BQ40Z50-R2: FCC update is low

Hello,

Can you please tell me how the chemID was acquired for this gauge? Was the GPCCHEM tool used?

Also, would it be possible to receive a log file from bqStudio of this occurrence happening? This will allow us to take a deeper look into the issue.

Regards,

Anthony Baldino

Hi,

I followed the instructions provided on the TI official website to obtain the chemical ID, and did not use the GPCCHEM tool. Attached is the charging and discharging data, battery specifications, and chemical ID information I obtained using Battery Management Studio. Please help me solve the problem. Thank you!0474.chemistry_ID_data.zip5265.5895test_data.zip

Hello,

Thank you for sending those files, it seems like the chemID was chosen properly.

I see that there was a learning cycle completed for this gauge already. If possible, can you please send the learning cycle log file over so we can confirm that part?

Regards,

Anthony Baldino

Hi，

I couldn't find any previous learning records, but I would like you to help me confirm if there are any issues with my configuration file and if my battery specification book conflicts with the configuration file. This way, I can retrieve the learning process data again and provide it to you for analysis. Thank you

Hello,

It is difficult to determine a root cause based on the .gg file alone, however after looking through the sent file, it seems like your Charge term taper current is decently large at 450mA. We typically recommend that this value is set between C/10 and C/100.

Regards,

Anthony Baldino

Hi，

Hello, I found that the chemical ID I previously obtained did not match well due to voltage calibration accuracy errors. This time, I recalibrated and retrieved a new chemical ID of 7583, which I learned according to the TI learning instruction document. After completing the learning, I exported the learned gg.csv file and found that burning the chemical ID and writing the configuration for the same battery was reset, and the full charge capacity of the battery was updated to 2768mAh. May I ask if there were any issues with my learning process? Please refer to the attached collection logs. Thank you!7823.chemistry_ID_data.ziplearn_data.zip

Hi Ian,

Based on the sent files, it seems like the learning cycle was completed successfully. If possible, can you please send the report given to you by the GPCCHEM tool so we may check that the ChemID is a good fit?

Regards,

Anthony Baldino

Hi，

Please see the attached chemical ID report. Additionally, I would like to ask if there are any issues with my load mode and load select settings, as the current fluctuations in my usage scenario range from 4A to 15A?56_new_id-report.zip

Hi Ian,

Thank you for sending the chemID report file. It seems like the correct chemID has been chosen for this application.

If possible, could you try changing the Load Mode to 0 for constant current and see how this affects the performance? Typically, we recommend using Constant Power only if there are large fluctuations in the current within each cycle.

Regards,

Anthony Baldino

Hi.

I will now modify the load mode to 0 and learn again. Also, may I ask why the full charge capacity update during the learning process is also too low? In fact, when the discharge reaches the cut-off voltage, it can release more than 2900mAh, and I will charge and discharge the learned battery. After discharge, the FCC will update 2300mAh. Can you help me solve the problem from the learning data?

Hello,

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

Regards,

Nick Richards

Hi Ian,

While looking into the past test data you sent as well, I see that the temperature reading is getting to a very high value (almost 50 degrees C). Can you give more detail into what is occurring at that time? Many calculations take temperature into account, which could also be causing an issue with the gauge.

Regards,

Anthony Baldino

Hi，

I couldn't find any data close to 50 ° C from obtaining chemical ID 7583 and learning data, with a maximum temperature of 30 ° C. If it refers to the test data I initially asked, the temperature reaching 50 ° C is due to my use of 6A current for discharge at room temperature, and the temperature rise of the battery meets the specifications of the battery cell. If there is any problem, is there an error in selecting the chemical ID? I have learned again with load mode set to 0 and load select set to 7. Please check the attached data for any abnormalities. I have written the exported gg.csv file to another battery, and after sending the reset command, the FCC is 2828mAh, and the health rate is only 94%. This is inconsistent with the actual charging and discharging capacity during the learning process. Please help me find the problem. Thank you!load_mode_0.zip

Hi Ian,

Below is the temperature reading from TS1 from the test data earlier sent:

I believe for the chemID, this should be fine since the instructions have been followed for the inputted data to the GPCCHEM tool.

If possible, can you please tell me more about the load of the application? This will allow us to get a better idea of how to set the load select.

How much is the load changing between discharge cycles?

Regards,

Anthony Baldino

Hi,

My usage scenario is applied to drones, with a discharge current of around 600mA before takeoff, a discharge current of around 5000mA during takeoff hovering, an average of 12000mA during rapid flight, and a peak current close to 16000mA. This is the usage situation on the product. At present, I am using 1500mA current charging and 6000mA discharge for testing after completing my studies. The test result shows that the FCC update is relatively small. Although calculating the discharge capacity of SOC to 0% is correct, the voltage has not reached the cut-off voltage. Continuing to discharge until the cut-off voltage can release the battery cell's specification capacity. So, what is the reason for estimating that more capacity cannot be released?

Hi Ian,

Thank you for explaining, below is the descriptions of each of the Load Select options:

I believe the second option "Present average load from the beginning of current discharge cycle" should be tested based on the description given above.

Also, for load select/load mode changes, the entire cell does not need to be relearned. These parameters can be changed, then running a few charge/relax/discharge/relax cycles should be able to show the difference.

lan xin said:

Although calculating the discharge capacity of SOC to 0% is correct, the voltage has not reached the cut-off voltage. Continuing to discharge until the cut-off voltage can release the battery cell's specification capacity. So, what is the reason for estimating that more capacity cannot be released?

In the sent files, Term Voltage is set to 5400mV, and within the log file the voltage reading does reach this. This is also represented if you look into the individual cell voltages where the both of them become less then 2700mV. 

Based on the table below, it also seems like the gauge is seeing the correct amount of passed charge, since the design capacity for this cell is 3000mAh.

Regards,

Anthony Baldino

Hi，

You are right. It is precisely because the capacity of charging and discharging meets the battery specifications that I have doubts about the low update of the battery FCC. What is the reason for the low update? Do you have any better improvement suggestions? Or what further data is needed to confirm the situation

Hi Ian,

After receiving council from our team, they have also brought up that the FCC can be calculated at a lower value when placed in high load situations. If possible, could you test reducing the current of the discharge to see if this is able to affect drop in the FCC calculation? I understand that your application will be running a high load, however this will give us a good place to start looking into.

Regards,

Anthony Baldino

Hi，

I tried two batteries here, wrote the learned gg.csv file and reset it. The full charge capacity was updated to 2909mAh, and at this time, the batteries were both fully charged. 6A and 3A were used for discharge, and 6A was discharged to SOC of 0%. The FCC of the battery was updated to 2601mAh, and the FCC of the other 3A was updated to 2758mAh. The above discharge capacities correspond to the actual discharge capacity data of the load meter, but I did not use Battery Management Studio to record the logs. Observing the phenomenon, it is indeed true that FCC updates at low loads are higher than those at high loads. If we study this aspect, what methods are available to optimize this issue. Can you provide optimization suggestions for using different loads to record log data in the future? Thank you!

Hi Ian,

Is it possible for SMOOTH to be enabled on your device, and try to see the difference between the FCC updates at the 6A and 3A currents again? I believe this could help the issue. This bit can be found in the IT Gauging Config register of the gauge.

Regards,

Anthony Baldino

Hi，

My device has always enabled SMOOTH. You can check the gg.csv file I provided earlier. My IT Gauging Configuration is set to 0x56be. I have recorded a 1.5A discharge test log and found that the FCC update is only slightly lower. However, the battery cells should support 6A discharge to achieve the same discharge capacity. Please assist in analyzing based on the log. Thank you!1_5A_learn.zip

Hi Ian,

Thank you for sending the log file, we will look into it.

Just to confirm, how many cycles have been ran on the gauge after the learning cycle has been done? The FCC accuracy should improve after the first few cycles since the gauge will better understand the cell and the type of cycle.

Regards,

Anthony Baldino

Hi.

I have tested the log of 3A discharge. Please refer to the attachment to help compare the 1.5A discharge situation. Do you need to optimize the relevant parameters or perform high current cycle learning?3A_learn.zip

Hi Ian,

I see in the sent log and .gg file that the "Max Average I Last Run" is being used, where since there is no previous run to compare when the simulation occurs for the FCC calculation, this could be throwing off the calculation. If either of the Last Run (0 or 7) load selects are being used, I believe that more than one charge-relax-discharge cycle should be completed to see its true accuracy. Also, when changing the Load Select, there is no need to complete a whole new learning cycle, these parameters can be changed and tested immediately.

Regards,

Anthony Baldino

Hi，

I have always set the load select to 7, but the log data I have provided has not been modified for load selection. Did you expect me to set the load select to 1 before conducting a charge discharge cycle test? I am currently completing the discharge test for 6A. After the test is completed, I will analyze the logs for you again. If it is confirmed that the load select needs to be changed to 1 before testing, I will conduct the test again and capture the log data. Also, may I ask if it is necessary to relearn the value of CELL-TERM when modifying it? Currently, I have set it to 1

6A_learnzip.zip

Hi Ian,

Yes, please try changing the load select to 1 so we can determine if that is able to improve the seen results.

Regarding CELL_TERM, this value is used to determine whether the individual cell voltage or the voltage of the entire pack wants to be used to determine termination. Changing this will not require relearning of the cell, you should be able to change it and see the results.

Regards,

Anthony Baldino

Hi，

I have checked the data of the 6A discharge of the battery sent to you and found that the TrueFullChgQ capacity has been updated from 2858 to 2749 for the first time. The update condition is that the temperature has changed by 5 ° C. Why does it change so much? The subsequent updates of TrueFullChgQ have relatively small deviations, with only the initial temperature change having a large deviation. As a result, when my SOC reached 0%, I only released 2710mAh, but when it was discharged to the cut-off voltage, I was able to release 2860mAh. After settling, the temperature decreased and was updated to 2947mAh.

Hi Ian,

When the gauge calculates FCC, it factors in the Remaining Capacity calculation. The remaining capacity calculation takes in several different calculations at a specific OCV measurement point and temperature point. Since there is a varying temperature, this can effect the measurements that are being seen. More information of how these values are calculated can be found below:

Regards,

Anthony Baldino

Hi，

After extracting the chemical ID of this battery at room temperature, I only went through learning at room temperature. After learning, the Update Status value is 0x0e. Do I need to learn at high temperature? Is high-temperature learning conducted directly based on the completion of learning at room temperature? Do you need to modify any parameters before proceeding? How many degrees Celsius should a high temperature environment be? Also, may I ask if modifying the cutoff voltage requires relearning?Attached is a 6A discharge data with a load select of 1.

load_select_1_6Azip.zip

Hi Ian,

No, I do not think there is a reason to relearn for high temperature in this scenario. If possible, can you please use the GPCRA0 tool below and send the report to us? This will give us a better idea of what parameters should be changed to better optimize the FCC calculation. This will also give us a look into whether the Ra table could be causing this or not.

https://www.ti.com/tool/GPCRA0 

Also, you do not need to redo learning when term voltage is changed.

Regards,

Anthony Baldino

Hi，

I provided the data required for the GPCRA0 tool as requested, and it sent me a copy of the data. The report pointed out that my chemical ID was incorrect. Could you please help me analyze the next steps and whether I need to replace the chemical ID and retrieve the data again?

5895_Gpca0.zip5895_Gpca0-report.zip

Hi Ian,

Regarding the ChemID, can you please input the same data into the GPCCHEM tool below and see if there is a difference in the received result?

https://www.ti.com/tool/GPCCHEM 

Regards,

Anthony Baldino

Hi，

I sent it to the GPCCHEM tool, but it shows that there is an issue with the report provided. I am using a typical 6A discharge, and it seems that this does not meet the requirements for obtaining a chemical ID. Please confirm if you need to try again. In addition, I followed the chemical ID extraction process and confirmed that the optimal chemical ID was 5895. The attached ls_chem_id.zip is the data I collected according to the chemical ID guide book and the report I obtained through the GPCCHEM tool is ls_chem_id report.zip. Then you asked me to submit the data sent to the GPCRA0 tool yesterday to the GPCCHEM tool, and the report I received was 5895_GpcChem-report.zip. Please refer to the attachment for specific details. Another question is, after obtaining the chemical ID, I only learned at room temperature. After learning, the Update Status value is 0x0e. Do I need to learn at high temperature? Is there any relevant guidance on the learning steps?

ls_chem_id.zipls_chem_id-report.zip5895_GpcChem-report.zip

Hi Ian,

lan xin said:

I sent it to the GPCCHEM tool, but it shows that there is an issue with the report provided. I am using a typical 6A discharge, and it seems that this does not meet the requirements for obtaining a chemical ID. Please confirm if you need to try again. In addition, I followed the chemical ID extraction process and confirmed that the optimal chemical ID was 5895. The attached ls_chem_id.zip is the data I collected according to the chemical ID guide book and the report I obtained through the GPCCHEM tool is ls_chem_id report.zip.

Your understanding is correct, the GPCCHEM tool requires the discharge rate to be C/5 of the capacity of the stack, which could have caused the GPCCHEM tool not to work when 6A was being used. However, for the GPCRA0 tool, this is expecting the maximum rate in application, which should be the 6A.

Also, looking back into the original data for the GPCRA0, it states that your configuration only has 1 cell in series. Can you please confirm this? If there are more cells, this could be where the issue is occurring.

lan xin said:

Another question is, after obtaining the chemical ID, I only learned at room temperature. After learning, the Update Status value is 0x0e. Do I need to learn at high temperature? Is there any relevant guidance on the learning steps?

There are other tools for configuring the gauge at low or high temp, such as the GPCRB tool below. 

https://www.ti.com/tool/GPCRB 

Regards,

Anthony Baldino

Hi,

You are right, my battery is 2S1P. After changing it to 2, I obtained the correct report from the GPCRA0 tool feedback. Please review the report content. I imported the gg_out.csv file of the report and conducted a charge and discharge test. Are there any items that need to be modified? After testing twice, it was found that there was a 5% jump in the battery SOC. The reason for the jump was that the cut-off voltage was reached. I used a personal upper computer to capture the log data in the data file 5757-20240614185406-Log.txt. How can I optimize the jump phenomenon?
5895_Gpca0-report (1).zip5757-20240614185406-Log.txt

Hi Ian,

Thank you for clarifying that, I am glad the GPCRA0 tool was able to produce a result.

If possible, can you please send or inform me of what .gg file was used before the gg_out.csv was programmed to the gauge so I may compare the results to what the gauge had before?

lan xin said:

After testing twice, it was found that there was a 5% jump in the battery SOC. The reason for the jump was that the cut-off voltage was reached. I used a personal upper computer to capture the log data in the data file 5757-20240614185406-Log.txt. How can I optimize the jump phenomenon?

Looking into this log file, it is difficult to understand the structure and interoperate the data. If possible, can you please give more insight to this?

Regards,

Anthony Baldino

Hi，

I will send the. gg file I used before programming ggouts. csv to the instrument panel, named old_memory.gg.csv. Then I may know why the SOC skipped 0, and I found out that my Charge Term Taper Current was set to 100mA. If I used a DC source to charge, I could achieve a constant voltage charging current of this value, but my battery charging hardware decided to set this value to 400mA to set the FC. I could only use a DC source during the debugging phase, so when I used a charging device to charge the battery, I couldn't achieve effective charging termination. I have collected data on reaching effective charging termination and then standing for 6A discharge, and there was no SOC jump phenomenon. Additionally, the FCC update for 6A discharge was correct. I am very happy. Finally, I would like to inquire about the impact of setting the Charge Term Taper Current to 400mA. I will make the necessary modifications before conducting further testing.

old_memory.gg.csv3323.5895_GpcA0.zip

Hi Ian,

I am glad to see that results are being achieved.

It seems like the GPCRA0 tool was able to offer new Ra Table values, which could also be assisting the reduction of the SOC jump.

Regarding changing the Charge Term Taper Current, the only issue I can see is that it could potentially reach charge termination at a faster rate then if it was lower, in which typically we recommend setting this value to between C/10 and C/100 of the capacity.

Regards,

Anthony Baldino