BQ27426: Reason why different battery Fuelgauge get different FCC values on initial power up?

The gauge uses measured voltage, current and temperature after a SOFT_RESET (=after configuring the gauge after power on) to calculate FCC and RM and SOC. It doesn't know the age of the attached cell so these results are not based on the specific cell that you connected but on the configuration and these measurements.

Hello Dominik,

Thank you for your reply. We have studied the formula and have a few more questions:

1. If the same battery is inserted into the Fuelgauge IC at different Vbat voltages, will the initial FCC obtained be the same? What is the reason?

2. According to the formula in the figure below, after performing a full charge, DODatEOC, DOD0, QStart, and FCC will be re simulated and corrected to tend towards approaching the real battery. Is this understanding correct?

3. According to the formula in the figure below, after performing full discharge, DODfinal, RM, FCC, and Qmax will be re simulated and corrected to tend towards the real battery. Is this understanding correct?

4. Battery fully charged ->Relax for 5 hours, OCV is corrected, DODfinal, RM, FCC, Qmax will be re simulated and corrected to approach the real battery. Is this understanding correct?

5. If full charging is not performed, only charging to SOC55% at SOC51%, and then maintaining Relax, waiting for the battery to self deplete to SOC51% before charging to SOC55%, will FCC gradually correct to close to the true battery capacity when running for a long time in this scenario?

5-1. If not, please let me know which parameters have not been changed, causing the FCC to be unable to be corrected?

5-2. If Yes, please let me know which parameters will change so that FCC can correct them to be close to the actual battery capacity?

Best Regards！

Hello，

Please help reply to the above question, thank you!

Hello，

Additional information：Fuelgauge IC is a system side application.

Please help reply to the above question, thank you!

#1: FCC may change a little bit but it should be close between insertions. The gauge takes an OCV measurement, calculates DOD0 and then performs a discharge simulation. FCC is calculated with Initial Capacity and the simulation results, hence this is not predominantly dependent on DOD0.

#2: This formula is an approximation. There are other details in the algorithm (e.g. Initial Capacity instead of Qstart, which can be different in some cases).

#3: DOD final is not a function of actual discharge but of termination settings and load prediction and cell model.

#4: Qmax is not simulated but measured. Other than that the statement is fairly accurate.

#5: DOD0 may not be of good quality depending on chemistry. Qmax may not update (flat zone).

In general, the algorithm will adapt and improve FCC quality over time, assuming that the chemistry is compatible and that the conditions for Qmax and Ra updates are met. And if load prediction is configured correctly.

Hello Dominik，

Thanks for you reply.

Regarding # 5,

1. Can the Filtered FCC of FuelGauge match the actual capacity of the battery only when both Qmax and RA are corrected?

2. Can the Filtered FCC of Fuelgauge match the actual capacity of the battery with only Qmax or Ra correction?

3. What actions need to be performed to match the value calculated by Qmax with the battery?Full charge？Full discharge？

4. Qmax will be updated during quiescence or when OCV is stable (dV/dt<4uV/s). If only charging+Relax, PassedCharge will change, will Qmax be corrected? Will it gradually match the true capacity of the battery after long-term operation?

chengxiaobin said:

5. If full charging is not performed, only charging to SOC55% at SOC51%, and then maintaining Relax, waiting for the battery to self deplete to SOC51% before charging to SOC55%, will FCC gradually correct to close to the true battery capacity when running for a long time in this scenario?

5. The RA Table is divided into 15 DOD grid-points, and the SOC we use is approximately 51%~55%. Can we cover one DOD grid-point?

6. Only covering one DOD grid point, is RA Table corrected?

7. Our usage scenario was asked in another case, and the response at that time was: Filtered Fcc can be updated.

There is a discrepancy with your reply, which one is correct? What is the correct basis?

Best Regards！

#1: yes

#2: both Qmax and Ra must be correct. And load prediction must be configured correctly

#3: The gauge has a complex rule set to ensure the integrity of Qmax and Ra. You can run a learning cycle to let the gauge measure both with high quality. The gauge will update Qmax and Ra during regular use if conditions permit (they are too numerous to explain on E2E). See https://www.ti.com/video/5836376324001

#4: Qmax will update if DOD changed enough (either charge or discharge) and various other conditions are met (coulomb counter error thresholds, temperature range, voltage range etc.)

#5: Grid points are spaced by DOD, not SOC. DOD is purely a function of coulomb count, OCV and Qmax. SOC is the result of a complex algorithm. DOD must change by 0.0975 to guarantee a grid point crossing for DOD between 0.0 and 0.78 and by 0.037 for DOD between 0.78 and 1.0.

#6: The whole Ra table will by updated with one grid point crossing that results in a successful resistance measurement.

#7: I don't see the discrepancy. Jonny said that Filtered FCC will be updated during regular use, which is correct. And I wrote that "FCC may change a little bit but it should be close between insertions.", which is also correct.

Hello Dominik，

Thanks for you reply.

1. What is the specific value of "DOD changed enough"?

The temperature is between 1-40 ℃, the voltage is not in a flat area, the design capacity is 1230mAh, and the RM variation range is 629-680mAh. Will Qmax be corrected in this scenario?

Dominik Hartl11 said:

#4: Qmax will update if DOD changed enough (either charge or discharge) and various other conditions are met (coulomb counter error thresholds, temperature range, voltage range etc.)

2. Design capacity 1230mAh. What is the RM corresponding to a change of 0.0975 in DOD between 0.0 and 0.78? What is the RM corresponding to a 0.037 change in DOD between 0.78-1.0?

3. Can it be understood that in our application, Filtered FCC can gradually approach the true capacity of the battery?

Dominik Hartl11 said:

#7: I don't see the discrepancy. Jonny said that Filtered FCC will be updated during regular use, which is correct. And I wrote that "FCC may change a little bit but it should be close between insertions.", which is also correct.

4. Directly insert the aging battery and run it for a long time in our application. Can filtered FCC be gradually corrected to the capacity of aging batteries?Why？

5. Please confirm if the understanding of Qmax correction is correct?

·After 10 minutes of Relax, use OCV table lookup to obtain DOD, read OCV every 100 seconds, and iteratively update DOD until OCV is stable enough or exceeds 5 hours. Recorded as DOD1

·When the battery is discharged or charged, the PassedCharge changes. After entering Relax for 10 minutes again, continue to use OCV to check the table and obtain DOD, with an interval of 100 seconds. Iteratively update DOD until OCV is stable enough or exceeds 5 hours. Recorded as DOD2

·According to the formula in the figure below, PassedCharge, DOD1, and DOD2 will all change, so Qmax will be corrected.

6. Fuel Gauge system side application. Directly insert batteries with unknown aging levels. Can only identify the degree of battery aging through full charge and full discharge? Is there any other way?

7. The battery is fully charged, as the Passedcharge change exceeds 37% and the DOD change also covers the grid points. Therefore, during the Relax period, Qmax will be corrected to be close to the actual maximum capacity of the battery. Is this understanding correct?

8. The battery is fully discharged, and because the change in DOD covers the grid points, the RA Table will be corrected to be close to the true impedance of the battery. Is this understanding correct?

Best Regards！

#1: The gauge has many safe guards for Qmax updates. It will reject a Qmax update if the voltage is in an area of the OCV curve that's too flat or if temperature is too high or low. All of that can be configured so I can't give you specific numbers unless I have the exact configuration (.srec or .bq.fs).

#2: RM isn't the result of a simple formula but the result of a discharge simulation (the core of the algorithm). Therefore it's not possible to calculate RM as a function of change in DOD.

#3: The smoothing algorithm is designed to eventually let filtered FCC converge to true FCC. The reason why there's a difference to begin with is that FCC is a function of temperature and load (among others) so it isn't constant but it can change (drastically). To avoid jumps, the gauge has a complex smoothing algorithm (not a simple formula), which is intended to prevent sudden changes in reported values.

#4: The gauge will have to learn Ra and Qmax before it becomes accurate for an aged cell. This will take several charge,relax,discharge,relax cycles as the gauge imposes limits on how much Ra and Qmax are allowed to change. Eventually, the gauge will adapt and report correct values.

#5: The concept is correct but the times depend on configuration and chemistries and charge termination. It's not quite a simple as you described. 

#6: The gauge requires at least one learning cycle (a special mode where Qmax and Ra filters are disabled; usually not implemented in the field) or several cycles (see answer #4).

#7: Yes, if all other (many) Qmax update conditions are met.

#8: Yes.

A lot of your questions cannot be answered in detail on E2E as this would require an in-depth discussion about details of the algorithm (e.g. all Qmax update conditions). Some of those aren't disclosed without an NDA. Please contact your local TI representative (FAE) for further assistance (e.g. NDA).

Hello Dominik，

1、Uploaded our FS file. Please help confirm the previous # 1 question.8400.0418_New_Bat_FS_withChemID_0419.gm.zip

Dominik Hartl11 said:

#1: The gauge has many safe guards for Qmax updates. It will reject a Qmax update if the voltage is in an area of the OCV curve that's too flat or if temperature is too high or low. All of that can be configured so I can't give you specific numbers unless I have the exact configuration (.srec or .bq.fs).

2、My question is: Our application is running for a long time. Can Filerted FCC be corrected to be close to the true battery capacity? Can True FCC be understood as being close to the actual battery capacity?

Dominik Hartl11 said:

#3: The smoothing algorithm is designed to eventually let filtered FCC converge to true FCC. The reason why there's a difference to begin with is that FCC is a function of temperature and load (among others) so it isn't constant but it can change (drastically). To avoid jumps, the gauge has a complex smoothing algorithm (not a simple formula), which is intended to prevent sudden changes in reported values.

Best Regards！

#1: The gauge won't qualify an OCV measurement if it's between 3244mV and 3338mV and if temperature was outside of 10deg.C to 40deg.C for your configuration.

#2: Both filtered and true FCC will converge to true cell capacity for your charging and load characteristics. Filtered will stay constant during a discharge even if conditions change as this by design of the smoothing algorithm

Hello Dominik,

1. My understanding is that after obtaining a qualified OCV, the DOD obtained by looking up the table is more accurate.

Dominik Hartl11 said:

#1: The gauge won't qualify an OCV measurement if it's between 3244mV and 3338mV and if temperature was outside of 10deg.C to 40deg.C for your configuration.

2. My understanding is that, in combination with our application, when the battery temperature is between 10 ℃ and 40 ℃, with charging (SOC 51% ->55%)+Relax, qualified OCV can be obtained, thereby obtaining accurate DOD. Therefore, during long-term use, Filtered FCC will gradually match the actual battery capacity. Is this understanding correct?

Dominik Hartl11 said:

#2: Both filtered and true FCC will converge to true cell capacity for your charging and load characteristics. Filtered will stay constant during a discharge even if conditions change as this by design of the smoothing algorithm

3、 How to understand this description?

Dominik Hartl11 said:

Filtered will stay constant during a discharge even if conditions change as this by design of the smoothing algorithm

4、Is it necessary to set update status=0x03? If only full charge and full discharge are performed, with update status=0x00, will Filtered FCC gradually match the actual battery capacity during use in our scenario?

Wang Dong said:

6. Fuel Gauge system side application. Directly insert batteries with unknown aging levels. Can only identify the degree of battery aging through full charge and full discharge? Is there any other way?

Dominik Hartl11 said:

#6: The gauge requires at least one learning cycle (a special mode where Qmax and Ra filters are disabled; usually not implemented in the field) or several cycles (see answer #4).

5. What is the measurement error of Filtered Fcc?

6. Directly inserting an aging battery requires at least one learning cycle. So, battery SOH≥？%, Does not require learning cycle?

For example, if SOH>80%, there is no need to perform a learning cycle, and during use, the filtered FCC can gradually match the actual battery capacity.

Wang Dong said:

4. Directly insert the aging battery and run it for a long time in our application. Can filtered FCC be gradually corrected to the capacity of aging batteries?Why？

Dominik Hartl11 said:

#4: The gauge will have to learn Ra and Qmax before it becomes accurate for an aged cell. This will take several charge,relax,discharge,relax cycles as the gauge imposes limits on how much Ra and Qmax are allowed to change. Eventually, the gauge will adapt and report correct values.

7. Is there only one method, which is to execute the learning cycle, to match the true capacity of the aged battery inserted into the Filtered Fcc of FuelGaugeIC?

Best Regards！

#1,2: Yes.

#3: The smoothing algorithm holds filtered FCC steady during discharge. This is by design.

#4: Update Status = 03 means that the gauge won't apply some filters for Qmax and Ra. This should be set to 03 before running a dedicated learning cycle (not regular use).

#5: Filtered FCC is not a true FCC. If you define the error as the difference between true FCC and filtered FCC then this can be large as filtered FCC, by design, is held constant during a discharge with the objective to avoid jumps in SOC due to changes in discharge conditions (load, temperature). The smoothing algorithm is too complex to fully explain on E2E and it also requires an NDA.

#6: The gauge will always require learning Qmax and Ra to adjust to an aged cell. There is no way around this. SOH will only be meaningful after the gauge learned Qmax and Ra for an aged cell.

#7: You can also let it run regular discharge and charge cycles and the gauge will eventually learn Qmax and Ra for an aged cell. But it will take several cycles (the exact number depends on how aged the cell is). A true learning cycle (with Update Status = 03) will accomplish this in one cycle.

Hello Dominik，

Thanks for you support.

1. What is the measurement error of True FCC?

2. Because the actual verification of Fuel The matching result between Flt Fcc and aging battery capacity takes too long, and feasibility needs to be confirmed theoretically. Therefore, please allow me to describe our usage scenario again:

    Usage scenario: Reserve capacity 350mAh

    If Fuel RM<280mAh, then charging to Fuel RM ≥ 320mAh. Then maintain the Relax until the battery self drains to Fuel RM<280mAh, Charge to Fuel again RM ≥ 320mAh. Long term cycle.

    In our application, the battery is only used as a backup power source, and the discharge scenario can be ignored. Because batteries have self loss, they will be recharged after the self loss reaches a certain value.

    2-1. In this scenario, inserting an aging battery without performing a true learning cycle, long-term operation, Can Fuel gauge's Flt FCC gradually match the actual battery capacity??

    2-2.In this scenario, use a battery that matches the FS file and run it for a long time, Can Fuel gauge's Flt FCC gradually match the actual battery capacity??

3. The SOH here is the actual value of the battery measured by the instrument.

What I want to ask is:

When the actual SOH of the battery is more than ?%, Can skip the learning cycle?

When the actual SOH of the battery is less than ?%, Need to execute a learning cycle?

4. There are two situations that can cause battery aging: one is that multiple charging and discharging lead to an increase in battery impedance(Cyclic aging); Another approach is: prolonged exposure to high temperatures can lead to a decrease in lithium electrons(Calendar aging).

Regarding the aging caused by the reduction of lithium electronics, can Fuelgauge IC only match the true capacity of the battery through full charge and full discharge?

5、During discharge, the Flt Fcc of fuelgauge does not change. During charging or relax, the Flt Fcc of the Fuelgauge will converge to the true battery capacity. Is this understanding correct?

Dominik Hartl11 said:

#3: The smoothing algorithm holds filtered FCC steady during discharge. This is by design.

Best Regards！

#1: This depends on many factors. With optimal configuration and load prediction, impedance tracking can achieve an error of 1 percentage point True SOC - Reported SOC.

#2-1: There is no way around performing charge/relax/discharge/relax cycles to let the gauge learn true FCC for an aged cell. If you can't run a true learning cycle, the gauge will still adapt if you run shallow cycles that pass enough charge for Qmax and Ra updates (and don't stop in the flat zone of the OCV; other conditions apply).

#2-2: As long as the system cycles deeply enough, the gauge will learn true FCC.

#3: SOH is ultimately a function of cell resistance and Qmax measurements. SOH is meaningless until the gauge learned Qmax and Ra, hence it's not a good idea to use SOH to decide if a learning cycle is necessary.

#4: You don't need to perform full cycles. The reason why TI distinguishes between learning cycles and regular use cycles is that regular use cycles are typically not well controlled (e.g. current may be dynamic, total passed charge depends on the user, temperature is not controlled etc.) hence the gauge applies heavy filters on Qmax and Ra in regular use. This is to prevent gauging errors due to spurious Ra and Qmax errors - it's better to reject and filter measurements than corrupt Qmax and Ra. A learning cycle disables these filters and some rules because the discharge is deep and the discharge is controlled (load, temperature), hence the measurements are of high quality.

The gauge will still learn Qmax and Ra (FCC etc. will improve) during regular use - but it won't allow for fast changes so if you replace the battery with a severely aged battery, it will take several shallow cycles until the filters allow Qmax and Ra to converge.

#5: There are several rules how true and filtered values sync up. I can't list all of them on E2E. It also depends on settings (e.g. RelaxJumpOK bit). In general, filtered FCC will adjust at synchronization events (full charge, full discharge, OCV measurement in relax if RelaxJumpOK = 1).

Hello Dominik，

Thank you for your support。

Using SOH description is indeed inappropriate. Change the description again.

Use an instrument to test the capacity of the battery from full charge to full discharge, noted as BubtFCC.

The FuelGauge recognizes the true capacity of an aging battery by performing a minimum of one learning cycle. So want to confirm.
1、When the battery capacity BubFCC is lower than how many mAh, need to perform a learning cycle?
2、When the BubFCC is greater than how many mAh, can we use the Fs file of the new battery directly without performing a learning cycle?

Dominik Hartl11 said:

#3: SOH is ultimately a function of cell resistance and Qmax measurements. SOH is meaningless until the gauge learned Qmax and Ra, hence it's not a good idea to use SOH to decide if a learning cycle is necessary.

3、After powering on the device and performing charging, two different phenomena were found. Could you please help analyze the reasons?

Fuel_Volt-->From FuelgaugeIC，Voltage(): 0x04 and 0x05

Fuel_Curr-->From FuelgaugeIC，AverageCurrent(): 0x10 and 0x11

Fuel_RM-->From FuelgaugeIC，RemainingCapacity(): 0x0C and 0x0D

Fuel_Fcc-->From FuelgaugeIC，FullChargeCapacity(): 0x0E and 0x0F

Bub Ntc-->Battery temperature

Note：

1、No significant change in Filtered FCC during Relax and Discharge of battery.

2、The real capacity of the battery used is close to 1230mAh.

Best Regards！

Hello Dominik，

Thank you for your support。

Using SOH description is indeed inappropriate. Change the description again.

Use an instrument to test the capacity of the battery from full charge to full discharge, noted as BubtFCC.

The FuelGauge recognizes the true capacity of an aging battery by performing a minimum of one learning cycle. So want to confirm.
1、When the battery capacity BubFCC is lower than how many mAh, need to perform a learning cycle?
2、When the BubFCC is greater than how many mAh, can we use the Fs file of the new battery directly without performing a learning cycle?

Dominik Hartl11 said:

#3: SOH is ultimately a function of cell resistance and Qmax measurements. SOH is meaningless until the gauge learned Qmax and Ra, hence it's not a good idea to use SOH to decide if a learning cycle is necessary.

3、After powering on the device and performing charging, two different phenomena were found. Could you please help analyze the reasons?

Fuel_Volt-->From FuelgaugeIC，Voltage(): 0x04 and 0x05

Fuel_Curr-->From FuelgaugeIC，AverageCurrent(): 0x10 and 0x11

Fuel_RM-->From FuelgaugeIC，RemainingCapacity(): 0x0C and 0x0D

Fuel_Fcc-->From FuelgaugeIC，FullChargeCapacity(): 0x0E and 0x0F

Bub Ntc-->Battery temperature

Note：

1、No significant change in Filtered FCC during Relax and Discharge of battery.

2、The real capacity of the battery used is close to 1230mAh.

Best Regards！

About phenomenon 1 and 2: The gauge will recalculate DOD at end of charge using measured VatChgTerm and configured Taper Rate at start of charge. This is by design, hence FCC will update if VatChgTerm was changed (e.g. when writing a .gm.fs file after power on)

FCC is a function of load, temperature, Qmax, Ra, charge termination, terminate voltage and delta voltage. It's not the same as "real capacity of the battery" as given by the datasheet of the battery.

Hello Dominik，

Sorry, didn't understand your reply.

The rated capacity of the battery is 1230mAh, and the Design Capacity of the FS file is also 1230mAH.

1、When the battery capacity is 1230±?? mAh, the FS file can be used directly and the Filtered FCC initially acquired by Fuelgauge matches the battery?

2、When the inserted battery capacity is ?? mAh, need to perform the learning cycle?

3, What are the steps to perform a learning cycle? update status=0x03, charge over 90% + full discharge? Or do I need to redo the gold learning?

Best Regards！

#1: The gauge will use what is configured in the .gm.fs file. Design Capacity in the FS file is only one component. The gauge also uses Qmax, Ra, charge termination and end of discharge settings (and temperature) to calculate FCC. Also, it doesn't matter what cell you connect, the gauge will use the info from the configuration only to calculate FCC. If you connect a compatible cell, then the results will be accurate. The gauge will then adjust Qmax and Ra (and other parameters) as you charge/discharge the cell. If you connect an incompatible cell (e.g. a different chemistry/capacity or a severely aged or damaged cell), then the gauge will not be accurate. It will become more accurate over time for a aged cell (same chemistry) but that isn't instantaneous.

#2: A learning cycle is the fastest method but the gauge will adjust automatically during regular use as well.

#3: A learning cycle is started by setting Update Status = 0x03. Then you have to perform a successful Qmax update followed by a discharge that crosses most Ra grid points.

Hello Dominik，

1、#1 and #2: Can't you provide specific mAh values?

In addition.

The rated capacity of the battery is 1230mAh, and the Design Capacity of the FS file is also 1230mAH。

The FS file is made based on the new battery(1230mAh),Reserve capacity 350mAh

Charging Scenarios:The battery has a charging capacity of about 50mAh, a charging current of 390mA, and a charging time of about 10 minutes

Discharge Scenarios:The battery discharge capacity is about 50mAh, the discharge current is about 45uA (0-28 ℃, battery self loss), and the discharge time is about 46 days

2、Does our charging and discharging scenario belong to normal battery charging and discharging?

3、With our charging and discharging scenarios, can FuelgaugeIC gradually match the real battery capacity?

Best Regards！

Hi Wang,

1- Can you please be more specific? The gauge will calculate capacity based on what Dominik explained in #1

2- For the gauge to learn Qmax, it needs a change in capacity by more than 90% for the first time. It looks like 50mAh does not meet this condition. 

3- I suggest that for the learning cycle, you charge/discharge the battery from full to empty and vice versa. Then, you can adjust a private parameter called "Min % Passed Chg for Qm" so that the 50mAh charge and discharge condition does not affect gauging. This way the gauge will learn the true capacity of the cell as it ages. 


Regards,
Jose Couso

Hello，

The rated capacity of the battery is 1230mAh, and the Design Capacity of the FS file is also 1230mAH。

The FS file is made based on the new battery(1230mAh),Reserve capacity 350mAh

1、When the inserted battery capacity is 1230±?? mAh, the FS file can be used directly and the Filtered FCC initially acquired by Fuelgauge matches the battery?

2、When the inserted battery capacity is ?? mAh, need to perform the learning cycle?

Jose Couso said:

1- Can you please be more specific?

Best Regards！

Hello jose，

Thanks for you reply。

Your response is slightly different from my question. I would like to ask "When the inserted battery capacity is 1230 ±?? mAh" and "When the inserted battery capacity is?? mAh". Please help me reply again, thank you!

Wang Dong said:

1、When the inserted battery capacity is 1230±?? mAh, the FS file can be used directly and the Filtered FCC initially acquired by Fuelgauge matches the battery?

2、When the inserted battery capacity is ?? mAh, need to perform the learning cycle?

Best Regards！

Hello jose，

Thanks for you reply。

1、Your response is slightly different from my question. I would like to ask "When the inserted battery capacity is 1230 ±?? mAh" and "When the inserted battery capacity is?? mAh". Please help me reply again, thank you!

Wang Dong said:

1、When the inserted battery capacity is 1230±?? mAh, the FS file can be used directly and the Filtered FCC initially acquired by Fuelgauge matches the battery?

2、When the inserted battery capacity is ?? mAh, need to perform the learning cycle?

2、"Min% Passed Chg for Qm" was not found in TRM. What does this parameter mean? What is the modifiable range? How to modify this parameter?

Jose Couso said:

3- I suggest that for the learning cycle, you charge/discharge the battery from full to empty and vice versa. Then, you can adjust a private parameter called "Min % Passed Chg for Qm" so that the 50mAh charge and discharge condition does not affect gauging. This way the gauge will learn the true capacity of the cell as it ages. 

Best Regards！

Hello Wang,

Our responses may be delayed due to the US holidays.

Sincerely,

Wyatt Keller

#1: The gauge will use the parameters from the configuration file to calculate FCC. If you insert a battery with different capacity than what the configuration file describes, assuming that the configuration is compatible with your cell, the gauge will need either one successful true learning cycle or several in-system charge/relax/discharge/relax cycles to report actual FCC.

#2: This parameter controls how much you have to charge or discharge between qualifying OCV measurements for the gauge to calculate Qmax. Qmax = passed charge / change in DOD. As Qmax is of utmost importance (and Qmax changes slowly), the gauge has many safe guards to ensure correct Qmax and the minimum passed charge is one of those safe guards.

Hello，

1、How to change the parameter "Min% Passed Chg for Qm"?

2、Our charging capacity is only 50mAh, can it be modified to match our application?

3、If it can be changed, what will be the impact?

7206.0418_New_Bat_FS_withChemID_0419.gm.zip

#1: TI doesn't expose this parameter. It must only be changed after careful analysis of the system as changing this parameter can lead to inaccurate Qmax, which will cause a lot of problems.

#2: I recommend using a higher sense resistor and scaling capacities by a factor of 10. The gauge uses integer math internally for capacity predictions and with a 50mAh cell, it'll likely run into accuracy issues due to integer resolution. Also, currents are going to be very small so you'll run into accuracy issues with the coulomb counter. See https://www.ti.com/lit/pdf/slua792.

#3: The impact is likely reduced Qmax accuracy, which will break gauging.

Hi dominik，

Our product has been mass-produced, and we can't change the IC anymore. so:

1、Please help us to change "Min% Passed Chg for Qm" to 50mAh based on our FS file, thanks!
2、Please tell us the detailed effect on FuelGaugeIC prediction after changing "Min% Passed Chg for Qm" parameter.

3、Please let us know the modification range of "Min% Passed Chg for Qm" parameter and how to modify it.

8875.0418_New_Bat_FS_withChemID_0419.gm.zip

I will handle this internally through the assigned TI contact. There's an internal email thread already where I explained the concerns about opening up the Qmax parameters. As this is a custom modification, I cannot attach this to a public E2E thread.

Hi dominik，

Thanks for you reply!

1. What is the configurable range of the "Min% Passed Chg for Qm" parameter?
2. What are the other parameters associated with this parameter? In what proportion should they be modified?

The range is 0% to 100%.

There are other parameters that TI doesn't publish which control Qmax updates. For example, a threshold for the estimated error due to accumulated coulomb counter errors.

I'm still actively working with your local TI contact on resolving this request. In general, we are very careful changing some of these parameters because of the potential to negatively impact gauging accuracy. In this case it seems that the motivation for the change is to get a Qmax update. While it is possible to get a Qmax update by reducing the min. passed charge requirement, it is not advised as this has a big potential to make things worse.

Actual chemical capacity of a cell doesn't change quickly. So it's better for the gauge to only update Qmax if conditions are good. And that includes a significant passed charge.

If the request is for a Qmax update after changing the battery, then it's even more important to allow for a large passed charge. For a first Qmax update, our recommended passed charge in 90%, not 37% of Qmax.

The correct approach is for your application to allow for a significant discharge, not for TI to change a critical threshold parameter.

Hello Dominik,

The risks we understand. We are just doing the verification first. Please help us with the FS files for the three different values of "Min% Passed Chg for Qm"=4%, 6% and 10%. Thank you!

Hey Wang,

Today's a federal holiday so no one is in the office. We will get back to you tomorrow.

Regards,
Nick Richards

Please attach your latest .gm.fs file and I'll provide three files with the changes.

Hello Dominik，

Thank you very much for your support!

We compared the 6% and 10% FS files and found that two subclassIDs were modified.

1、Can you provide the description of the modified values of the subclass ID?

2、Can you provide the method to modify "Min% Passed Chg for Qm"? Purpose: We may need to modify this parameter several times in the future to find the maximum value suitable for our application.

3、We read from other materials that modifying "Min% Passed Chg for Qm" will affect the accuracy of SOC, but we don't know the correlation between this parameter modification and SOC, please help to explain?

#1: The changes are both to the min.passed charge parameters and to track version info and Qmax error check limits. These are all private parameters.

#2: This will require a waiver in case performance suffers due to extreme changes to Qmax min. passed charge. The local FAE is looped in. Please work with the FAE for additional information about this private parameter.

#3: The accuracy of SOC depends on the accuracy of Qmax because Qmax determines the accuracy of DOD and DOD determines the accuracy of the cell model's voltage prediction. Qmax is critical, hence the gauge is designed to be very conservative updating Qmax. Which is also the reason why the 37% min.passed charge parameter is private and no public. Lowering this just to get Qmax updates can significantly corrupt Qmax and then gauging becomes pointless.

hello Dominik，

We verified using the three fs files you provided with different batteries. But none of the Qmax are updated, please help us analyze why, thanks!

new Fs file,Qmax not update.rar