Describe experience with this chip.Initially, the project - Car control and protect the battery cells 40.Initially tried to use it in standalone mode. As a result, the lower limit was too low for lithium-ion batteries and too high for LiFePo4. But balancing and sets the threshold she allegedly executed. But after using the prototype turned out that even with a minimum set takeoff balancing battery cells 10 to 90 millivolts diverges between the most charged and discharged itself.Then we decided to go to the host control.In this case, it was decided to use this chip only as a switch for transferring cells to the received voltage level of the ADC
The result revealed several problems that do not quite fit together with the description in the datasheet:
1) In balancing work schedules can be seen that the voltage measurement - balancing transistor is closed for 120 microseconds for measurement. It is true - the current balancing disappears. But not everything is so rosy! If you choose to translate the output cell number 1 and then set the flag balancing her - at the output of one-to-one passed all transients balancing. So kakbudto balancing is not disabled and measurements are made all the time.
Fair Enough. This problem, I decided to program - just once per second for 10ms to read all the cells I cut all balancing.
2) With this algorithm is also having problems. It turns out - the first cell, which I beg to withdraw the output Wout - periodically (1-2 times per second) falls by 10-20 mV against its real voltage (even on the testimony of the BKYU)!
Exit and then was found - had 4 times the software to drive the switch cells between the 3rd and 8th cells. only consistent proofreading cells is a more or less adequate.
3) attempts to calibrate all the offsets within the chip does not lead to proper results. Sdeluya procedure from datasheet spent all the measurements and calculations and vseravno. 1st cell does not correspond to reality at 2 mV, 2 to 4, the third 8 and so on ... The 10th cell does not correspond to reality at 200 millivolts! And the chips and chip samples obtained from Aliekspress received almost no different!
Had added to each cell to use a linear approximation of typing can welcome gain and offset. result began to look like a reality.
4) No matter how I tried but sometimes when you restart the CPU host controller chip is locked forever. I understand that this is due to the cliff in the middle of a team but it should not be a cause complete failure of the serial interface. resuscitation occurs only after the ACK 100 via interface chip.
Also even when the host control load balancing still continues to work on the old algorithm - 120 microseconds on a regular basis is disabled for measuring voltage is within the chip is not affected my calibration factor and displacement as a result of control of the power transistors also had to endure an external controller.My personal opinion - chip ill-conceived and flawed.The very idea of this chip and its implementation is pretty good. But many small intractable problems, which in general should not be, spoil all the convenience of working with her.If someone uses it, too, in their designs - tell how you are struggling with these "features"?