# To the question of batteries

### If you read the inscription on the elephant’s cage: buffalo, do not believe your eyes

In one product under development, a rechargeable battery of considerable capacity was used from several parallel sections with six cans of LiIon battery cells of the type “1s1p MP 176065 IntegrationTM” for 6000mAh manufactured by Saft in each. One battery returned to us for repair and I decided to assess the degree of balance of individual cans after long-term use. To do this, it was necessary to disengage the banks, charge individually and evaluate the spread of the received charge.

But for starters, I decided to look at the voltage in the sections as a whole, expecting to see zero, because any discharged bank disconnects itself from the consumer circuit, and we must either see the voltage of at least 6 * 2.4V = 14.4V, or nothing.

Imagine my amazement when 12V was found on one section and 7V at all (I won’t write further on Volta and not because of lack of respect for the great Italian, but because of laziness).

The result is somewhat discouraging, we continue research and measure each jar in section at 12 - on four the expected voltage is 2.8-3.2, on two - 0, everything coincides in total, but why it is 0 on disconnected cells and not open circuit is not clear.

Okay, I suppose that a circuit break takes place, but it is not absolute, there is a small current through the control circuit, because it does not disconnect from external contacts - we look at the circuit (it is located at the end of the post, otherwise the whole intrigue disappears).

We connect the load to the section in the form of a 1k resistor (current 7mA), the output voltage drops slightly to 11.8, which means that the internal resistance of the control circuit is on the order of 0.2 / 7E-3 = 200/7 ~ 30 Ohms, as the resistance is not enough, I counted on a dozen kilos but, maybe it’s non-linear, although it’s still strange. We increase the current to 50mA, the voltage drops to 11.6, which corresponds to a resistance of 0.4 / 50E-3 = 400/50 = 20, we increase the current to 100mA, the voltage practically does not change, which gives 10 ohms.

Indeed, the resistance is nonlinear, but it is clearly bent in the wrong direction. Most of all, it looks like a direct branch of the I – V characteristic of the diode, but where it comes from is not clear. Okay, remove the protective film from the battery and remove the protection board (you understand that you cannot use the battery without it if you are not extreme). We see a typical circuit of two transistors and a voltage control chip, but on the reverse side of the board we observe a large element marked STPS104B.

Yeah, this is a Schottky diode, connected back to the polarity of the battery, that's where the voltage drop of ~ 0.2 per cell is practically independent of the current (at low currents). The riddle was solved, it was not difficult, now we must evaluate the technical solution of a well-known company.

The first is a definite plus - this scheme allows you to continue to draw energy from the section, even if one cell is completely discharged. Moreover, the total voltage drops by 2.4 (the minimum allowable per cell) + (0.2 ÷ 0.6) <= 3, but this is better than a complete shutdown, although nuances are possible.

The second is the weak minus - the diodes in general, and Schottky in particular, have reverse current, which will be a stray load and will reduce the available cell power. In this particular case, the reverse current will be no more than 100 μA (under normal conditions), that is, during the day of storage of the cell we will lose 0.1 * 24 = 2.4mAh ~ 0.04% capacity, very little compared to self-discharge of ~ 0.5%. By the way, the self-discharge current in the documentation for the cell is simply not indicated, I come up with general considerations and recommendations for re-charging the cells during long-term storage every 6 months.

Third - an invented minus - if all cells at the same time decide that they have reached the required level and are disconnected from external terminals to prevent overcharging (not a very likely event, but nonetheless), then the input voltage of the charger will be applied to a series of back-connected diodes and it may well turn out to be on one of them, which limits the charging voltage to the breakdown inverse value for the diode at 45 and, accordingly, the number of cells in the section 45 / 4.2 = 10. Although I’m not quite right about the idea - if all the cells are completely discharged (and this is possible with us) and disconnected from the external terminals, then at the moment the charge starts, there will be a similar situation, albeit not for long, therefore certain voltage limits should still be indicated .

But the fourth - a definite minus - the complete absence of an indication of the presence of such a circuitry in the documentation for the device. Well, you can’t do this, this is an essential feature of the device and it should be reflected in the ED. Although, probably, they taught me this way, and in modern engineering practice it is customary to neglect such considerations. In the end, I don’t have a firm of the SAFT level, and those who produce these cells with such documentation have it and they feel good on the market. And what do you think? - poll at the end of the post.

Well, actually on the subject of the study, it turned out that after ~ 100 charge cycles (without balancing) and discharge (taking into account the features described above), the spread of the residual charge in individual cells was 12%, the figure is not negligible, but I expected the worst, so that offset.

Let us evaluate the possible efficiency of the standard passive balancing circuit, for which we take a current through a ballast resistor of 50 mA (the standard value for Chinese devices), then when charging with a recommended current of 0.2 ° C and a charge time of not more than 7 hours, we get a possible decrease in cell charge of 50 * 7 = 350mAh, which corresponds to approximately 5% of the declared capacity. Actually, of course, it’s much less, because the shunt will turn on under constant voltage mode, and you still need to get to it, so this is an upper estimate, but at least something.

Then we can assume that the aforementioned variation in the cell capacity can be compensated for 5-10 charge cycles with balancing. That is, if balancing the non-integral part of the section is done, we can expect that the negative effects can be compensated completely in each cycle with these cells. Of course, all this is only an estimate and the exact answer to the question about the possibilities of balancing requires additional research.

And here is the promised scheme:

Only registered users can participate in the survey. Please come in.