FOLLOW-UP Reported full charge capacity many times higher than design capacity of Li-ion battery using gas gauge Bq20z95

Hello Rasmus,

Update status 02 means that Qmax was updated

04 = IT algorithm and lifetime are updating

06 = Both of the above

Was update status set to 02 in the golden image? If yes, then check calibration. Wrong current calibration can make the gauge think the gauge is charging/discharging and count all the current that was passed.

Also is the battery put into storage right after the cells are connected or do you charge/discharge before putting into storage?

If Qmax was not 06 during the learning cycle, then the learning did not complete. You will need to redo learning in that case

Dear Shirash,

I checked with a functional battery from a different batch and different cells and there the updates status was set to 06, see attached parametersFunctionalBatteryOlder.csv

Should Qmax be a static value or will it update dynamically? Or does it depend on whether the update status is 04 or 06?

Sorry I don't have information regarding the golden image or the manufacturing yet. Will try to get this information.

Will update status be set to 06 after the learning cycle is complete and does that mean given our batteries having update status 04 that the learning cycle has failed?

Hello Rasmus,

Qmax will automatically update and that is when update status changes.

After successful learning update status will be 06, but it is then changed to 02 when creating the golden image. If on the learning cycle it ended with update status 04, then learning cycle failed.

Hi Shirish, 

So Qmax is only really updated during the learning process to create a golden image. Once the learning process is completed, Qmax is locked. The golden image along with the learned parameters (including Qmax) is then exported to all other batteries, correct? 

So it could be that our golden image (we don't know for sure because we do not have it currently) has not locked the Qmax value. Upon exporting it to the other batteries, the learning continues even after being sent to customer. That is why we get these strange Qmax values of -27622 mAh. Would that be a plausible hypothesis?

In order to solve this we need to redo the learning process and recreate the golden image, and then export the new golden image to the existing batteries so they get correct IT Cfg parameters, no?

Best Regards, Rasmus

Hello Rasmus,

Qmax will continue to update in normal operation but the maximum allowed change is limited.

A negative Qmax value usually indicates some corrupted values in data flash. Do you have a log file that shows Qmax being updated?

Hi Shirish,

Ok, so Qmax should update. But the reason I had the hypothesis that Qmax should remain static is that I have now compared GG-files of various batteries. The batteries with no problems have Update status 06 and a Qmax identical to the design capacity see here: okBatteryParameters.csv but the defect batteries with inflated capacity have Update status 04 and Qmax varying from -27000 to +30000 mAh. See here: notOkBatteryParameters1.csv and notOKBatteryParameters2.csv

I'm trying to figure out why and the big difference I see in a functioning battery and a defect one seem to be the Update status and whether the Qmax is different from design capacity.

I do unfortunately not have periodic logs of the Qmax values

Hello Rasmus,

It looks like the pack notOKBatteryParameters2.csv may be in permanent fail state. Both have negative value in Ra and Qmax. 

Check if the pack calibration is right. It must read 0mA current through SMBus when there is no load or charger.

Hi Shirish,

Thank you for your patience with me.

I have measured the current on rest with no load och charger, and it reads 0mA.

I also finally got the golden image: goldenImageParameters.csv of which all the defect batteries have been programmed. Can you help me check the parameters and check if something looks immediately wrong?

Hello Rasmus,

The most obvious problem is that Update Status = 0x04. It must be set to 0x02 in the golden image.

Based on this image, i would guess that learning did not complete successfully. The first step would be to get a successful learning and then create a new golden image file.