li-ion series charging - equalising

[lleonidas]

Last weeks I ve been looking on series li-ion balancing circuits out there for a project I had in mind a topology behaving like a zener connected in parallel with each cell with a voltage knee around 4V and a transition from 0A@4V to say 1A@4.01V
The greatest problem was the temperature coefficient which had to be around 10mV for 40oC temperature swings.
Last days I started getting results and by combining subcircuits with different but tunable tempcos.
The circuit lives in Ltspice simulations only but I used active components for which I verified that Ltspice gives correct tempco behavior (for example an IRF6620 as the power MOSFET). It is fully tunable and predictable as of knee voltage and tempco (you can modify tempco from negative to positive by changing a single resistor)
I have not run a sensitivity analysis yet but I guess that with low tempco resistors the behavior will not change much and with random resistor variation I hope the circuit will remain to specs
The IV characteristic(current axis is log scale)i depicted below. Blue trace is at 0oC and red trace is at +40oC

What is your opinion on the result? Does it worth pursuing the idea?
As I am quite old now and not affiliated with any university professors what is the best way to have the circuit peer reviewed? Could you please recommend a Journal?
My job is physics teacher in a secondary school and I had an msc in microelectronics in the past but I am a little afraid to publish this. The most probable thing is that either the achievement is not important (if you know other circuits please point them to me) or other unforeseen problems of my idea. On the other hand it would be a same to not let other people use this idea if it is good.
The idea will probably have usability also in the field of I-V references and not only on charging series lion batteries


Thank you!

 

[KlausST]

HI,

impressive results.

but how is it related to a balancing circuit. Isn´t it just a limiter?

From a balancing circut I expect that each cell in a series has same voltage.
So when fully charged all cells maybe have 4.1V, but all the same, and when discharged all may have 3.0V ... agin all the same voltage.

My idea is when let´s say two cells are in a string. surely both never are perfectly matched in available capacity. So say one has 1000mAh and the other 900mAh. .. just as example values.
Now you charge both to 4.1V. both are full. But one may output 1000mAh until down to 3V while the other can only give 900mAh down to 3V.

So maybe you have an under discharge limiter of 6V (for both in series).
voltage curve approximated to https://siliconlightworks.com/li-ion-voltage (just a random source. Not knowing how valid these values are)

So after drawing 900mA the one is true empty at 3.0V while the other just gave 86% of it´s available charge. Resulting in a cell voltage of 3.7V.
So one cell is at the limit and the total series voltage still is 3.7V + 3.0V = 6.7V.

If you go down to the limiter´s voltage of 6.0V then the expectable cell voltages are around 2.4V and 3.6V.
2.4V is less than the recommended discharge voltage limit.

***
For me a balancing circuit tries to maintain equal voltages in each state of operation.
And a more advanced circuit should do this by pushing energy form the healthier cell to the weaker cell without much loss.

But maybe I´m over complicating things ... or expect too much from a charge balancer.

Klaus

 

[lleonidas]

"but how is it related to a balancing circuit. Isn´t it just a limiter?"

thank you for your reply!
for the charging application the circuit just bypasses a cell when full just like a parallel zener would do but with the ability for bypassing a hefty current and the small temperature dependence of knee voltage.
I do not care for the discharge protection as the baterry bank will be oversized for the application and can be checked every year for capacity degredation to ensure that remains oversized.
The only problem is when it is charged. I want to prevent cell overcharge without stop charging the rest not yet fully charged cells.

 

[KlausST]

Hi,

O.K. I understand.
Should work.

But have in mind that there will be huge power dissipation. Worst case: cell_voltage x charge_current.

Klaus

 

[BradtheRad]

A suggestion which might be almost as good...
Build monitor circuits, one for each cell in your battery pack. If a cell weakens, then its voltage drops under load. This triggers its monitor to activate an led, or beeper, or wireless signal, etc.

It should be possible by using a stack of window comparators.