Low cost BMS

[kovgab3]

Originally Posted by harlequin2

All this to shift the heat dissipation from a resistor to a transistor.

Not "ONLY" for that.
But to raise the dissipating current in a quite cheap way.
As you also had written some posts before, if you want to dissipate with 5A, then you need a large and expensive high power resistor. And it's very difficult to remove the heat that gets generated there.
If you use a resistor with metal housing, that will be even much more expensive, and that's a pain to assembly onto a heatsink.

If you use a "cheap" FET, then the heatsink is easy to attach. If you monitor the temp. of the FET, and you play with the PWM of the FET (and the current through on that), then the FET lives long, and dissipates much more current, then the resistor.

[kovgab3]

Originally Posted by filip

, but we
could screw the transistor to the positive battery terminal and have it dump
the heat into the battery.

That's definitely a bad idea, if you have LiFePO batteries.
The heat kills the electrodes.
ScienceDirect.com - Journal of Power Sources - A review on the key issues for lithium-ion battery management in electric vehicles

I don't know, how the Pb batteries react to heat...

[harlequin2]

OK, if you want to use a balancing current of maybe 5 A, then you have 18 watts to dissipate and a fet bolted to a heatsink might be a reasonable way to do it. In my car, with 45 cells, that would mean 810 watts to get rid of. Not very efficient is it?
Practically, if you want to use a balancing current significantly greater than 1/2 amp or so, then you really have to move to an active balancing system where power from a "high" cell is shunted to a "low" one. There are some quite good designs out there for such things, but they all cost a great deal more than the simple, lossy resistor.
By far the simplest and most cost effective way to do it is to balance all your cells before installing them into your battery and then keep them in balance with the little 1/2 amp shunt which is all it takes once you have done the initial balance.

[kovgab3]

You are totally right.
However, if you have to replace 1-2 cells in your pack, then you might think other way about the necessity of the 5A dissipating modules

An other "reason" can be if your charger can not decrease the current on an outer signal. If you have to switch it off when the first (old) cell reaches the limit, then the new ones will never get charged to their capacity.

[kovgab3]

Hello Harlequin,

I attached 3 images, please check them.

I was thinking about the wires between the cell modules.

On the first image I have rearranged your CellTop module schematic, and for the better visibility, I have removed everything that I don't need for the "experiment"

On the second image I put two of this circuits behind each other, and I added a shape representing the copper connector between the two battery cells, and I have added simple 2 lines for the twisted pair wire also.

Nothing special, just to show that the VCC of the first cell, and the GND of the next cell is connected.
And because of this, I can reposition the resistor R2, as shown on the third image.
This way we don't need the twisted pair, only 1 wire (just as they did: Battery Management System (BMS)).
However, this way the optocoupler becomes unnecessary also, am I right?

The BMS on the above link really doesn't use optos, and they can handle cells up to 255. They use twisted pairs only at the first and at the last cell, to connect them to the master unit. (and in one more case: if the battery pack is divided into multiple banks, and the banks are quite far from each other. Only the last module of Bank0 and the first module of Bank1 are connected this way.)

What is your opinion, can this make the design even simpler? (by the cost of the opto. Maybe a transistor needed to invert the signal, or it even can be handled inside the PIC?)

[harlequin2]

Yes, it can be connected that way and I realised that way back when I started this project. However, I wasn't sure how much interference the comms would suffer if one of the lines was also carrying the motor current with its high level of emi and so I decided the twisted pair would give the best protection. For that reason also I only run the data at 9600 bps which is quite enough to do the job and less prone to interference than higher Baud rates.
I haven't looked at doing away with the opto, but I guess that would be possible too.
I see the one described in the link you included runs at 56 kbps and uses a single wire, so maybe I was being too cautious. I have my doubts though, especially when I see a 2R7, 5 watt resistor being asked to carry 1.5 A of balancing current and thus dissipating 6 watts! In my experience, a 5 watt wire wound resistor can dissipate only about 2 watts in free air before catching fire.

[kovgab3]

I understand, thanks.

The e-motus BMS also uses PWM (you can configure the top limit for the dissipating current), and they also measure the temperature. However, the sensor is on the board, so it measures not the cell, but the PCB and the resistor together.
Hopefully they decrease the duty cycle, if the temp goes up

[harlequin2]

I'm sure they do that, but its not right to claim a balance current of 1.5 A, is it?
Also, I found that when balancing is turned on, there is a significant voltage drop along the connecting traces/wires and this affects the cell voltage reading unless you take care to separate the tracks carrying that current from the voltage sensing path. Their connecting wire to the +ve terminal shares both and will result in erroneous readings - unless they only measure the cell voltage while the pwm is in the off state.

[kovgab3]

these guys dissipate with 10A:
EV-Power | RT-BMS 2.0 - Cell Ballancing Unit (terminal) - High Power 10A
I'm sure, they have to deal with this problem.

[harlequin2]

Yah, they all seem to lack a basic understanding!
I re-iterate: If you balance your cells before installing them into your battery, a 1/2 A balancer is adequate to keep them that way, certainly for all the sizes commonly used by the home EV converter.
I balanced all mine with an individual cell charger made from an old 5 V/ 20 A computer power supply modified to put out 3.6 V instead of 5 V. Might take a while if you have a few hundred cells!
Also, weber and coulomb on the AEVA balanced theirs with only 1/2 A just by leaving the charger in balancing mode for a few days - they have about 220 cells.