BQ27Z561-R2: RSOC jump to 0% during low-temperature boot

RSOC becomes zero because the gauge estimates that there is no capacity left at 15deg.C. It is not about Term V Hold Time or the Term Voltage Delta (which is not added to Terminate Voltage but controls how the gauge schedules simulations and calculates cell resistance close to end of discharge; There's another parameter, Delta Voltage in class State, which is added to Terminate Voltage for *simulations*, not to test actual measured cell voltage).

Assuming that the ChemID has the correct temperature adjustment tables, the most likely reason for the sudden drop is how the gauge models self heating.

The bq27z561 tries to estimate how much the cell heats up during a discharge. This will affect the estimated voltage during a discharge and, especially at cold temperatures, can unlock a lot of capacity as the voltage curve is typically flat. This is especially difficult for the gauge to model well, if Terminate Voltage is set so that it ends up in the flat area, which 3300mV likely is for -15deg.C and a typical load.

The gauge learns how to estimate self hearting with state class parameters temp k and temp a. If these are learned at a different temperature (e.g. 25deg.C) then they may not be applicable to -15deg.C and lead to excessive heat loss with this extreme temperature difference in simulations, causing simulated voltage to drop significantly, more so than the actual cell voltage during a discharge.

The best way to fix this is to run GPCRb (GAUGEPARCAL) with the battery in the final enclosure in a temperature chamber set to e.g. -15deg.C with the airflow blocked from the device. GPCRb then calculates temp k and a and it's possible to then restrict the gauge to use these values with temp k and a min/max in class IT Cfg.

As a quick fix, increase the temp k value in the State class by a factor of 4 and set the temp k min/max parameters in the IT Cfg class to this value, +/- 10%. Note that temp k min is address 0x4622 and temp k max is at address 0x4620. Both are not visible in bqStudio. The actual value that you put in temp k in bqStudio is multiplied by 100. So if old Temp k is 0.88 [0.1C/256mW], then new Temp k = 3.52, Temp k min = 100 * 0.9 * 3.52 = 317 and Temp k max = 100 * 1.1 * 3.52 = 387.

Hi Dominik，

Dominik Hartl11 said:

RSOC becomes zero because the gauge estimates that there is no capacity left at 15deg.C.

Could you pls help to understand the principle? 

Regards，
Hailiang

Hi Dominik,

Customers get mass production files after using GPCRDB. If you directly write this file to the fuel gauge, and do a low-temperature start-up test (initial RSOC=45%), the RSOC will drop to 0%. If it is cycled at room temperature for 3-4 cycles and then tested at low temperature, the degree of drop will be reduced and will not drop to 0%. Why is this? Could you pls comment?

Regards,
Hailiang

The gauge will adjust the thermal model parameters when you cycle it in room temperature (and it will also update the cell resistance table). It's likely that the gauge estimates more accurate self heating when you run the low temperature test after 3-4 cycles at room temperature (esp. if GPCRb was done with forced cold air onto the cell, leading to conservative temperature model parameters; this assumes that they actually took the temp a/k values from GPCRb and programmed them to the gauge).

Hi Dominik Hartl11,  I have encountered the same problem. Can you tell me the detailed method for writing these two addresses 0x4622/0x4620 to modify the values?

Hi Lx Luo,

Dominik is out of office this whole week. Please expect some delay in his response.

Regards,
Jose Couso

You can use the method described in the TRM, 13.2.42 Data Flash Access() 0x4000–0x5FFF. Note that you also have to write the check-sum and length information as described in 13.2 0x00, 0x01 ManufacturerAccess() and 0x3E, 0x3F AltManufacturerAccess()

Hi Dominik, I conducted a comparative test:

1. Current project electricity meter model: bq27z561-R2, firmware number: 2_ 01；

2. Model of electricity meter for comparative testing: bq27z561-R1, firmware number: 0_ 04,；

3. The configuration parameters of the electricity meters tested by comparison are the same, and the same ChemID # 7153 is written. The tested data can be changed normally by FCC, RSOC does not jump, and can be discharged normally to the set cutoff voltage of 3.4V;

4. So it seems that there is a problem with BQ27Z561-R2?

Please confirm how we need to improve?

4571.log.rar

Are you ok attaching both .srec versions for R1 and R2 to this thread? There are several private parameters that are only visible to TI and I'd like to check, if there is a difference in these parameters between both versions. For example, if the temperature adjustment tables are different between your R1 and R2 configuration (this will not transfer automatically if you switch from R1 and R2), then you will see a difference in performance.

Please confirm the attachment, thank you!

8154.SREC.rar

Hello lx,

Thanks. I will check the files tomorrow

Make sure to program GPCRB after chemistry programming.