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

[alex_audi]

Hi that's my bms capacity from bmsbattery.com

It 's connected to 16 headway 38120 10Ah
The charger is a 400 watt charger for lifepo set at 58,4 v
After 1 hour charging, cells in the negative pack side go in over voltage.
This cause stop and go the charger. It's impossible to arrive to 58,4v because other cells are at 3,40v.
More one goes towards positive and the voltage decreases.
First cell in negative side is 3,9v and last cells is 3,3v

Is this a bad BMS ?
Why does this happen ?
Do your bms have this problem or i can buy it without any problem with headway cells?

 

[GGoodrum]

I think we have the layout changes sorted out now. I was able to keep the width the same (75mm...), and keep the number of channels per board at six. I had to increase the length, though, back to the original 100mm, or just under 4". I'm also going to do an 8-channel board that is 100mm wide, which will fit in the next size up Hammond box.

I've now completed testing the new functionality in the charge controller, the new adjustable current limiter, which can provide the missing CC mode for use with inexpensive power supplies. The "smarts" for a controlling the complete charge profile are now all on the charge controller card, an adjustable CC mode, followed by individual cell CV modes. I tested the new logic on a simplified charge controller card that I use with 6s4p LVC/HVC-only boards. In this type of "split" BMS setup, the LVC/HVC boards get buried in the packs, and then the shunt boards are used in an external fashion, as balancers. The reason for this is that typically, these Turnigy and Zippy LiPo-based packs just don't go out-of-balance very easily. In fact, the only significant drift I've seen with these is when I take the packs all the way down to where the LVC circuits start tripping. Anyway, most of the time I just need to bulk charge the pack, but I still want the individual cell overcharge protection, during the charge process, so I split out the charge controller into a separate small box, that is used in series with the supply. When the packs do start to get a bit out-of-balance, I plug in the shunt balancer boards, and then balance charge the pack.

Here's a couple shots of the simplified charge controller:

View attachment Simplified Charge Controller-v.4.0.6-01.jpg
View attachment Simplified Charge Controller-v.4.0.6-02.jpg


What makes it "simplified", vs the version that will be with the regular BMS, is that the timer is omitted. In its place, there is an auto-shutoff function that shuts everything down, when the current drops below about 200 mA. This feature also ensures that the charger control logic doesn't stay on if the charger/supply is disconnected.

The new charge controller card also has a completely isolated LVC opto output, which is essentially disabled, during charging. This will keep an LVC buzzer or other alarm from going off once the HVC circuits start activating.

Anyway, we will have new BMS boards this week, so if all goes well, they will be made available sooner, than later. :lol:

-- Gary

 

[dbaker]

Great news, Gary. I looked for this on your TP Packs site. Will you be posting circuit boards there with a parts list link?

Thanks for all of your (and Richard's) hard and frustrating work developing the BMS for the ebike community.

Dave

 

[manfred59]

Good news!
I am waiting for the new board because my v.2.3 makes big problems. I had this problem once before, but now it is even more stubborn than ever. Perhaps I connect the BMS to the batterypack in the wrong order? I am doing like this:
First I connect the multi-pin connector to my 16s battery pack (PSU is not pluged), then I connect the charger output to the BMS (red main led lights) and when I connect the power coming out of the BMS to the batterypack (nearly every try) one of the first or/and the last KSA931 or the LM431 immediately are damaged. The first and the last orange leds shine. I have changed these parts about ten times but without success, every time the same game. What could I do? Changing the KSA931-PNP transistor to a stronger one?

 

[fechter]

Manfred, I'm not sure I understand your sequence, but if the charger (PSU?) is connected to the board when you plug in the multi-pin tap wires, it will blow things up. It does not matter if the charger is on. All the cell connections to the BMS board must be made with the charger disconnected.

Most people leave the BMS board permanently connected.

In your case, it would be possible to add a unidirectional transient voltage suppressor across each cell circuit to prevent any damaging spikes. These things look like diodes.

Gary and I are also working on a "split" version of the new BMS that has a small HVC/LVC section that always stays with the pack, and a detachable power stage for the big shunt resistors. The HVC/LVC board does not generate heat and can be wrapped up with the pack. All the sensitive parts stay permanently connected to the cells. The split version is still in prototyping stage, so don't hold your breath.

 

[GGoodrum]

Good news!
I am waiting for the new board because my v.2.3 makes big problems. I had this problem once before, but now it is even more stubborn than ever. Perhaps I connect the BMS to the batterypack in the wrong order? I am doing like this:
First I connect the multi-pin connector to my 16s battery pack (PSU is not pluged), then I connect the charger output to the BMS (red main led lights) and when I connect the power coming out of the BMS to the batterypack (nearly every try) one of the first or/and the last KSA931 or the LM431 immediately are damaged. The first and the last orange leds shine. I have changed these parts about ten times but without success, every time the same game. What could I do? Changing the KSA931-PNP transistor to a stronger one?

I think what you are seeing is a voltage drop, in the main charge wires. Try running extra wires from the Pack - and Pack + connections, on the board, to the main pack +/- connections.

-- Gary

 

[GGoodrum]

Gary and I are also working on a "split" version of the new BMS that has a small HVC/LVC section that always stays with the pack, and a detachable power stage for the big shunt resistors. The HVC/LVC board does not generate heat and can be wrapped up with the pack. All the sensitive parts stay permanently connected to the cells. The split version is still in prototyping stage, so don't hold your breath.

Here's what the LVC/HVC boards look like:

View attachment 6s LVC-HVC-02.jpg


This version is for six channels, but I'm also doing one for 8 channels.

-- Gary

 

[manfred59]

@fechter

Manfred, I'm not sure I understand your sequence, but if the charger (PSU?) is connected to the board when you plug in the multi-pin tap wires, it will blow things up.

My BMS board is not permanently connected
1. the charger is off
2. connecting the multi-pin tap wires via one big connector (ok)
3. connecting the three powerpole connectors coming from the charger to the BMS board (ok)
4. connecting the batterypack main connector to the BMS board (bang - - the first and the last led lit and the KSA931's are killed some times the LM 431 too)

In your case, it would be possible to add a unidirectional transient voltage suppressor across each cell circuit to prevent any damaging spikes. These things look like diodes.

Does that mean, that I have to install 16 voltage supressors directly to the battery pack? Or is it possible to install only one at the main battery pack output?
Fechter could you say exactly what unidirectional transient voltage suppressor I should use? http://www.diotec.com/service/files/tvs-dioden.pdf.

 

[AndyH]

My BMS board is not permanently connected
1. the charger is off
2. connecting the multi-pin tap wires via one big connector (ok)
3. connecting the three powerpole connectors coming from the charger to the BMS board (ok)
4. connecting the batterypack main connector to the BMS board (bang - - the first and the last led lit and the KSA931's are killed some times the LM 431 too)

Manfred - if you swap steps 3 and 4 you'll get a long life from the BMS. This is exactly what I did with my V2.3 board for about eight months. As long as the charger was not connected, I can connected either balance plug first, or the pack +/- Andersons first - doesn't matter. Connect all of the BMS to the pack first, charger last, and it just seems to work. Andy

View attachment bms.jpg

 

[ejonesss]

what is the dimensions of the boards?

i know it is probably under 2 inches more like 1.5 inches wide but how long is the board?

Gary and I are also working on a "split" version of the new BMS that has a small HVC/LVC section that always stays with the pack, and a detachable power stage for the big shunt resistors. The HVC/LVC board does not generate heat and can be wrapped up with the pack. All the sensitive parts stay permanently connected to the cells. The split version is still in prototyping stage, so don't hold your breath.

Here's what the LVC/HVC boards look like:




This version is for six channels, but I'm also doing one for 8 channels.

-- Gary

 

[ejonesss]

can placing a normal diode like those found on relays that protect the driver transistor work?

also how can the cells place a spike since there is no inductors (unless the resistors are wire wound)?

My BMS board is not permanently connected
1. the charger is off
2. connecting the multi-pin tap wires via one big connector (ok)
3. connecting the three powerpole connectors coming from the charger to the BMS board (ok)
4. connecting the batterypack main connector to the BMS board (bang - - the first and the last led lit and the KSA931's are killed some times the LM 431 too)

Manfred - if you swap steps 3 and 4 you'll get a long life from the BMS. This is exactly what I did with my V2.3 board for about eight months. As long as the charger was not connected, I can connected either balance plug first, or the pack +/- Andersons first - doesn't matter. Connect all of the BMS to the pack first, charger last, and it just seems to work. Andy

 

[GGoodrum]

what is the dimensions of the boards?

i know it is probably under 2 inches more like 1.5 inches wide but how long is the board?

The LVC/HVC boards, including the full 6s4p parallel adapter are 1.4" wide and 3.25" long.

-- Gary

 

[ejonesss]

ggoodrum and fechter: are you getting close to making the boards for sale?

i am getting worked up now for them .

also are they going to be just the naked board and we buy the parts like last time or are they going to be kitted or pre built?

i hope the first option because the boards being 3.25 inches long makes them too long to fit inside the cyclone battery tubes and for my project i can simply not install the other jst hx connectors and trim the board to fit.

 

[fechter]

My BMS board is not permanently connected
1. the charger is off
2. connecting the multi-pin tap wires via one big connector (ok)
3. connecting the three powerpole connectors coming from the charger to the BMS board (ok)
4. connecting the batterypack main connector to the BMS board (bang - - the first and the last led lit and the KSA931's are killed some times the LM 431 too)

Manfred - if you swap steps 3 and 4 you'll get a long life from the BMS. This is exactly what I did with my V2.3 board for about eight months. As long as the charger was not connected, I can connected either balance plug first, or the pack +/- Andersons first - doesn't matter. Connect all of the BMS to the pack first, charger last, and it just seems to work. Andy

Yes, I think Andy is right.

The problem is there are big capacitors sitting at zero volts inside the charger supply. When the main pack connection is made, there is a path from the batteries back through the body diode in the BMS FET to the charger. This creates a huge current spike as the pack charges up the capacitors in the charging supply.

If all the connections between the pack and the BMS are made before connecting the charger, then the high current path will be directed through the large wires that can handle it.

On the TVS diodes, something like a Mouser# 576-P6KE10 or 576-P6KE10C (leaded) or 576-SMBJ8.0A (surface mount) would be possible parts. These are like 8v-9v versions, which is needed to keep the standby drain current low. These are about $0.20ea, so not too bad. I've never tried these things, but they should prevent damaging voltage levels on the cell circuits. If you hook up some wires the wrong way, I think they would fail shorted to try and protect the cell circuits. These things might fit on an existing board between cell connections.

 

[manfred59]

Manfred - if you swap steps 3 and 4 you'll get a long life from the BMS

Yes, thank you Andy, this was my fault, now the BMS is working perfectly.
First do all connections to the BMS an then connect the charger!

 

[GGoodrum]

Okay, boards are on the way, finally. Here's what they look like:

View attachment 18s BMS-v4.0.6a.png


There's several, "last minute" changes/additions. First of all, we widened the traces, between cells to .12", and added holes that can be used to add wires, if need be. I actually don't think these will be needed, especially if you just add solder to the wide traces, which is pretty easy to do. Even this is only necessary for 1A+ shunt currents. We've added two filter caps that, in conjunction with the wider traces, kills the oscillation problems we were seeing before.

One new feature is the addition of a temp-controlled switch, and power resistor, for a fan (or fans...), on each shunt board. This eliminates the need for a fan switch on the controller section. This also lets the fans run a bit longer, after the charge/balance process stops, as the shunt resistors and transistors will still be warm. The temp switch comes on at 45C.

The controller section has the new CC mode adjustable current limiter, which also includes a function that shuts off the power when the current drops below 200mA. This keeps the controller logic from staying on if the charger/supply cuts off prematurely. The adjustment pot can be used to set the initial charge current from full off to full on. Anyway, I've tried this with unmodified Meanwell S-350-24 supplies, and it works great. I have it set for 14A.

Also new is a completely isolated LVC output, which is disabled when charge controller logic is on. This will keep an external LVC alarm from going off, once the throttling starts.

The balance timer is activated the first time a cell hits the HVC point, and the "throttling" starts. The time can be set from essentially zero time (30 seconds...) up to four hours.

I'm working on the preliminary instructions, and Andy has done a preliminary BOM, which I will post as soon as it is complete. I will update my website this weekend, so that when the boards get here, I can make them available immediately. Initially, it will be PCBs only, but once we come up for air, I will look into pre-assembled versions.

-- Gary

 

[DaveAK]

Great news Gary! This will be the final piece of my project, but the question is, should I DIY it, or wait until you or Andy have pre-assembled units? I know how to solder, and have done small projects but is there anything to assembling this board that would put it out of reach of a newbie like myself?

 

[GGoodrum]

Great news Gary! ... is there anything to assembling this board that would put it out of reach of a newbie like myself?

Only a lack of a clear set of assembly/test instructions, which I'm working on now. Seriously, although a bit time consuming, it goes together pretty easy. We made some improvements in showing how parts are oriented, and have tried to label everything. It should be easier than the v2.x series.

-- Gary

 

[DaveAK]

Great news Gary! ... is there anything to assembling this board that would put it out of reach of a newbie like myself?

Only a lack of a clear set of assembly/test instructions, which I'm working on now. Seriously, although a bit time consuming, it goes together pretty easy. We made some improvements in showing how parts are oriented, and have tried to label everything. It should be easier than the v2.x series.

-- Gary

Just the kind of news I like to hear. There are a few small items I need from Mouser, so I'll hold off until the BMS BOM is ready. Next job will be to clear my desk so I actually have room to make this.

 

[fechter]

For about the last two months I've been spending about every minute of spare time working on the Ver4 system. I'm lucky if I get more than 5 hours of sleep a night. The folder on my computer for Ver4 stuff has over 300 files of schematics, datasheets and board layouts. The workbench is a disaster area.

Anyway, after all that, it finally seems to be coming together and I have high expectations for the most current design which we should have back from the board makers sometime next week for testing. There is always a chance that something will need to be tweaked anytime the layout changes, but we should get that verified fairly quickly.

I'll be starting a Ver.4 thread soon, with more details on the design. This thread is looooong enough!

 

[AndyH]

Great news Gary! This will be the final piece of my project, but the question is, should I DIY it, or wait until you or Andy have pre-assembled units? I know how to solder, and have done small projects but is there anything to assembling this board that would put it out of reach of a newbie like myself?

Please DIY it - or at least let Gary and Richard get some sleep first.

Seriously - after building about 50 V2.3, V2.5 and 2.6 boards and three complete V4.x assemblies, I can say that the new units will be faster and easier to build. Gary has done an excellent job with layout and it's much easier to insert a bunch of parts and solder the group without thinking about 'build order' or interferance from other parts. The modular design makes it easier to build over multiple work sessions. The balance lead connectors and terminal blocks make it much quicker to wire the finished BMS. I really like the temperature-controlled fans!

We used to joke in the Air Force that when the weight of the pre-flight paperwork equaled the weight of the airplane the airplane was allowed to fly. I think that the weight of paperwork (and plasma-related-events) is more than sufficient to show that this will be a very worthy successor to the V2.x series BMS.

 

[sjoerdvdd]

Naa, i'll wait for the V5..

No seriously, great job, can i donate some extra money when i order these boards from tppacks?

Kind regards

 

[DaveAK]

Please DIY it - or at least let Gary and Richard get some sleep first.

It'll be my pleasure.

Seriously - after building about 50 V2.3, V2.5 and 2.6 boards and three complete V4.x assemblies, I can say that the new units will be faster and easier to build. Gary has done an excellent job with layout and it's much easier to insert a bunch of parts and solder the group without thinking about 'build order' or interferance from other parts. The modular design makes it easier to build over multiple work sessions. The balance lead connectors and terminal blocks make it much quicker to wire the finished BMS. I really like the temperature-controlled fans!

I'm sure it's within my capabilities, just trying to be cautious as I've only done a few small boards up until now. But you make an excellent point about the modular approach that will allow me to tackle just one board at a time. I'm looking forward to it!

 

[zega]

I have played with something similar some time ago, and it had the same/similar flaw...
In some cases current flow during turn on or some glitch, can run trough TL431 and trough BE path of bleeding BJT, causing too high base current, frying BJT and sometimes TL431...
I would add resistor in series with base of the BJT to limit base current (at least 100 ohms), and protect TL431 (altrouh i know that 431 has a limited output current but (just in case).
this all deal is from fig12 in the datasheet but load resistor is placed a little bit different and that circuit is little bit more safer, but i would not trust it anyway...
Even when i do constant current circuitry with 2 BJT-s i typically place resistor in base from shunt, it makes circuit more stable...

 

[ejonesss]

nice idea.

unfortunately being so close to the release of the board would result in ggoodrum having a $70+ pile of useless boards

http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=5416&start=1742 the first post was made on by GGoodrum » Fri Apr 02, 2010 3:52 pm that was on the doorstep of the weekend.

ggoodrum if you have not sent in the order yet then i think the change can be added .

if not can the transistor be replaced with one of higher value or 2 be paralleled or a trace be scraped away and the resistor connected in the gap?

I have played with something similar some time ago, and it had the same/similar flaw...
In some cases current flow during turn on or some glitch, can run trough TL431 and trough BE path of bleeding BJT, causing too high base current, frying BJT and sometimes TL431...
I would add resistor in series with base of the BJT to limit base current (at least 100 ohms), and protect TL431 (altrouh i know that 431 has a limited output current but (just in case).
this all deal is from fig12 in the datasheet but load resistor is placed a little bit different and that circuit is little bit more safer, but i would not trust it anyway...
Even when i do constant current circuitry with 2 BJT-s i typically place resistor in base from shunt, it makes circuit more stable...