BQ27520-G4: Learning cycle failed

Hello Morita-san,

I am not able to find any issue with the procedure. I assume you sent Enable Impedance Track (0x21) and issued a reset command (0x41) at the beginning since the flags seems to be correct.

I noticed that QMax Cell in OCVx table is set to 1000. I would expect it to be set closer to your Design capacity based on the chemID.

Here are some things that help resolve the issue
1. One of the recommendations is to ensure that the battery is at a low SOC and relaxed when sending IT_ENABLE.
2. Resistance tables may not update if the battery chemistry does not match what was programmed. RUP_DIS bit is set in middle of discharge in such cases. Verify battery chemistry match with GPCCHEM

Hi Shirish,

1-1.What percentage should the SOC value be below?

1-2.How much relaxation time do you need?
Do you need about 2 hours as written in slua903?
slua903："Typically it takes 2 hours in a CHARGED state・・・"

2.If the battery chemistry does not match the programmed one, may the initial Qmax not be updated?

It is an additional question.
I want to quickly determine if the learning cycle has failed.
Where should the learning cycle stop?

Best Regards,
Kazunori　Morita

1. Below 30% is considered low

2. Relaxation depends on the specific battery. The dv/dt rate must be reached. See https://www.ti.com/lit/an/slua450a/slua450a.pdf section 2.3 Update of Qmax

3. If Qmax does not update at the point it is supposed to update, then the learning cycle failed.

Hi Shirish,

Please answer the following questions.
2.If the battery chemistry does not match the programmed one, may the initial Qmax not be updated?

3.About updating Qmax
I don't understand whether the offset error, which is the update condition of Qmax, satisfies the condition.
The sense resistor is 10mΩ.

For example, please tell me specifically how the offset error changes in the following results.
BQ27520-G4_Learning cycle failed_20230316_001.xlsx

Best Regards,
Kazunori　Morita

If the chemistry does not match the programmed one, then Qmax may not be updated because the gauge will be expecting different characteristics for Qmax update.

dv/dt is the rate of change of voltage over time. To make sure there is zero current, physically disconnect both load and charger completely and wait 2 to 5 hours and the battery should relax enough to satisfy this condition. Let me know if you meant something else by "offset error".

Hi Shirish,

3.About updating Qmax
I want to know when the learning cycle failed.
Do both results satisfy the initial Qmax update criteria?
Qmax is not updated during relaxation after the first charge in either result, so can we conclude that the learning cycle has failed?

BQ27520-G4_Learning cycle failed_20230308_002.xlsx
BQ27520-G4_Learning cycle failed_20230316_002.xlsx

BQ27520-G4_Learning cycle failed_20230308_002.xlsxBQ27520-G4_Learning cycle failed_20230316_002.xlsx

4.Qmax update timming

We have previously observed that Qmax does not update during relaxation after the first charge, and Qmax updates during relaxation after discharge.

BQ27520-G4_Learning cycle failed_20230323_001.xlsx

BQ27520-G4_Learning cycle failed_20230323_001.xlsx

Not meeting the initial Qmax renewal criteria?
Also, which condition is not met?

Best Regards,
Kazunori　Morita

Yes, the learning failed.

I would check if the battery has been discharged enough. I see that it is discharged to 2790mV. Most batteries have a lower voltage on datasheet. The condition of "full discharge" may not have been met.

This is based on the observation that SOC is reaching 100 before significant tapering of current starts.

Hi Shirish,

>I would check if the battery has been discharged enough. I see that it is discharged to 2790mV. Most batteries have a lower voltage on datasheet. The condition of "full discharge" may not have been met.
>This is based on the observation that SOC is reaching 100 before significant tapering of current starts.

What does the above describe?
Is it an answer to "3.About updating Qmax"?

4.Qmax update timming
Why Qmax is not updated on first charge and Qmax is updated on discharge
I would like to confirm the following result.
Have you checked BQ27520-G4_Learning cycle failed_20230323_001.xlsx?

The voltage when discharging is 2586mV
The current when SOC is near 100% is already small at 30mA.

Not meeting the initial Qmax renewal criteria?
Also, which condition is not met?

Best Regards,
Kazunori　Morita

3. It is answer to Q3. Specifically step 3.2.1 Discharge Battery to Empty. If it is not discharged low enough, then the value stored after relax is not good for Qmax update

4. I think i looked at the other log and commented. Will check the new one and let you know.

Sorry, i did not look at the new data when i answered previously. The learning cycle completed successfully in the new log.

When update status 0 changed to 1, then the first part was successful

When update status 0 changed to 2, then the second part was successful

4.Qmax update timming

The answer I'm looking for is not whether the learning cycle succeeded or failed
I want to quickly determine if the learning cycle has failed.
Where should the learning cycle stop?

If Qmax is not updated on the first charge, how do I know if the learning cycle should continue or stop?

Best Regards,
Kazunori　Morita

You learning cycle was successful.

Learning cycle is successful if update status changes as expected. If update status does not change, then stop and restart

I prepared a new file because it seems that the intention of the question is not conveyed.
please confirm the contents.
BQ27520-G4_Learning cycle failed_20230329_001.xlsx

BQ27520-G4_Learning cycle failed_20230329_001.xlsx

Best Regards,
Kazunori　Morita

Back to the main topic of this thread.
I went through a learning cycle based on the advice below, but it failed.

What is the problem that the learning cycle fails?

>Here are some things that help resolve the issue
>1. One of the recommendations is to ensure that the battery is at a low SOC and relaxed when sending IT_ENABLE.
Attached is the result of running the learning cycle.
BQ27520-G4_Learning cycle failed_20230327_01.zip

BQ27520-G4_Learning cycle failed_20230327_01.zip

>2. Resistance tables may not update if the battery chemistry does not match what was programmed. RUP_DIS bit is set in middle of discharge in such cases. Verify battery chemistry match with GPCCHEM
I ran GPCCHEM and the Best ChemID was 2033.
I am attaching the GPCCHEM report.
No.210501553_20230227-report.zip

No.210501553_20230227-report.zip

Attached is the result of performing a learning cycle with ChemID2033.
In addition to changing the ChemID, we changed the Design Capacity and QmaxCell0,1 so that the initial FCC is closer to the True FCC.
BQ27520-G4_Learning cycle failed_20230403_01.zip

BQ27520-G4_Learning cycle failed_20230403_01.zip

Best Regards,
Kazunori　Morita

Hello,

I checked the log BQ27520-G4_Learning cycle failed_20230327_01.zip

In the learning cycle log, the discharge needs to be lower. It is recovering to more than 3V. This probably results in a restricted range of learning.

I would recommend using a c/10 rate and get it down to 2700mV for this cell.

The Discharging End Voltage of this cell is 2750mV.
Should I still go down to 2700mV?

Best Regards,
Kazunori　Morita

In that case go down to 2750mV using a C/10 or lower rate upto C/20 (c/5 rate terminates much sooner than C/10 rate)

A learn cycle was performed at 115mA and failed.
Attached is the result of running the learning cycle.
BQ27520-G4_Learning cycle failed_20230412_01.zip

BQ27520-G4_Learning cycle failed_20230412_01.zip

What is the problem that the learning cycle fails?

Best Regards,
Kazunori　Morita

I think there has been some confusion.

Only the first discharge should be 115mA until it goes to 2750mV. There is no learning in the first discharge, it just needs to empty the battery. The next discharge is for learning and this must be between C/5 and C/10 like you did before.

Next time I do a learning cycle, I will try to set the second discharge current to 350mA.

About the first discharge current value:
The battery relax voltage after discharge at 115mA was about 3020mV.
Is it okay to leave the first discharge current at 115mA?
Or should it be lower than 115mA?

Best Regards,
Kazunori　Morita

It really depends on how much capacity is left even if voltage recovers above 3020mV. To be sure that battery is completely discharged, it is best to reduce the current below 115mA when it starts approaching terminate voltage.

The first discharge was performed at 50mA, but the learn cycle failed.
Attached is the result of running the learning cycle.
BQ27520-G4_Learning cycle failed_20230419_01.zip

BQ27520-G4_Learning cycle failed_20230419_01.zip

What is the problem that the learning cycle fails?

Best Regards,
Kazunori　Morita

The log seems ok. However the following may be contributing to the issue.

1. Terminate voltage is set to 2750mV but the discharge does not go this low. Set Terminate voltage to 2800mV or 2850mV to ensure that a full discharge is detected.

2. The closest ChemID match according to previous post was 2033 but the GG file has 2678. Make sure chemistry ID 2033 is programmed using the Chemistry tab of bqStudio. Do not update the Chem ID in data flash directly. Note that the chemID process must use bare cells with no protector for matching.

3. I would also update "OCV Tables","OCVa1 Table","Qmax Cell"  and "OCV Tables","OCVa0 Table","Qmax Cell","1000","mAh" to the cell capacity.

1.From the forum below, I thought the Terminate Voltage value had nothing to do with the learning cycle. Is it a mistake?

https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/1195053/bq27520-g4-a-near-empty-battery-has-an-fcc-value-of-0/4529499?tisearch=e2e-sitesearch&keymatch=BQ27520-G4#4529499

2.ChemID2678 was chosen from the cell model number.
I mentioned it at the beginning of this post.

＞There are 3 Chemistry ID values (1146、2026、2678) for Panasonic UR18650ZY cell. 

I will try the learning cycle in ChemID2033 as well.

3.Ok

Best Regards,
Kazunori　Morita

1. Terminate voltage is not related to learning cycle directly. But it must be set to a reasonable value that allows a little margin. The value that is set is at the lowest of the battery spec and the discharge does not reach this level. Therefore the gauge will think the battery is not fully discharged yet. That can cause learning to fail.

2. The chemistry match does not include any of those IDs. That usually does not happen. I would update the chemistry database and check if the model/ID is marked obsolete. www.ti.com/.../GASGAUGECHEM-SW

I tried a learning cycle with the following settings, but it failed.
Terminate Voltage:2850mV
OCVa0 Table Qmax Cell:2400mAh
OCVa1 Table Qmax Cell:2400mAh

Attached is the result of running the learning cycle.
BQ27520-G4_Learning cycle failed_20230419_01.zip

BQ27520-G4_Learning cycle failed_20230426_01.zip

What is the problem that the learning cycle fails?

2.GPCCHEM was performed with the protection IC attached.
ChemID1146、2026、2678 is also included in the latest database Ver1025.

Best Regards,
Kazunori　Morita

I would refer you to this document https://www.ti.com/lit/an/slua902/slua902.pdf

Section 2.4 - has the importance of getting the right chemID match. The match should without protector.

Section 2.5 - Ideally we want the terminate and final voltage to be 2750 because that is what the battery specifies. The protector is interfering with this requirement and changing the terminate voltage as a workaround did not work. The learning cycle may need to be completed without the protector. I think Final voltage was not updated, you can try setting that higher but there is low probability of success.

>I think Final voltage was not updated, you can try setting that higher but there is low probability of success.
Ok.

I agree with you about the low probability of success.

Please provide the voltages covered by the following ChemID.
DOD = 1.0 @ 0deg.C OCV, DOD = 0.0 @ 0deg.C OCV

Best Regards,
Kazunori　Morita

1146: 4198mV, 2889mV
2026: 4192mV, 2727mV
2678: 4192mV, 2708mV
2033: 4193mV, 2847mV
1293: 4196mV, 2959mV
1105: 4196mV, 2866mV
2029: 4192mV, 2884mV
1106: 4200mV, 2988mV
1126: 4204mV, 2883mV