Dear TI-Support Team,
I have an issue with the BQ76PL536A-Q1 used for cell surveillance in our BMS.
System overview:
Our automotive battery consists of two battery stacks with a 72s2p configuration resulting in a total of 144s2p connected cells. Each Stack is subdivided into 12 modules with a 6s2p configuration for simplified maintenance. Each module is supervised by one bq76pl536A-Q1. ). The two bottom BQ76´s of the Daisy Chain are both connected to a single µC via one digital Isolator ISO7241C per BQ76. Please note the system overview below.
Each of the voltage measurement inputs is protected by a melting fuse(VC0-, VC0+, VC1+,…., VC6+), which is required due to rules (fusable link wires or fusing like this are required for reasons of short circuit protection) of the racing series we participate in.
The complete modules are connected by power bars made of chemically nickel-plated aluminum. Please not the following partial electric circuit diagram.
(Please visit the site to view this file)
The safety fuses and the complete measuring track (including contact resistances etc.) have an electrical resistance of 1.5 – 3 Ohm. The capacity of the cells are 6250mAh and their inner serial resistance is below 1.5mOhm (in average 1.25mOhm, 99% of the cells between 1.1 and 1.38mOhm).
The problem:
As long as there is no charging current applied, the voltage measurement works perfectly. The measured Voltages differ by a maximum of 1mV from the voltage of the cells (verified with external measurement devices). The batteries are charged with the Constant Current Constant Voltage principle, even though from our data we never reach the constant voltage range and so far only charge at constant current.
Before charging alle cell voltages are at 3.8V +-50mV.
As soon as charging is started (two battery stacks are connected in series during charging) some of the measured cell voltages start moving towards 4.2V rapidly. The voltage measurement of the BQ76 is correct, at its entries we measure with external measurement devices the same voltage.
Moreover the measured voltages are not stable. They vary by +-20mV. Sometimes one voltage starts to run away or even bounces stronger (up to 100mV).
What strikes the eye is the fact that only the voltages at the end and at the beginning of each module are affected. These are the inputs from which the BQ76 is supplied as well. This leads us to the conclusion, that this effect is not due to damaged battery cells, because statistically it is highly unlikely that we would only have damaged cells in this area.
This issue is very critical for us, because it makes charging almost impossible. At very low charging currents (1A or below), the process is still unstable, but the bouncing does not exceed 4.2V, which would open the insulation relays and by that terminating the charging process. Charging at these low currents causes charging times up to 13h, which is absolutely unacceptable during racing events (we only have a few hours for charging).
After terminating charging, all measured voltages go back to normal, which means a range of +-50mV. The first and the last cell of the modules do not show any higher state of charge than the other cells, they are normally distributed within this range.
The circuit design is from our point of view absolutely equal to the reference design displayed in the datasheet. The only difference is the resistance of the measuring track.
To give you an idea how the measurements look like, here are some graphs made of data taken of the battery can. All the graphs show the highest and the lowest voltage of each Battery over a period of time.
This graph shows a part of a constant charging process at 1A. As you can see the highest measured voltage of the left battery, the dark blue curve, (the batteries are located at the side of the car) drops down and rises up again without any further changes of the parameters. The temperature during this process was constant and this voltage was measured at the first cell in the fifth module (always counted from the lowest potential).
This graph shows the same measurement, only a few moments later. Suddenly there is a huge voltage drop, even though the current remained constant.
The next graph shows that there is some kind of correlation with current, even though the highest voltage of the container this time is not measured at module 5 cell 6, but at Module 12 cell 1. The lowest voltage of the same battery shows a similar behavior, but it is weaker. The lowest voltage was measured at module 9 cell 6.
All the previous measurements only show highest and lowest cell, but the cells surrounding these bouncing cells are not shown. For this reason another measurement was recorded, showing not only the lowest and the highest voltage, but also the cells around the bouncing cell. The bounding cell is in module 12 on the left side, the first cell. So the Voltages of cell 5 and 6 of module 11 and cell 2 and 3 of module 12 are displayed, which are the direct “neighbors” of the bouncing cell. These cells do not show any special behavior, even when we applied this current change.
Even though these measurements only illustrate strange behavior in the left battery, we observed the same kind of behavior in the right battery as well. Sometimes they coexist on both sides and the affected modules change without a system (as far as we recognized by now).
I hope you can help our team to find the reason behind this mistake. We have no ideas left, how we can solve this issue.
If you need any further information (pictures of the batteries, electric schematics, layouts etc.), please do not hesitate to contact me via private message. I would not like to upload all data to the public forum.
Thank you very much!
Best regards from Stuttgart, Germany,
B. Spatafora