BQ27541-G1: Out of range FilteredCompensatedCapacity

Hi Faizal,

Sending the reset force a simulation recalculation of Full Charge Capacity, Remaining Capacity, etc. The incorrect reporting of state of charge came mainly from Full Charge Capacity being incorrect. State of Charge is Remaining Capacity / Full Charge Capacity.

What is your QMAX value reported? Full Charge Capacity can't exceed Qmax. I'm assuming your Data Flash is configured with SMOOTHEN = 1, so reported Full Charge Capacity is the Filtered Full Charge Capacity that's why your filteredCompensatedCapacityMah[24] = fullChargeCapacityMah[18].

The issue may be that you charged at a higher temp than the current temp and Qmax and FCC was based on the higher temp. The reset forced an update of FCC and RemCap.

The error in SOC seen would have been corrected by the gauge when conditions were met to trigger the next simulation.

Damian,

Thank you for your reply. Here is sample Qmax from another unit with this affliction.


nominalAvailableCapacityMah[12]:1668
fullAvailableCapacityMah[14]:2466
remainingCapacityMah[16]:1631
fullChargeCapacityMah[18]:7506
averageCurrentMa[20]:-35
timeToEmptyMinutes[22]:2796
filteredCompensatedCapacityMah[24]:7506


---subclass82_State---
Qmax Cell 0[0]:2549

In this case, FCC = 7506, Qmax = 2549. A hypothesis - Is this some kind of data corruption related to the way the filtering data recovery is implemented?

Our manufacturing is in controlled conditions (~25C always) . These devices never left manufacturing. So I do not suspect charging at actual elevated temperature. In addition, the temperature sensing seems accurate as well.

In any case, the error in FCC is ~300%! What explains this much error?

Best,
Faizal.

Hi Faizal,

In your application do you charge to full and discharge to empty or that never occurs? Also, what is your sense resistor value and what is the value of your cc deadband. Your issue might be either related to not having proper reference points by having charge to full or discharging to empty from time to time OR it might be due to ghost current from the coulomb counter. The resoluton to the later typically is to increase the value of your cc deadband so your gauge doesn't  accumulate ghost current. If  the former, is the culprit, you will have to either have full charge occurring or move to the bq27542 which is the next gen of the 541.

thanks

Onyx  

Full discharge cycles are not guaranteed to occur. Full charges are anticipated. Deadband = +/-5ma. Sense = 0.01ohms. I have verified that the average current measured (by BQ27541) tracks the actual current with better than 1ma offset. So, I believe, 5ma deadband seems adequate. In any case why would this number count above Qmax or design capacity.

Performing a full charge / discharge has not fixed the issue.

Can a tab disconnection cause this issue?

In addition, if 'ghost current' is suspect, why is it that only the filtered data is affected?

Hi Onyx,

    Thanks a lot for the detailed info and test steps. Since all units exhibiting this problem has been 'reset' I will have to wait until the problem re-appears. Also, since the IC is installed in the finished product, it will be tricky to get this configuration into the IC via bqStudio.

I will devise a way to do this. Perhaps you could tell me the offset and format for CC-Deadband field such that I can program the data-flash without having to un-power the fuel gauge or otherwise disassemble the system. 

I would like to note that the appearance of the large error in filtered values was rather abrupt and not as a result of long-term accumulation of error. The only suspicious activity between 'known good state' and 'bad state' was a tab disconnection. I have set aside the suspect units (now all 'reset') for long term observation to see if they act up on their own. 

Best,

Faizal

Damian,
Onyx,

Thanks for your responses!


What are the conditions, other than a reset command or a discharge below 'Terminate Voltage', under which the FG is expected to perform a simulation/internal operation that corrects this error.

I have observed that the following conditions does not correct the problem.

(a) Relaxation after Charge.
(b) Relaxation after Discharge.
(c) Relaxation (for several hours) after Charge Termination detection by taper current and voltage thresholds.





Best,
Faizal.

Onyx,

    Thank you for your response and recommendation. We are planning a move to bq27542-g1.  

     Based on observation, I am also considering setting SMOOTHEN = 0 for the bq27541s is stock. 

    Could you please point me to any TI documentation/errata which describes the problem? Or any specific change(s) in Bq27542-g1 which addresses this issue. 

Best,

Faizal.