BQ27520-G4: A near-empty battery has an FCC value of 0

1: Terminate Voltage should be below the knee of the voltage curve. For most chemistries, 3000mV is fine.

2: The gauge uses Terminate Voltage for discharge simulations (capacity predictions). It uses Final Voltage to force RM and SOC to 0mAh and 0%.

3: FCC = 0% is likely related to something other than Terminate Voltage. Please check Ra0. If this is excessively high, set Ra0 and Ra1 to the value of Ra2. Some ChemIDs use an artificially high Ra0 and that can interfere if charge termination is close to DOD = 0 (Ra0).

4: The optimal values depends on the chemistry and also the system requirements. 3000mV works well for a lot of chemistries. What's the chemistry that you use?

Hi Dominik,
I appreciate for your response. 

3.I tried changing Ra Base(Ra0?) and Ra1, but FCC was 0.
Added result to BQ27520-G4_20230209_001.xlsx sheet: log,gg

Please check how to change Ra Base and Ra1.
I changed the Ra Base and Ra1 values in bqStudio's Data memory, but isn't it wrong?
For example, is it necessary to enable IT ENABLE when changing the value of Ra?

4. Battery is UR1865ZM. I found 2822 from the Chemistry ID list in bqStudio.
The Discharge End Voltage of the cell is 2.5V.
Battery pack overdischarge detection voltage of protection IC is 2.60±80mV.

Best Regards,
Kazunori

BQ27520-G4_20230209_001.xlsx

#3: This gauge stores Ra with relative values (from base). Therefore changing Ra0 and Ra1 isn't as simple as setting Ra base to 0. You'd have to calculate the delta values for Ra2 correctly to keep the integrity of the rest of the table intact. However, as a first test, your experiment is still valid. It shows that the problem isn't related to Ra0 and Ra1.

#4: The protection setting doesn't directly affect gauging. If the protector cuts off at around 2600mV +/-80mV, then I'd set Terminate Voltage higher than 2680mV.

About the FCC issue, based on your log file: The only suspicious item is that WAIT_ID is always set. This will clear, if the gauge identified the battery (and selected the correct profile). It looks like the gauge isn't able to complete this. If you don't use the battery identification feature, I recommend disabling IDSELEN in OpConfig (change OpConfig from 0x0973 to 0x0933).

Hi Dominik,
I appreciate for your response. 

#4 Is 3000mV the optimal value for ChemID2822's chemical properties?

About the FCC issue, based on your log file: 
I would like to use the battery identification function for a product that requires battery replacement.

Regarding the reason why WAIT_ID is not cleared:
Could it be that there is a problem with the learning cycle that prevents me from choosing the correct profile?
I am attaching the results of the learning cycle.
BQ27520-G4_20230214_001.xlsx

Please advise if there is a problem with the learning cycle.

Best Regards,
Kazunori

BQ27520-G4_20230214_001.xlsx

ChemID 2822 ends at 3000mV (OCV 3000mV -> DOD = 1.0). Terminate Voltage refers to the minimum loaded voltage (not OCV) for a specific application. This ChemID supports 3000mV Terminate Voltage because loaded voltage is always less than OCV.

The learning cycle looks fine.

Please check, if disabling pack identification makes a difference with your FCC = 0 test.

Hi Dominik,
I appreciate for your response. 

About Terminate Voltage:
From the SLUA903 and STEP-BY-STEP GUIDE and cell datasheets,
Discharge End Voltage: I thought it should be close to 2500mV, am I wrong?
BQ27520-G4_20230217_001.xlsx 　sheet: datacheet

SLUA903 has the following description.
”For learning cycle purposes, this should be set to the minimum voltage of
the battery as specified in the manufacturer’s data sheet."

Also, in the STEP-BY-STEP GUIDE below, Termination Voltage is the discharge Cut-off Voltage in the datasheet.
training.ti.com/how-perform-successful-learning-cycle-gauges

I can't find the optimal value for ChemID.
1. Do I have to ask TI every time I use a new cell?
2. Is 3.0V desirable for the Terminate Voltage when performing a learn cycle?
　Is it the reason why learning was not completed during relaxation after discharge in the learning cycle?

I disabled IDSELEN in OpConfig but FCC was 0.
BQ27520-G4_20230217_001.xlsx　　sheet: log,gg

Best Regards,
Kazunori

BQ27520-G4_20230217_001.xlsx

Terminate Voltage is a system (application) specific parameter. It's not determined solely by the chemistry. It applies to the voltage of the battery under load and is typically set to a value required by the system (and one that is safe for the battery).

For a learning cycle, Terminate Voltage isn't actually relevant in the sense that it controls gauging results like FCC and RM and SOC. Hence the suggestions from the app note are a bit misleading. For a learning cycle, the cycle itself should terminate as close to the minimum allowed voltage from the cell/battery datasheet, regardless of Terminate Voltage. This will ensure most complete coverage for cell resistance measurements (and good DOD change for Qmax).

About ChemID: Yes, this must be determined for every make/model cell. It is the same for a specific make/model cell but every time you change the cell manufacturer (make) and/or the cell model, you'll have to determine a compatible ChemID. This can be done by the customer directly (GPCCHEM on the TI site) or by requesting a custom ChemID via your local TI contact person.

Hi Dominik,
I appreciate for your response. 

About the Terminate Voltage during the learn cycle:
I thought it would have to be below the Terminate Voltage I set during the learn cycle, but now I understand that it doesn't matter.

About FCC =０ test:
I disabled IDSELEN in OpConfig but FCC was 0.
What about the results above?

Best Regards,
Kazunori

If IDSELEN doesn't make a difference, then we know that the problem isn't with the gauge fetching an incorrect Ra table. In general, FCC = 0 indicates that the gauge calculated an excessive voltage drop for full charge conditions and that is sometimes caused by an incorrect cell resistance table (esp. at grid point 0 and 1). If the gauge calculates a voltage drop that results in a simulated voltage that is below terminate voltage for a full charge, then FCC will be zero. I don't see anything else in the supplied data that causes this. You can try configuring load prediction for a user rate and use low load (e.g. C/10) for a user rate. If this fixes FCC = 0 and the Ra table is correct, then chances are that the load prediction used an excessively high load.

Hi Dominik,
I appreciate for your response. 

About FCC =０ test:
Load select = 6, User Rate = -235mA, but the FCC was 0.
BQ27520-G4_20230222_001.xlsx　　sheet: log,gg
What about the results above?

Best Regards,
Kazunori

BQ27520-G4_20230222_001.xlsx

This looks like a bug in the gauge FW where if you use an over-discharged cell (with an OCV < 3000mV), the gauge will zero out FCC until it calculates DOD <1.0. That's why this is ChemID dependent.

Regarding the FW bug,
1.Can you explain why the FCC is normal even if the battery voltage is 2500mV when the Terminate Voltage is set to 3000mV?
2.Please let me know the proper way to work around this bug.

Best Regards,
Kazunori

The gauge calculates FCC by taking an OCV measurement to establish "QStart" (which is QMax*(DOD(OCV) - DODatEOC)) plus remaining capacity (which is a result of a discharge simulation).

#1: I can't say why the gauge reports "normal" FCC in this case but 0mAh when Terminate Voltage is smaller. It looks like the gauge will zero out FCC (bug) if the discharge simulation terminates immediately if OCV < Terminate Voltage.

#2: 2500mV OCV isn't covered by this ChemID (it can't calculate DOD(2500mV) because DOD = 1.0 for 3000mV). A possible workaround is to ignore FCC and SOC until voltage recovered into the DOD range of the ChemID (=FCC is not 0mAh). In your host uC driver, set SOC = 0% in case FCC = 0mAh (the gauge reports 100% because it divides RM by FCC, which is a division by 0 and the gauge FW then saturates SOC at 100% as a result).

#1: Tell me more.
#1-1　What can be the reason why the discharge simulation ends immediately?
#1-2 If the Terminate Voltage is set to 3000mV, is it possible that the "normal" FCC will be calculated because the discharge simulation will not end immediately?


#2: Is the Terminate Voltage setting 2750mV?
As a trial, I obtained charge/discharge characteristics at Terminate Voltage=2750mV.
BQ27520-G4_20230228_001.xlsx

#2-1 Even if it exceeds 3000mV (DOD=1.0), FCC is the same value as RM and SOC is 100%.
Are there any other workarounds?



#3 From the datasheet of UR1865ZM, EndVoltage is 2500mV.
Why does ChemID2822 only cover up to 3000mV?

ChemID=2822 is the ChemID of UR1865ZM as described in Chem.ini below.

[1921_Chem]
ChemID=2822
Description=LiMn2O4 (Co,Ni)/carbon, 4.16V
NumKnownCells=1
HF_Data=0
2OCV_Data=1
1=Panasonic


#4 Even if I request a custom ChemID for UR1865Z, will it only cover up to 3000mV?

#5 Is it possible to improve by increasing the coverage of ChemID from 3000mV to 2700mV?

#6 This issue may also affect other products.
Please provide the voltages covered by the following ChemID.

ChemID
0129
2797
3608
2039
2050
1280
1146
2026
2678
2033

Best Regards,
Kazunori

BQ27520-G4_20230228_001.xlsx

#1-1: If the gauge calculates a loaded voltage that is less than Terminate Voltage, then it will return zero capacity.
#1-2: If you set Terminate Voltage to 3000mV, the simulation will terminate sooner so the chance that #1-1 happens is higher.

#2-1: If FCC = 0 and RM = 0, then the gauge calculates SOC = 100 * (0/0) = 100%. This is due to the division by zero.

#3: The original request from the customer who asked TI to characterize this cell was to characterize it for a charging voltage of 4160mV and a Teminate Voltage of 3000mV, hence this is the range for DOD = 0.0 to DOD = 1.0

#4: It's possible to re-characterize this cell for lower Terminate Voltage. However, the voltage drops off quickly so it may not make a big difference in usable capacity. It's probably more efficient to just not discharge the cell to below 3000mV.

#5: This will likely fix the observed behavior.

#6: DOD = 1.0 @ 0deg.C OCV:

0129: 2987mV
2797: 2702mV
3608: 2662mV
2039: 2509mV
2050: 2457mV
1280: 2962mV
1146: 2889mV
2026: 2727mV
2678: 2708mV
2033: 2847mV

#2-1：Why is FCC abnormal even if it exceeds 3000mv (dod = 1.0)?

From the previous answer, I thought that if it exceeded 3000mv (dod = 1.0), FCC would be a normal value (eg .2380mAh). Is it a mistake?
「A possible workaround is to ignore FCC and SOC until voltage recovered into the DOD range of the ChemID (=FCC is not 0mAh). 」

#7：For reference, please tell me the process of sending a battery to Ti for characteristic evaluation as in the following questions.

e2e.ti.com/.../bq27520-g4-what-is-the-process-of-sending-the-battery-to-ti-for-new-chemical-id-characterization

It is an additional question.
Please tell me the number of days you need to create a chemid.

Best Regards,
Kazunori

#2-1. FCC is supposed to be a normal value if DOD > 1.0. It looks like a bug that it isn't.

#7: Please contact your local TI FAE, who can ship the cells to Dallas for characterization.

#2-1　：You mean there is no workaround? Can the bug be fixed?

#6　：Please provide the voltages covered by the following ChemID.

ChemID
0100
2162
2176
2157
1120

#8　：Is there an application note that explains ChemID's coverage?

Best Regards,
Kazunori

0100: 2973mV
2162: 2450mV
2176: 3070mV
2157: 2987mV
1120: 2920mV

8: No.