I've never seen a single cell fail, on its own, and pull down other cells paralleled with it. That just doesn't happen, as long as you have LVC protection.

 

[fechter]

Any ideas on how to scale this design up to more cells? I am building a pack that will be 112S4P of A123 M1 cells. My only real hurdle is the BMS side. Everyone seems to want to move to 10 to 100 Ah cells but they don't fit my form factor for space reasons.

112s? that's like 400v! Yikes! One touch and you're toast.

The cell circuits are optocoupled, so there is no limit on those. The existing LVC circuit should work also. The only issue will be the charge control circuit. With a higher (much higher) voltage rating on the FET and voltage regulator, it should work. It would be better to interface the optos directly to the charger to avoid the need for a FET.


What would you use to charge something like that? How much current?

 

[ejonesss]

sounds like you are powering a car or an extreme bike or motorcycle

other facts

1. 400 volts actually that is 369.6 volts 3.3 times 112 to 403.2 volts 3.6 times 112 at 12.8 ah.

2. being dc you have a chance of survival because i heard somewhere that dc you can pull away but ac you will tend to hold on to the power and can be deadly.

also you may want to make sure to have the the connections well covered because a short at those voltages can be bad.

Any ideas on how to scale this design up to more cells? I am building a pack that will be 112S4P of A123 M1 cells. My only real hurdle is the BMS side. Everyone seems to want to move to 10 to 100 Ah cells but they don't fit my form factor for space reasons.

112s? that's like 400v! Yikes! One touch and you're toast.

The cell circuits are optocoupled, so there is no limit on those. The existing LVC circuit should work also. The only issue will be the charge control circuit. With a higher (much higher) voltage rating on the FET and voltage regulator, it should work. It would be better to interface the optos directly to the charger to avoid the need for a FET.


What would you use to charge something like that? How much current?

 

[AndyH]

Fechter,

I've run thru the troubleshooting steps posted so far. I have a successful V1.5 build before this.

12 channel board connected to 8 fully charged cells on channels 1 thru 8. Cells 9 thru 12 are about 90% charged.

When connected to the charger, channels 1 thru 8 go 'orange' quickly. Almost immediately, the resisters for 1-8 get very hot, as do Q101 thru Q801. Cell voltages run up thru 3.9 to 4.1V after 10 minutes. During the same 5 to 10 minutes, channels 9 thru 12 start a faint orange glow (faint as in 'if you look very closely you can see an orange glowing square'). Cell voltages here are 3.58 thru 3.61.

[edit] 'very hot' above is 260 to 280F per the IR thermometer [/edit]
[edit] The V1.5 board LEDs respond the same as the 2.2 board, so no trouble there. It doesn't appear that power is being shunted away from the charged cells even after the LEDs are blindingly bright.[/edit]

I made a bad pin in the molex connector between the board and cells - between channels 8 and 9. It's possible on the first powerup that I fed too much power into the first 8 channels.

Thoughts on what I might have fried? :?

On the orange LEDs... There aren't any flats on the plastic that I can see, even when aided by a 10x lens. Short lead goes into the square pad.

[final edit] Swapping the shunt LEDs took care of the board. The board isn't latching into 'end of charge' because my smart charger drops out of CC before the last couple of cells are fully charged, then it just sits there and makes heat. The board wasn't designed to 'play with' smart chargers, so no biggie. Nothing a Mastech won't fix. THANK YOU Gary and Richard for your help and to the group for making this design happen! [/edit]

 

[webfootguy]

112s? that's like 400v! Yikes! One touch and you're toast.

I'm ok with high voltages. I've got lineman gloves good to 600V and a healthy respect for high voltage. I'm replacing 280 c sized nicads (2 to 3 packs) each with their own BMS. RS485 based communication and I almost have a software emulation of the BMS fully functional (enough to fool the car into a ready state).

The cell circuits are optocoupled, so there is no limit on those. The existing LVC circuit should work also. The only issue will be the charge control circuit. With a higher (much higher) voltage rating on the FET and voltage regulator, it should work. It would be better to interface the optos directly to the charger to avoid the need for a FET.

That might be impossible as I can not get any technical info on the charger. I will try to pursue this as it might be possible.

What would you use to charge something like that? How much current?

The controller and charger are integrated together. The charge current is settable, runs off 120v, and unfortunately is currently expecting to do a charge profile for nicad cells. My choices are to emulate a behavior that gets it to do what I want, or use a secondary charger. Not sure yet what I will do.

 

[methods]

Oh boy, 140 solder joints and one step closer to BMS nirvana.

My harness for the 24S 2P 10Ah Lipo pack
Complete with sacrificial gender changing end caps
Built of JST-XH 7-pin (6S) connectors
Plug and play with most all Hobby City Lipo packs (Zippy etc.)

-methods

 

[ejonesss]

where can i get these plugs and gender changers?

i need to get 9 pin sets but will settle for 8 pin since i can use the junction of the + of 1 pack to the - of the other pack with a single plug.

Oh boy, 140 solder joints and one step closer to BMS nirvana.

My harness for the 24S 2P 10Ah Lipo pack
Complete with sacrificial gender changing end caps
Built of JST-XH 7-pin (6S) connectors
Plug and play with most all Hobby City Lipo packs (Zippy etc.)

-methods

 

[Tiberius]

Oh boy, 140 solder joints and one step closer to BMS nirvana.

Built of JST-XH 7-pin (6S) connectors
Plug and play with most all Hobby City Lipo packs (Zippy etc.)

Hi Methods,

Is there a standard connector and pinout for these battery to balancer connections?

Nick

 

[GGoodrum]

Oh boy, 140 solder joints and one step closer to BMS nirvana.

Built of JST-XH 7-pin (6S) connectors
Plug and play with most all Hobby City Lipo packs (Zippy etc.)

Hi Methods,

Is there a standard connector and pinout for these battery to balancer connections?

Nick

I think what he has don is make four "Y" connectors that each allow two 6s packs to be connected together in parallel, at the cell level. These connectors have 7 pins. They usually have one red wire and 6 black wires, or they are multi-colored, with red at one end, and black at the other. The red wire goes to the pack +, the black wire next to it goes to the junction between the 5th and 6th cell, and so on. The last black wire goes to pack -.

Anyway, the connectors are pretty common, but the female versions are harder to find.

-- Gary

 

[fechter]

When connected to the charger, channels 1 thru 8 go 'orange' quickly. Almost immediately, the resisters for 1-8 get very hot, as do Q101 thru Q801. Cell voltages run up thru 3.9 to 4.1V after 10 minutes. During the same 5 to 10 minutes, channels 9 thru 12 start a faint orange glow (faint as in 'if you look very closely you can see an orange glowing square'). Cell voltages here are 3.58 thru 3.61.

[edit] 'very hot' above is 260 to 280F per the IR thermometer [/edit]
[edit] The V1.5 board LEDs respond the same as the 2.2 board, so no trouble there. It doesn't appear that power is being shunted away from the charged cells even after the LEDs are blindingly bright.[/edit]

Thoughts on what I might have fried? :?

Cell voltages should never get that high. Excessive voltage will increase the heating.
It sounds like the control circuit is not throttling when it should. You can power it up and try shorting the output of one optocoupler to see if the main LED goes solid green (see pic below). Every cell circuit should work the same way.

This should stop the charge current and make all the cell LEDs go out. If the main LED does not go green, then there is a problem with the diodes or the gate driver. If the LED goes green, but the cell LEDs stay lit (it might take them a while to go out), then the FET could be shorted.

 

[Malcolm]

Fechter, or anyone else who knows: Since I have to replace the FET on my board anyway, is there a higher rated FET that I could use in place of the listed FET to reduce the risk of blowing it again with my heavy-duty charger? I'm taking the other steps you mentioned (including using longer, smaller gauge charging cables) but would like the BMS to be as foolproof as possible (i.e. proof against me :lol: )

 

[fechter]

Since you're running a relatively low voltage, there may be some other options. It was the peak current rating of the body diode that got you. I'd have to check some datasheets.

I think a IRFB4110 won't be much better and you don't really need 100v rating.

Two in parallel might be a good option. You can stack them on top of each other and just parallel the leads. Not optimal, but I bet it would solve your problem.
To be a bit more optimized, the gate leads would need separate resistors.

 

[methods]

Hi Methods,
Is there a standard connector and pinout for these battery to balancer connections?
Nick

Nope.
Only standards for different RC batteries. You can follow the Thunder Power connector scheme or the .100
I own more TP than Zippy but I prefer the .100 spacing because I have lots of other gadgets that work on that spacing
There are 4 or 5 different .100 spacing schemes that will fit together with a little razor blade work. Just different clips and what not.

Most importantly you can plug right into Radio Shack proto boards for soldering up custom circuits.

-methods

 

[AndyH]

Cell voltages should never get that high. Excessive voltage will increase the heating.
It sounds like the control circuit is not throttling when it should. You can power it up and try shorting the output of one optocoupler to see if the main LED goes solid green (see pic below). Every cell circuit should work the same way.

This should stop the charge current and make all the cell LEDs go out. If the main LED does not go green, then there is a problem with the diodes or the gate driver. If the LED goes green, but the cell LEDs stay lit (it might take them a while to go out), then the FET could be shorted.

Excellent - Thank you! It's all good - the cells don't get close to 3.7 even when the board is sending them 3.84. It looks like I stressed one of the shunt LEDs when I pulled/replaced it. It tested good but failed soon into a charge cycle. I kept waiting for that last LED to light... The basically dead channel didn't absorb any power and that didn't help heating any.

On the testing side - the picture is excellent - thank you! I used an A123 cell, a 100 ohm resister, and a pair of test leads to energize the other side of the optocoupler (pins 1 and 2) as well - it let me know that I didn't cook the chip.

(sigh...no more soldering after 1AM.)

(edited to repair quote)

 

[methods]

Has anybody souped a board up yet with lower resistance values?
Hook up a little fan, keep an eye on it sort of thing.

My packs have relatively low capacity (10Ah), are of reasonably good quality (RC) and I now charge very aggressively (almost 2C) so any time those things are actually conducting will be minimal.

*(thinking as I write)*

It just occurred to me that I will be running those resistors at 4.2V instead of 3.7V

3.7/6.8 = ~544ma
4.2/6.8 = ~618ma

I guess that by design I will already be running a little on the aggressive side but. . .
I would be overjoyed if I could get that up to 1000ma

I think i will give it a go and report back
We will see what I can fit on the board.
Maybe single resistors, maybe piggyback.

-methods

 

[GGoodrum]

Has anybody souped a board up yet with lower resistance values?
Hook up a little fan, keep an eye on it sort of thing.

My packs have relatively low capacity (10Ah), are of reasonably good quality (RC) and I now charge very aggressively (almost 2C) so any time those things are actually conducting will be minimal.

*(thinking as I write)*

It just occurred to me that I will be running those resistors at 4.2V instead of 3.7V

3.7/6.8 = ~544ma
4.2/6.8 = ~618ma

I guess that by design I will already be running a little on the aggressive side but. . .
I would be overjoyed if I could get that up to 1000ma

I think i will give it a go and report back
We will see what I can fit on the board.
Maybe single resistors, maybe piggyback.

-methods

If you do go down in resistance, upping the shunt current, you should probably add some of those clip-on heat sinks for the shunt transistors. I think they are spec'd to 800mA, or maybe 1A, but I know they already get pretty warm with about 500mA going through them now.

-- Gary

 

[GGoodrum]

Richard did a lot of head work and as a result the nesxt version will support LVC cutoffs of 2.1V, 2.6V, 2.7V and 3.0V. It will also have shunt crossover voltages of either 3.68V of about 4.19V. For this next through-hole version, if you want a higher LVC cutoff, you simply add one or two extra resistors and drill out one or two "jumper" holes, both per channel. Same for the higher shunt cutoff, add a resistor and drill out a hole, for each channel. For the surface mount version, we'll just have the resistors already there, and we'll drill out the jumper holes, for a non-LiFePO4-based setup.

 

[methods]

... you should probably add some of those clip-on heat sinks for the shunt transistors. . .

Good tip, I had not thought it all the way through yet.

I will shoot for 1A and report back.
Worst case scenario I will burn up a resistor and a bjt, no bid deal. :wink:

-methods

 

[fechter]

I think you'll find that after the cells get initially balanced, the balancing time will be quite short even at 500ma. The resistors are only heating during the balancing phase. During bulk charge, the circuit has no limit on the charge current, so you should be able to charge as fast as the batteries and charger can handle.

Going to a higher shunt current will be putting the transistors at the edge of their spec. for heat dissipation. If you really want a higher shunt current, I would recommend using bigger transistors with a TO-126 package that allows bolting on a heat sink.
Here's one: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=BD14016STU-ND
These should fit in the existing holes and have the same pinout. Even with no heat sink, these can dissipate quite a bit more heat than the TO-92L packages.

 

[fechter]

Since this issue seems to be coming up frequenly, here's how to test the auto shutoff feature on the control board:

Power the board and get the cell LEDs lit (or at least some of them).
Use a small jumper wire to breifly bridge the control ground to the 'all shunts' line.
This should trigger the main LED to go green and latch. This terminates the charging current and the cell LEDs will go out after a few seconds. Disconnecting the charger and re-connecting should reset the shutoff and enable charging again.

 

[methods]

...Here's one: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=BD14016STU-ND
...

Thanks Fechter, I will definitely take one of your two advisories: either the bigger transistor or I will leave the circuit alone.

The reason i am shying away from the 500ma balance limit is based only on my experience using similar RC type balancers. It just occurred to me that perhaps I am not comparing apples with apples. . . I now realize that a "500ma RC balancer" does not behave the same as this circuit will.

Let me take some measurements on how effective my current balancer is and then I will reconsider.

-methods

 

[GGoodrum]

With decent quality RC packs I'd be very surprised if you ever had healthy cells that were more than about 2-3% out-of-balance, worst case. With a 10Ah pack, that is like 200-300mAh. With a shunt current of 500mA, it would take about 1/2 hour to let the low cells catch back up to a full 100% level.

-- Gary

 

[velias]

I think my BMS is working fine, but I wondered if anyone has ever had this problem with their packs?
It appears that the very first and very last cell in the pack never get up to the 3.69 volts like the other cells do when charging.
They get up to 3.4 volts and thats it. I'm using a 16 PSI cell pack. I even bought a brand new spare BMI cell to replace one of the low cells and that cell does the same thing, 3.4 volts max.

I remember reading a thread a few months ago about someone who always had lower voltage cells as the first and last cell in a pack as well. I don't remember if they ever figured it out.
It's kind of strange its the first and last cell. I've tried hooking the charger directly to the pack to make sure its not the BMS but the 2 cells still max out at 3.35-3.4 v. I haven't tried charging with the BMS for a long period of time (>15 minutes) to see if the two leds corresponding to those cells will ever come on (they do get slightly dim sometimes). Could it be some issue with the charger? It's one of those Taiwan QQE technology chargers with a no load voltage of about 59.9 volts. I've just started using the pack so maybe things will change after get some use and cycles out of it.