BQ27520-G4: Cannot finish learning cycle (Update status never become 01 after charge)

Hello Benjamin,

The chosen Chemical ID is not a problem since it only has 1.23 % error in DOD. What the gauge does is, it uses the voltage and resistance table associated with ChemID 2530 to perform gauging. This is fine since the voltage table and resistance table is very similar to your cell, Almost identical. 

Looking at the log file, I see that the impedance track algorithm running on the gauge is working as expected. What kind of chemistry are you working with? I suggest waiting 24 hours after charge to see if it makes a difference. But the VOK bit is getting cleared, this means the gauge learned a Qmax value. 

Please send a configuration file before and after learning cycle. Discharge the cell to empty, relax for 5 hours, charge to full, relax for 24 hours, discharge to empty, wait for 5 hours. Please send the gg file after all these steps. Please don't forget to save configuration file before starting this cycle.

Please change load mode to constant current since you are using a constant current to charge and discharge.

Regards,
Jose Couso

Hello Jose,

Thank you for your advice, I will re-do the learning cycle according to your advices.

But I still have a question.

According to you, the VOK bit was cleared, so the gauge should learned a Qmax value. But the Q max is still default value (1000) rather than 3200 ( the capacity after charging finished).

I looked at the record of a successful learning cycle before (for another battery), the VOK bit was cleared about 8 minutes after the charging ended, and the NomAvailCap, FullAvailCap and Qmax Cell 0 were updated after that.

But in this time, the VOK bit was cleared right at the ended of charging, and the NomAvailCap, FullAvailCap and Qmax Cell 0 were not updated.

Would you think this is the reson that cause Update Status 0/1 not change to 01?

How can I adjust the setting to prevent this from happen?

Thanks for your help.

Benjamin Kao

Hello Benjamin,

A good starting point before Qmax is learned will be to set Qmax cell 0 equal to the nominal capacity stated on cell's manufacturer datasheet.

Benjamin Kao said:

According to you, the VOK bit was cleared, so the gauge should learned a Qmax value. But the Q max is still default value (1000) rather than 3200 ( the capacity after charging finished).

Did you make sure to unseal the gauge? While VOK is set, the gauge is able to do Qmax updates if OCV measurement qualifies (if gauge understands that cell is well relaxed)

Benjamin Kao said:

I looked at the record of a successful learning cycle before (for another battery), the VOK bit was cleared about 8 minutes after the charging ended, and the NomAvailCap, FullAvailCap and Qmax Cell 0 were updated after that.

I suggest you follow this app note (section 4.2), it's for a different gauge, but same principal. Please ignore update status register mentioned here (all I will need is the before and after gg file).

https://www.ti.com/lit/an/slua777/slua777.pdf

Benjamin Kao said:

How can I adjust the setting to prevent this from happen?

I suggest to unseal the gauge, and set Qmax to nominal capacity as a starting point. Also, don't forget to change load mode to 0.

Regards,
Jose Couso

Hello Jose,

Jose Couso said:

Did you make sure to unseal the gauge?

Yes, I did unseal, IT_ENABLE, reset and unseal the gauge at the start of the learning cycle.

Jose Couso said:

I suggest you follow this app note (section 4.2)

I had read this app note, it seems to be the same steps as SLUA903 I followed for learning cycle.

I had re-do the learning cycle after set the load mode to 0, but my learning still not sucess.

The data memory .gg file before and after learning are below:

(Since the learning not success, the only difference in these two files is the cycle count)

7360.data_memory_before_learning.gg.csvdata_memory_after_learning_failed.gg.csv

And the learning record is attached:

learning_record_102.csv

Thanks for your help.

Benjamin Kao

Hello Jose,

Jose Couso said:

1- Taper Current > Chg Current Threshold > Quit current
2- Dsg Current Threshold > Quit Current

Thank you for this hint. 

In the failed learning, we set taper current = 30 mA, but other three threshold all larger than this value.

After set Dsg Current Threshold = Chg Current Threshold = 25 mA and Quit Current = 15 mA, the learning was success.

Jose Couso said:

I see some negative resistance values in the Ra-table. This is a sign of an unwell matched cell.

The .gg file I provided was not learned at all, so the negative Ra-table is directly from programming the Chem ID.

If the Ra-table should not be negative, why is this Chem ID in the database?

Thanks again for solving our problem.

Benjamin Kao