.

Hello, then... lets go.

Here a circuit for 4s. Just as an explainer.

Sorry comments still in german.

You may think thats a first of an newbie in the electronics beginners lesson hour 1, then I understand your point.

But you are wrong.

Perhaps i describe how this has developed, without going into too much details, that would explain a lot. Interests can still be asked for.

About 3 months ago i stumbled over the following problem:

people of the mobile home field started to build 4s1p with lifepo 280Ah, and an "unknown" BMS with the typical 50 mA balance current. For the first installation, the cells had to be balanced, but even the minor differences of this where to much to be really be dealt with the 50 mA balance.

The resulting problem is, is it possible to place an "after burner" between BMS and battery, which measures the balancing current sunk from the BMS per cell, and add an additional higher sink current on the same cell ?

First problem, obviously shunts in all BMS lines and Op amps to detect current. No problem there.

(checked)

Second Problem, how is the correlation between the 5 currents and the 4 cells?

OOps!

It is 5 currents, some in two directions.

I assumed to solve that problem with trinar logik, but got lost shortly after trying to learn the basic laws.

The second aproach was truth table of all possible balance cases. That led in the end to binary logic equations for the 4 balance signals, assuming the different current directions as separate logic signals.

I even could press them in logic gatter circuit using 5 CMOS ICs, no processor necessary.

(Checked)

Third problem, sink current for the 4 cells. Except for heatsink, no problem.

(checked)

In total, not a really difficult circuitry, but still lots of stuff.

Then Chance, or luck, struck.

I got hands on the Type of the IC, what organized Voltage Measurement and Balancing : AN84910UA

On Page 13 - something I found details about the balancing procedure, which said that the

**circuit NEVER sinks two adjacent cells**. That was du to a slightly different connections of the balance resistors.

When i checked my truth table and eliminated all cases with adjacent cells, plus the all 4 and the nothing, then i got

**one simple specific current on one line for each cell.**

No logic anymore, just the shunts and the sink transistor!
Now, still the Shunts and their OPs where a nuisance. I had checked that the 1 Ohm / 50 mV i had assumed in the beginning was not interfering with the exact voltage measurement.

Could it be possible to direct drive a Transistor? Even a Germanium would need 190 mV. I started to dive into my Museum to find Germanium samples, but before i really started to test that i retreated back to a proper analysis.

And that was an a anlysis of a real LiIon battery, a JBD 10s BMS, some resistors and a small oszilloskope to see what happens. The battery was intentionally disbalanced, to have the balancer continouously working,

**on two adjacent cells. **

The result was, the JBD behaves exactly as the description in the datasheet.
It doesnt balance two adjacent cells simultanbeously. The voltage measurement is not disturbed by the baancing. the balancing is not disturbed by the 0.6 V loss of. the resistor / the Ube loss. That is because the JBD switches between balance current on, and current off , obviously for measurement purposes, with a ratio of (guessed) 80% to 20 %. That reveals the secret why the measurement is not disturbed at all, i even tried 150 Ohms, without transistor ( making 2 V loss).... no effect in Voltage measurement.

I did intentionally disbalance the battery to 1 cell lower as the rest, making the balancer try to balance each an every cell (9) except the lower one.

for two cells i had the balancer istalled, for another twi i had the 39 Ohm resistor in the line, the rest of the 6 cells where running as usual.

The disbalance was about 1 Ah. Would have been 20 Hours to balance, since the adjacent cells have to be balance alternating, about 40 hours.

the balancing Resistors where 3,9 Ohms, making the balance current short to 1 Amp. in effect, with the 80 % PWM about 0.8 Amp. Distributet for two cells 0.4 Amp.

And indeed i saw, as could be expected, in just over about 2 hours ( i didnt catch the exact minute) the balancing for this two cells ended.... and the reast took far more than another day.

Thats what i can tell at moment. The above circuit WORKS.

There is room for some improvement or different adjustments, where i give a few by statement:

- the balancing current is defined by the resistor. Up to 2 Amps, limited by Hfe from the transistors. Base resistor should be adjusted a bit.

-i have tested 2 A. there is really heat coming from the collector resistor. ;-)

- i have used bipolar power transistors, care for high Hfe. Darlingtons have higher Ube, see below. Mosfets go not low enoug in Opening voltage.

- you can use all NPN transistors by using 4 resistors in the balancer lines. Leave the plus wire without resistor because its the Plus Voltage for the BMS. I used the circuits above to save 2 resistors, just as piece of Art.

- the transistors do not really good switch down to Uce sat very well, its a plaything between base current, Hfe and the collector resistor. Evaluate. So there can be something like 0.8 V pass the transistor, giving need for cooling if you exceed 0,5-1 Amp in total.

-i have tested a darlington. It did work as well, but i have not thorougly tested it. This could be the way to really exceed 5 A, if you like, and should care for the heat sink.

- take in account for the heatsink that the total power loss of a coll. resistor PAIR is matching the single current, they can only be switched alternating.

- there is the possibility to double use the Collector resistors as well. means only 5 resistors for 10 cells. not tried yet. Not inherently safe.

-there is the possibility to implement automatic temperature regulation, means power limiting if the temp rises too much. ( no more components.... (Chuckle))

- the really lowest component count leaves the Collector resistors all off. The Power is sunk from the transistors, the current would be a function of base current and Hfe. The absolute component count would be 15 for 10 cells. But beware, not tested yet.....

- if you stay on, say, 0.5 Amps it can be a separate board with just two 5 pin JST XH Plugs, just to place it into the existing balance lines, without soldering.

- a board for 4s is ready, if you like.

Now guys, comments please.