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

[mikereidis]

I may also be interested in tear offs. I just started reading this thread and am trying to come up to speed.

Some of us in Canada are using Canadian Tire/Yardworks batteries: http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=4908

They are Lithium Manganese 6 amp-hour, 5 cells per battery, and come in a case with BMS for $110. The BMSs having been giving me grief so I've gone BMS-less, but want to build and/or buy something better and configurable. Eventually I want a way to hook my laptop or whatever to the BMS and be able to read all cell voltages; perhaps one A->D converter and something to switch both + and - to 20 different cells; 20 A->Ds might be nice, but costly, and I don't feel I need constant millisecond level monitoring of every cell voltage.

I'm using 4 batteries in series currently, for fully charged voltage of 84v and don't expect to go higher at this point. So I need a solution for 20 cells.

Each LiMn cells needs LVC at 3v or so, and reaches full charge at 4.2v or so. So I'd have to mod the relevant voltages.

My current "dangerous charger" mostly consists of a full wave bridge rectifier, connected in series with a 120vac "switchable load" (heaters, lights on dimmers) and the 120vac line. Yes, it works But over-charging is too easy if you walk away for too long. It provides a pseudo CC mode, and when CV mode is needed, it's simulated manually by turning off heaters and semi-constantly adjusting the dimmer. All charging equipment is, and will be, mounted on bike.

Yes, I have some somewhat bulged cells from overcharging. I may remove them and use for bike accessories etc. And/or sacrificial cells ?

 

[Anonymous]

I have been meaning to ask how this BMS project could work with LiMn (YardWorks) also.

 

[mikereidis]

I have been meaning to ask how this BMS project could work with LiMn (YardWorks) also.

I searched this thread for LiMn and found a bit of info. Different resistors or pots (15-20 pots !) to set LVC I think, and a different TC54 set to 4.0, 4.1 or 4.2v for CV. I looked at Digikey and they don't seem to have stock or have high minimums for voltages like 4.x v, but the recommended Mousser seemed to be a better source for small quantities that are in stock. http://endless-sphere.com/forums/viewtopic.php?f=14&t=5416&p=92785&hilit=mouser#p92785

From the minimum of reading I've done on this thread so far, I guess the CV voltage and LVC voltages are set and fixed in hardware, and are done by one component, or set of components, per channel. So to change something would require X channels of pots to turn/calibrate or X channels of resistors/TCXXs be modified or replaced.

I presume that single voltage reference sources for all channels were considered in the design, and that option was rejected. Perhaps to make cell/channel expandability easier ?

I think I'd prefer to have one adjustable, calibrated and compensated voltage reference for LVC and one for CV for all cells (of same chemistry of course). These references would be fed into X channels of comparators or somesuch.

Anyway, not trying to "diss" a design I don't know too much about yet. Every design has tradeoffs and I'm just wondering if this design has the right tradeoffs for me.

 

[fechter]

Yes that was considered in the design, but due to the voltage offset of each cell, it was too expensive to implement. The microprocessor based BMS designs would allow reprogramming the voltge set points.

In this circuit, the low voltage cutout is pretty much stuck at 2.7v unless you get custom TC54's. I don't know what a safe low voltage level for LiMn is. My guess is 2.7v is high enough to prevent damage.

The upper voltage set point is easily changed by swapping a resistor value or by installing pots. 4.2v or whatever is easily within the range.

 

[biggs]

I have been meaning to ask how this BMS project could work with LiMn (YardWorks) also.

I searched this thread for LiMn and found a bit of info. Different resistors or pots (15-20 pots !) to set LVC I think, and a different TC54 set to 4.0, 4.1 or 4.2v for CV. I looked at Digikey and they don't seem to have stock or have high minimums for voltages like 4.x v, but the recommended Mousser seemed to be a better source for small quantities that are in stock. http://endless-sphere.com/forums/viewtopic.php?f=14&t=5416&p=92785&hilit=mouser#p92785

hello,

my name is alex and i am new to this forum but i am reading it for a while.

in the moment am building a BMS based on this thread.

i ordered my TC54CV210 directly from microchip they have an web shop and i got my parts promt to vienna.

i have paid a little more but i have got it.

i use a fet for shunting because i shunt 5A and with bipolar that is not possible. (Uce sat)

bg
alex

 

[mikereidis]

Yes that was considered in the design, but due to the voltage offset of each cell, it was too expensive to implement. The microprocessor based BMS designs would allow reprogramming the voltge set points.

In this circuit, the low voltage cutout is pretty much stuck at 2.7v unless you get custom TC54's. I don't know what a safe low voltage level for LiMn is. My guess is 2.7v is high enough to prevent damage.

The upper voltage set point is easily changed by swapping a resistor value or by installing pots. 4.2v or whatever is easily within the range.

OK, thanks. Is microprocessor based version being designed now ? I'm looking to get some A->D converters and something to switch Analog voltage inputs (And one or more current sensors) from different cells and have this in a USB peripheral. Then I could use any chip with USB or a laptop running Linux or whatever to do whatever code is capable of. I'd like to see that code under GPL license, and perhaps something similar for design/schematics.

Normally, LiMn is LVC'd at 3.0v, but I've heard that it may be OK to go lower than that under load for short periods, as long as open circuit voltage remains above 3.0v.

At this time, I'm thinking I will create a custom, simple and expandable/modular BMS, but I'd like to "steal/borrow" whatever good ideas and experiences have come from this and other BMS designs.

 

[fechter]

OK, thanks. Is microprocessor based version being designed now ? I'm looking to get some A->D converters and something to switch Analog voltage inputs (And one or more current sensors) from different cells and have this in a USB peripheral. Then I could use any chip with USB or a laptop running Linux or whatever to do whatever code is capable of. I'd like to see that code under GPL license, and perhaps something similar for design/schematics.
....

At this time, I'm thinking I will create a custom, simple and expandable/modular BMS, but I'd like to "steal/borrow" whatever good ideas and experiences have come from this and other BMS designs.

I'm no longer working on a MCU based BMS, but many others are: http://endless-sphere.com/forums/viewtopic.php?f=14&t=3486
I suck at writing code, so I'm staying analog for now.

 

[GGoodrum]

Update...

Richard and I have a new version to test that should implement thefeature where the charge current will be cutoff completely at the end of the charge cycle, and the gree portion of the main LED will come on solid. Basically, we went back to what we had in Bob's original versions and added a diode matrix that generates two signals, "ANY SHUNT ACTIVE", and "ALL SHUNTS ACTIVE". Richard has come up with a scheme that uses an SCR to latch the gate driver so that the FET is cut off, once the "ALL SHUNTS ACTIVE" signal is activated. In theory, at least, this won't come on until all the shunt optos stop oscillating, which only will happen when the cells are all as full as they are going to get. We need to test this, of course, but at least we know the diode matrix logic works.

My plan is to turn in a new run of boards, in a smaller quantity, on Monday. I should get these back by Wednesday. We'll test the new changes, and assuming we don't need to do anything more than maybe tweak a value, here and there, I will start making the rest of the boards available, and I'll order a larger quantity. Since there has been enough interest in the "tear-off" concept, this is what I will go with for this next version. This will support versions from 8 channels all the way up to 24 channels.

-- Gary

 

[JB_EB]

Gary & Richard, that's great news. Hope the testing goes well. I'm itching to get an 8-way board.

JohnB

 

[PJD]

Gary,

Can you give us some dimensions for the tear-off boards?

 

[GGoodrum]

Gary,

Can you give us some dimensions for the tear-off boards?

Right now it is 3" by 11", but it will end up a bit wider, in order to make room for the "ANY SHUNT ACTIVE" and "ALL SHUNTS ACTIVE" signal lines.

-- Gary

 

[Deepkimchi]

Sh--t, I'll be just interested in getting a board..

Been running up to 7-8 ah just to be safe.

DK

 

[michaelplogue]

Subscribed - VERY interested in 24 unit bms as well......

 

[GGoodrum]

Update...

I finally got enough time to do the new layout, which includes the logic required to shut off the charge current when all the shunts are in full bypass mode, which says the cells are as full as they are going to get. Here's what the PCB now looks like:

The dimensions are 3.325" by 11.800". The plan is to test the new logic this weekend and then assuming no major changes I'll get a bunch more boards made by mid-next week.

-- Gary

 

[michaelplogue]

Nice! So, how many amps would this BMS be rated for?

.

 

[PJD]

Couple questions:

1. Shouldn't you use multiples of four cells on each board?

2. This system board doesn't seem as space efficient for 16-cell applications. is it possible for you to still make the original design - just the PC board available?

Thanks...

 

[GGoodrum]

Nice! So, how many amps would this BMS be rated for?

.

For discharge, it doesn't matter, as there is no current control. The LVC portion of the circuit monitors each cell, and then trips the controller brake line if any cell dips below 2.1V. Since all the current always has to flow through all the cells, you can do current limiting at the pack level. The controller and/or a CA can handle this just fine, and you can add a really big fuse as a faile-safe, last resort level of protection. The only reason most of the Chinese BMS variants come with current limiting functions is because in most cases the controllers can draw more than the cells can safely put out, so they need to have the BMS limit the current. Even if you use this new BMS with the cheaper, less powerful cells, you can limit the current using a CA, or by adjusting the current limit on the controller.

For charging purposes, the FET is rated high enough that current limiting should not be an issue, with the BMS anyway. I think a "practical" limit is around 30A, due to the trace sizes on the board for the FET. The shunt control part of the logic doesn't kick in until the cells are almost full, and by then the current is down to around an amp.

-- Gary

 

[GGoodrum]

Couple questions:

1. Shouldn't you use multiples of four cells on each board?

2. This system board doesn't seem as space efficient for 16-cell applications. is it possible for you to still make the original design - just the PC board available?

Thanks...

1. Not sure I understand what you mean here, because the board is laid out in multiples of four cells.

2. Actually, the old layout would have to grow a bit in order to fit the new charge cutoff logic, so the differences are not that much. By having "one size fits all"-type of design, the board cost can be less because I can get a large quantity of one size. Sorry, but I'm not interested in doing a small quantity of the old design, and I'm not going to spend the time it would take to add the charge cutoff logic to that version either.

I can't seem to please everybody, so I'm going to go with what pleases me for a change. If I added up what all I've spent in the development of these, and subtracted what I've received for the ones I've sold, I think it would work out that the few BMS boards that actually make it into my own setups will have cost on the order of hundreds of dollars each.

-- Gary

 

[rf]

The new board looks perfect, Gary! Perfect size. Suspect they'll disappear like hotcakes.

Your efforts are most appreciated.

Now I can upgrade my 24-volt electric skateboard, too. Finally, a universal BMS!


Richard

 

[flashedarling]

This 24 cell board looks to be perfect for my situation. I'm planning on using 24 Thundersky 60ah cells and it sounds like 30A till the first shunt turns on sounds like it should be enough to charge the pack in short order. Is there anything I'm missing about this working with these larger cells? The only thing I'm seeing is that I thought the recommended LCV for thunderskys was 2.6 rather than 2.1V. Could this be changed with a simple part swap? I'm really excited to see these boards.

 

[rf]

This 24 cell board looks to be perfect for my situation. I'm planning on using 24 Thundersky 60ah cells and it sounds like 30A till the first shunt turns on sounds like it should be enough to charge the pack in short order. Is there anything I'm missing about this working with these larger cells? The only thing I'm seeing is that I thought the recommended LCV for thunderskys was 2.6 rather than 2.1V. Could this be changed with a simple part swap? I'm really excited to see these boards.

Yes, the LVC voltage is a simple choice of parts. I think the closest value for you may be 2.7 volts.

Richard

 

[GGoodrum]

Now I can upgrade my 24-volt electric skateboard, too.

 

[GGoodrum]

This 24 cell board looks to be perfect for my situation. I'm planning on using 24 Thundersky 60ah cells and it sounds like 30A till the first shunt turns on sounds like it should be enough to charge the pack in short order. Is there anything I'm missing about this working with these larger cells? The only thing I'm seeing is that I thought the recommended LCV for thunderskys was 2.6 rather than 2.1V. Could this be changed with a simple part swap? I'm really excited to see these boards.

Richard is right. There is a 2.7V version of the TC54 voltage detector. I used to use these with my a12e3-based packs.

-- Gary

 

[fechter]

This 24 cell board looks to be perfect for my situation. I'm planning on using 24 Thundersky 60ah cells and it sounds like 30A till the first shunt turns on sounds like it should be enough to charge the pack in short order. Is there anything I'm missing about this working with these larger cells? The only thing I'm seeing is that I thought the recommended LCV for thunderskys was 2.6 rather than 2.1V. Could this be changed with a simple part swap? I'm really excited to see these boards.

Besides getting the 2.7v TC54's, you might have heating issues with the charge control FET at 30 amps. Nobody has ever tried one that high, so we don't know for sure, but the math says you'll need a heat sink on the FET above 20 amps or so. It could be a fairly small heat sink.

 

[cycle9]

So, the board looks great, I want one, or several! How is the best way to order those? I tried emailing you twice at TPpacks, but no response so far.
Thanks,
Morgan

Update...

I finally got enough time to do the new layout, which includes the logic required to shut off the charge current when all the shunts are in full bypass mode, which says the cells are as full as they are going to get. Here's what the PCB now looks like:

The dimensions are 3.325" by 11.800". The plan is to test the new logic this weekend and then assuming no major changes I'll get a bunch more boards made by mid-next week.

-- Gary