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

[fechter]

Only for my personal understanding: everything is fixed and I can go on and solder my board to the end, even change the R3 resistor?
Please tell my that it's so simple and everything works fine!
(I can't await to get everything together and charge the first time! )

Yep, make R3 4.7k, make R6, R7, R8 10k and C2 0.1uf and you should be good.

Eric! Nice to see you back. Yes a technical reference is on the to-do list, but I keep making changes to the design so it's hard to keep up.

AndyH has been making and selling fully built boards, so you might check with him. He's also been a great help in the testing process.

I'm presently working on an improved charge control circuit that should eliminate some of the issues with the present design and provide faster charging and wider compatibility with charging sources. Things just aren't moving along as fast as I'd like them to due to being busy with other things.

 

[bapou]

Hi fechter and other BMS builders,

some posts ago, I wrote about the problem with the charging finishing to early. You were right that this is caused by the
resistance of the wires and plugs between the BMS and the battery that cause a voltage drop of ~ 100mV @ 20A in the moment. This causes
the orange lights of cell 1 and 4 to light up to early.

I'm now integrating the battery and the BMS in a waterproof case and want to place the circuit close to the battery but have to put the resistors
to the cooling plate at the top of the box to keep the box cool. Now I'm wondering what to do with the MOSFET that is also warming, but should be close to the battery.

Do you think using a MOSFET with ~1.3mOhm would help to keep the MOSFET cooler (other version of the question: is the MOSFET warming
by I*Ri or are these more the switching losses (U*I*tau) that cannot be avoided even with a lower Ri. If it is caused by the Ri, a 1.3mOhm MOSFET should be nearly
3x cooler).

Thank's

Thomas

 

[fechter]

Do you think using a MOSFET with ~1.3mOhm would help to keep the MOSFET cooler (other version of the question: is the MOSFET warming
by I*Ri or are these more the switching losses (U*I*tau) that cannot be avoided even with a lower Ri. If it is caused by the Ri, a 1.3mOhm MOSFET should be nearly
3x cooler).

Thank's

Thomas

Yes, a lower Rds will help a lot since the FET is not switching during bulk charge. Once throttling (switching) starts, the average current will drop fairly quickly which will keep heating to a minimum.

You could also use an insulated heat sink bar and attach to the case or use a board mounted heat sink.

 

[Patriot]

fechter,

Today I am trying to charge up my battery, and see where the higher voltages on channel-7 is coming from. Channel 15 seems to lag behind a bit, but not by much. It also flckered a bit while balancing, just like channel-7. I'm thinking it's not my connector now, and must be something else on the BMS board itself that is malfunctioning. All channels at my connector are reading good voltages.

To start, all cells were within 0.05v of each other.

After three hours of charging, they are all at 3.38-3.40v. So, all cells are within .02v of each other. Even channel-7 that peaked at 4.2v, is charging fairly evenly with the other channels at 3.4v.

I'm thinking it has problems at the end of the charge when the shunt needs to trip? Thus, it may be tripping at too high a voltage? Not sure. I'm no pro at this.

I'm going to check it closely when they start to balance, and see when the problem happens. I'll be careful not to let it go much over 3.7v, as not to damage the cells.

 

[Patriot]

^^^^UPDATE:

As the battery got close to balancing, a couple things started to happen in the last 10-15mins. Most of the channels were close to each other in voltage. However,

Channel-15 started lagging behind the rest (though still coming up slowly).

Channel-7 started taking off ahead of the rest.

As some of the channels started lighting up around 3.7-3.72v on my cheap meter, the channel-7 did not. Channel-7 kept on going. Channel-15 continued to lag behind by about 0.7v from the rest.

Eventually, as I cycled through voltage readings of the channels that were lighting up, I got back to channel-7, and it had shot up to 3.95v in a matter of only a couple minutes, and it still had not started bypassing, and the LED was off when many of the others had already finished.

So, I pulled the plug.

Most of the channels hit about 3.7-3.72v, but channels 7 & 15 were not the case.

Channel 15 had a final reading of 3.65v and not lit.

Channel-7 had a final reading of 3.95v, and also had not activated. This is the problem child. I also include channel-15 as a concern, because it had a bit of flickering when balancing in the past as well.

Whatever the problem is, I've got two channels that aren't balancing as they should. After readin your post form the previous page, I still don't know excatly what to look for on the board between all the components, so it may be getting pulled form the bike in a couple of days. I think I'll let voltages settle out overnight, and disconnect the battery tomorrow.

 

[GGoodrum]

Sounds to me like you have a bad connection between the 7th channel and the 7th cell. Clearly, the shunt for that channel is not operating. You might also check the 7th LM431. If you have a multimeter with a diode check mode, you can check the various combinations of pins 1, 2 and 3. Check a good channel first. I haven't seen too many of these fail, but if they do, it is usually an open circuit.

-- Gary

 

[pm_dawn]

It could pretty well be the lm431 that is gone.
I have had all kind of faults on my 40 cells.
I've had to replace almost all kinds of components, the lm431, the transistor, the opto, the TC54.
Probably due to ESD faults, and in some cases bad connection to pack or wrong mounting order.
So now I always test the BMS for HVC and LVC on the bench before I put it back on the Pack.
I have updated my ESD protection in the lab and it seems that it has gotten better since.

regards
/Per

 

[brainzel]

Yep, make R3 4.7k, make R6, R7, R8 10k and C2 0.1uf and you should be good.

C2 was and is still 0.1uF, so I must change nothing at all (i think). I can't locate R3 ... there are only the resistor values printed on the boards, and "10K" is there several times . Is there a plan where I can get the names of your parts-list? Or an image with a big arrow spotting the right resistor? That would help me a lot

 

[Roy Von Rogers]

I have to admit it would be helpfull to have a better view of the first board, I havent put mine together yet cause I've been waiting for all the updates to get resolved, and I dont need the board yet. But if the first board was better viewed, the rest is just a repetition of the stuff on the right hand side of the first board, so one could reference to such a better view.

When one looks at the pdf file at Gary's site, even using magnification doesnt help, it just too fuzzy.

Roy

 

[AndyH]

C2 was and is still 0.1uF, so I must change nothing at all (i think). I can't locate R3 ... there are only the resistor values printed on the boards, and "10K" is there several times . Is there a plan where I can get the names of your parts-list? Or an image with a big arrow spotting the right resistor? That would help me a lot

There's only one parts change from the labels on the board and that's moving R3 from 10K to 4K7. Install this resistor first then let the board be your guide from then on. :wink:



Enjoy!

 

[fechter]

Thanks Andy!

 

[__Tango]

Quick question about this BMS:

Can the LED's be connected via wires as opposed to being mounted directly to the board? For example, i'd like to have the BMS protected in the interior of my motorcycle, but have the LEDs be mounted on the exterior somewhere? If this is possible, what is the minimum gauge of wire that can be used (e.g, can i use 24 gauge wire from cat-5)? The distance of the wire would be pretty short...probably less than a 12 inches.

Thanks!

...tango...

 

[fechter]

Quick question about this BMS:

Can the LED's be connected via wires as opposed to being mounted directly to the board? For example, i'd like to have the BMS protected in the interior of my motorcycle, but have the LEDs be mounted on the exterior somewhere? If this is possible, what is the minimum gauge of wire that can be used (e.g, can i use 24 gauge wire from cat-5)? The distance of the wire would be pretty short...probably less than a 12 inches.

Thanks!

...tango...

Yes, that could be done and the current in the wires is only around 10ma, so any gauge will be fine. You just need to be very careful none of the connections come loose, since the LEDs are an active part of the circuit. If you were to do that, I would recommend using a piece of perforated circuit board to hold the LEDs and solder wires between boards.

Another option would be to use plastic optical fiber glued to each LED, then run to a remote display panel. I was thinking you could drill a tiny hole in the end of the LED (not easy), and epoxy the end of a fiber into the hole.
It might work to just use heat shrink tubing to couple the fiber end to the LED. I haven't seen good cheap plastic optical fiber around lately (the fat kind).

Somewhere in the future, I plan to add a resistor to each cell circuit to make the LED no longer an active element, since they are easily damaged.

 

[PJD]

Gary or Richard,

Couple questions and comments:

1. Are new v 2.5/2.6 BMS boards available? I notice Gary's site says "out of stock".

2. My old V1.5 boards continue to work well on both the the scooters (16 cell 40AH, 8A charger) except for having to remember to unplug the charger at some point when done. However, I once inadvertently left it plugged in all night and the BMS still survived - and I'm using the 6.8u 5W resistors.

3. I'm thinking of a 4-cell upgrade to a scooter. Could the new board be adapted to the old V1.5 board to obtain the shutoff feature - or just use a 4-cell middle segment to expand the board without the shutoff feature?

Thanks!

 

[fechter]

You'll have to check with Gary about the stock of boards.

Unfortunately, the 1.5 board lacks the diode array to work with the auto shutoff feature, so you'd have to make a complete 2.6 board in order to have that function.

You could add a 4 cell section of 2.6 board and make it work with the 1.5 control circuit, but it would lack the auto shutoff.

It shouldn't be a problem for the cells if you leave the charger on for a long time, but those resistors will get toasty.

I think you have more long term experience than anyone I can think of using this BMS.
How are your cells holding up?
Any idea how many cycles you've put on them?

 

[ejonesss]

i checked tppacks and the boards are out of stock for now.

You'll have to check with Gary about the stock of boards.

Unfortunately, the 1.5 board lacks the diode array to work with the auto shutoff feature, so you'd have to make a complete 2.6 board in order to have that function.

You could add a 4 cell section of 2.6 board and make it work with the 1.5 control circuit, but it would lack the auto shutoff.

It shouldn't be a problem for the cells if you leave the charger on for a long time, but those resistors will get toasty.

I think you have more long term experience than anyone I can think of using this BMS.
How are your cells holding up?
Any idea how many cycles you've put on them?

 

[GGoodrum]

I should have more boards next week. You might also check with AndyH, at Rechargeable Lithium Power. He normally sells assembled/tested versions of this BMS, but he might also just sell the board.

-- Gary

 

[AndyH]

I should have more boards next week. You might also check with AndyH, at Rechargeable Lithium Power. He normally sells assembled/tested versions of this BMS, but he might also just sell the board.

-- Gary

Sorry...I just shipped my last board. I'll reorder once I'm done with my current 16 board build. The last thing I want to see right now is another resistor.

 

[AndyH]

For those of you building your BMSs -- add one part to your Mouser order: a 5-circuit Molex MiniFit Jr connector.

When it arrives, carve a divot in one side to hold a 1/8 watt resistor.

View attachment bend_divot.jpg

The side of the connector is perfect for bending the resistors. Place the resistor in the divot, hold with opposite thumb, the bend the leads down with thumb and index finger.

View attachment 1

The top of the connector is perfect for bending the 1N914 diodes. The openings center the diode. Same drill - pin with thumb while bending both leads at the same time with the other hand.

View attachment bend_3.jpg

It's the best 56 cents you'll spend on the project!

 

[DaveAK]

Just finished reading the entire thread. All my questions that I had have been answered in here already.

Do I get a prize?

 

[ahambone]

I don't have a prize to give, but welcome to the "read the whole thread" club! I started building my boards (2 x 24ch) back when the thread was only 42 pages long. I thought it was exhausting then; for you to make it through 90+ is some real staying power.

Now I just need to finish building my pack so that I can actually use my BMS boards as intended!

Cheers,
--Adam E. Hampton

 

[Patriot]

fechter,

Thanks for all your help.

Now I have another problem I can't figure out.

Today I got my BMS setup with a whole new set of channel leads. I eliminated the 17-pin connector, and just decided to go with eyelet terminals for each channel, and hook directly to the cells.

Well, everything went great, until I hit the 8th channel.

When I connected the 8th channel, the LED lit up, and the big resistor starting getting warm. So, I pulled it, and connected the 9th channel, with the 8th channel disconnected. Then, both 8 & 9 lit up. Weird. Then, I connected the 8th channel, and left it lit for a minute while I also connected the 9th. The 9th did not lit up again, but the 8th channel stayed lit.

I checked the board very carefully. All solder connections for the new leads are good, and have no cross connections or shorts that I can see.

What the heck would cause this?

 

[fechter]

Measure the cell voltage on #8. If the cell had been previously overcharged, the shunt may be bleeding off the excess charge. Let it run for a while and see if it goes out after a minute or so.

 

[Patriot]

^^^ I checked the voltages, cells 7,8 & 9 all are at 3.35v by my meter.

I kept thinking it may be another flux short from installing the new leads, like I had when first building. but I cleaned with a Q-tip/acetone between the solder points, and there's nothing. Plus, to get the resistor warm, I wouldn't think enough current could flow through flux to do that. It's almost like I've got a bad diode or something, causing the current to flow through.

If I hook it up for a few minutes like you say, it won't hurt it will it?

 

[fechter]

Just double check the voltage at the board (between two wires or power resistor legs) to make sure you don't have two cells in series going to one shunt.

Flux around where the wires attach shouldn't be a problem. It's around the LM431 divider resistors where it can be an issue.
All the voltages were correct when I was looking at it, so it's strange that something changed. I suppose there could be another intermittant solder joint, again it would be most likely around the input to the LM431.

One thing I noticed on your board was many of the solder joints looked "dry" as though the flux was completely cooked off and the surface oxidized. This could result from a tip temperature that is too high, or from heating the joint for too long.
One way to rework that would be to add a tiny dab of paste flux to the suspect joints and heat them just long enough to get a melt. No solder needed unless the hole is not filled. The flux will do its thing and reflow the solder nicely. After it cools off, then you try to remove the excess flux (not so easy). The solder should look smooth and shiny when it's done right. Heating time should be very fast, like about one second.

On the other hand, maybe all your connections are fine, and there could be something else going on.

The part of the circuit that turns the resistor on does not have many parts, so the list of possible suspects is short. The LM431 and its divider resistors determine the voltage where the shunt turns on, and the PNP transistor does the work. Most common failure mode for both semiconductors is to short out so they are stuck on. I think in this case the shunt would always be on no matter what the cell voltage was.