BQ4050: battery capacity jump

Hi Howard,

Have you configured the CEDV parameters for the gauge? This is necessary for gauging accuracy. When the SOC drops to 7% it means that the EDV2 voltage has been reached. Please see the online GPCCEDV tool for more information on this process: http://www.ti.com/tool/GPCCEDV

Here is the recommended flow for using fixed EDV or compensated EDV (better). This is from the app note 'Using the BQ78350-R1' gas gauging section. (BQ78350-R1 is also a CEDV gauge so the instructions are similar).

Setting Design Capacity

The default maximum chemical capacity of the battery is 4400 mAh. Design Capacity should be set based on the battery pack specification. Design Capacity is set in both mAh and cWh. The [CapM] setting in BatteryMode() determines which units are used to report data for capacity parameters. It is recommended to initialize Learned Full Charge Capacity to the same value as Design Capacity mAh. This will improve the accuracy after a reset until a learning cycle has completed to update the Full Charge Capacity.

Setting the CEDV Gauging Parameters

Setting the [EDV_CMP] bit to '1' in CEDV Gauging Configuration puts the gauge in CEDV mode. This selection will compensate both temperature and load. The Compensated End-of-Discharge (CEDV) gas gauging algorithm requires seven coefficients to enable accurate gas gauging. The default values are generic for Li-CoO2 chemistry, but these coefficients should be recalculated and updated based on the battery. The CEDV coefficients ensure gauge accuracy over temperature and current load. The procedure to gather the required data and generate the coefficients can be found at http://www.ti.com/tool/GPCCEDV.

The GPC tool requires six data log files of a continuous discharge (3 different temperatures and 2 different discharge rates). The logs should contain columns for time (in seconds elapsed), voltage (in mV), current (in mA where discharge current is negative), and cell temperature (in degrees C). Data files should be stored in csv format.

There are seven CEDV parameters in the report that should be programmed into the bq78350 Data Memory:

• EMF (EMF)

• EDVC0 (C0)

• EDVC1 (C1)

• EDVR1 (R1)

• EDVR0 (R0)

• EDVT0 (T0)

• EDVTC (TC)

A simpler but less accurate way to set up the gas gauging feature is to use EDV gauging. Setting the [EDV_CMP] bit to '0' in CEDV Gauging Configuration puts the gauge in EDV mode.

EDV mode uses fixed values for EDV1 and EDV2 that will not be compensated for temperature or current load. These values can be selected by one of the below approaches:

Rough Estimation

– i. Find the discharge curves for the cell from the manufacturer

– ii. Pick the capacity at the rate you intend to discharge and the voltage where you intend to stop. This is the design capacity and the EDV0.

– iii. Calculate 97% of the capacity and find the voltage on the curve. This is EDV1

– iv. Calculate 93% of the capacity and find the voltage on the curve. This is EDV2

Better Calculation

– i. Set the battery in the normal operating conditions and let the temperature stabilize.

– Charge the pack to full. Make sure overcharge protections and voltage protections are set to allow full range while learning the battery.

– iii. Log discharge until you reach the “empty” voltage. If using average cell voltage gauging allow some margin for the CUV threshold in setup so you can reach empty.

– iv. Calculate discharged capacity by adding the current x incremented time at each log point.

– v. Select EDV0 voltage where the pack is “empty”, either the manufacturer stated cutoff voltage or some selected value with margin

– vi. Calculate 97% of the passed charge to the EDV0 point, find the corresponding voltage in the log. This is EDV1.

– vii. Calculate 93% of the passed charge to the EDV0 point, find the corresponding voltage in the log. This is EDV2.

After EDV1 and EDV2 values are determined, load the values to Data Memory and run confirmation cycles as needed to verify performance.

Best regards,

Matt

Hi Howard,

Matt is on vacation this week, so his response will be delayed.  I have not worked with this device in detail, but I can try to offer a few comments based on my understanding.

While in CEDV mode, the EDV0/1/2 levels are generally recalculated regularly during operation, I believe this is every 1-sec.  If you have a log file of a discharge and note the EDV levels reported, you should see them changing as the load and temperature change.

However, the EDV voltage recalculations can be overridden, with them forced to stay fixed, depending on settings within the device.  The [FIXED_EDV0] bit forces the EDV0 voltage to not be modified, and the [EDV_CMP] bit can disable all dynamic calculations for the EDV voltages, even if in CEDV mode.

The jump that occurs at 7% is correcting errors that have accumulated earlier in the discharge.  While it appears the capacity slowly reduced to 17% before the jump, actually the capacity had really been slowly reducing to ~7%.  The device corrects the error when it reaches EDV2.  Have you completed a learning cycle on the battery?  If not, then it is not unexpected to have a jump, since your Full Charge Capacity is probably incorrect, it needs the learning cycle to learn the capacity of this specific pack.

You can also use the CEDV Smoothing feature, which rather than the SOC dropping immediately to 7%, it smoothly ramps the SOC downward as it nears this voltage.

CEDV gauging can be used in systems which incur voltage jumps, you may just need to tune the smoothing feature for your specific case.  Per your question #3, even if the voltage was decreasing very slowly and smoothly, if the gauge reaches EDV2 and the reported SOC is above the 7%, it will still need to correct the error, either through a jump or a smoothed transition.

Thanks,

Terry

Hi Howard,

The coulomb counter in the bq4050 can detect currents even below the reported LSB of the measured current, which I recall is around 3.5uV, although I don't have a firm value for the minimum.  Note the device also estimates self-discharge of the cells and includes this in the coulomb count, see the Self Discharge Rate parameter.

Regarding the power in SLEEP mode, you can extend the Sleep:Voltage Time and Sleep:Current Time parameters longer, so the device wakes less often to do measurements.  You may also consider whether you can instead use SHUTDOWN mode, which will have much lower current.

Thanks,

Terry