• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 2 answers
  • Subscribers 63 subscribers
  • Views 530 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

BQ34Z100-G1: VOK does not clear after fully charged during learning cycle

Priit Tänav
Prodigy 20 points
Part Number: BQ34Z100-G1
Other Parts Discussed in Thread: BQ34Z100EVM, , BQ34Z100

I am using a bq34z100EVM to perform a learning cycle for a 4S32P 200Ah LiFePO4 battery pack.

  • The voltage divider resistor R1 is replaced with 270 k, the R28 with a 18 k resistor and a jumper is placed to pins 1 and 2 on the header J2 to set the voltage divider to 270 k/18k.
  • The current shunt resistor R30 is replaced with two 5 mR 4 W resistors in parallel resulting in a 2.5 mR shunt.
  • Current and capacity values have been scaled down in the bq34z100-G1 configuration by a factor of 7.

According to the Technical Reference Manual sluubw5 and the Achieving the Successful Learning Cycle Application Report slua903, the VOK bit should clear in 2 h after charging the battery to full, but it has been 10 h after completing the charging cycle and the VOK bit is still HIGH, OCVTAKEN is LOW.

Should I wait for longer and for how long or should I move on to discharging the battery?

Has the learning cycle failed and if it has then why?

Also, the Learned Status changes from 0x04 to 0x05 immediately after completing the charge not after two hour rest period. Is this behavior expected?

I have attached the learning cycle log up until this point and the Data Memory exports. Chemistry ID is set to 0418. Let me know, if some more information is necessary.

bq34z100-G1_200Ah_LiFePO4_learning_cycle_2021_09_29_02_10-2021_09_30_09_15.zip

Feedback is appreciated!

  • 0
    TI__Guru** 105590 points

    Hello Priit,

    Thank you for providing all the logged data along with .gg file and chem ID!

    I believe the currents you are using on the EVM may be too high, if you are scaling by 7 and the log shows 3A, you're at over 21A during some portions of the cycle.

    If you have enabled the LFPrelax functionality the gauge will take a DOD at the end of charge and use that as the reference, so that is normal to have update status change immediately if you have that feature enabled. You should be able to continue to discharge between C/5 and C/10 to complete the learning cycle with the Ra table update.

    VOK may not be clearing due to the voltage not settling and being in a flat voltage region (specified in the chem ID)

    For more information on the OCV flat region: e2e.ti.com/.../bq34z100-g1-cell-term-v-delta

    Sincerely,

    Wyatt Keller 

  • 0 in reply to Wyatt Keller
    Prodigy 20 points

    Hi Wyatt,

    Thank You for the information and the insight regarding the flat voltage region.

    I understand that the EVM is not designed for high currents and therefore I have bypassed the TB3 screw terminals for BAT- and PACK- and soldered 5 cm long 2.5mm2 wires to the BAT- and PACK- pins onto the bottom side of the PCB. These short wires are then connected to 4mm2 wires using a WAGO connector since there wasn't enough room to directly solder 4mm2 wires to PCB. One of the parallel shunt resistors is soldered directly to BAT- and PACK- pins and the other one is at the "normal" R30 position. That way at least roughly half of the 21-22A current does not flow through the EVM PCB traces.

    I completed the learning cycle and after relaxation after discharge the Learned Status incremented to 0x06.

    I noticed that the Remaining Capacity and the State of Charge decreased very rapidly to 0 during discharge right after the State of Charge had dropped below 89%. Is this normal when the learning cycle has not yet been completed or is it indicative of a poorly matched Chem ID?


    I attached a log file of the whole learning cycle.

    bq34z100-G1_200Ah_LiFePO4_learning_cycle_2021_09_29_02_10.log

    I have also a question regarding the Chem ID.

    The battery pack consists of four 1S36P "sub-packs". This sub-pack has been assembled by the battery manufacturer and I am unable to run a relax-charge-relax-discharge-relax cycle on a single cell basis, therefore I ran a cycle to determine the correct Chem ID using one of those sub-packs, but the Gauging Parameter Report replied with an error:

    The following errors were reported by the calculation engine:
    Accuracy Error: Deviation is so high that it is most likely due to anomaly in the data. Please check that data files have recomended format, units and test schedule

    I have attached the log file and the report. The GPC report returned the best chemical ID with max deviation 33.54%.

    LiFePO4_chem_id_cycle_2021_10_01.zip

    LiFePO4_chem_id_cycle_2021_10_01-report.zip

    What could be the reason for such high deviation and can You recommend further action regarding determining the Chem ID?

  • 0 in reply to Priit Tänav
    TI__Guru** 105590 points

    Hello Priit,

    I would definitely recommend performing the chem ID matching with one cell, LFP chemistries are difficult to gauge so ay leg up with reliably we can get I would do. Also the learning cycle should be performed on the final systems PCB, the golden image that gets exported has the Ra table, which includes trace resistances which can vary from the EVM to your board. Usually we like to see under 5% DOD error, so 33.5% is most likely too high and will lead to gauging errors.

    The learning cycle you shared looks like it completed successfully, I would be more curious of the accuracy on the next cycle since the gauge had not updated the Ra table on the discharge cycle you shared so the SOC won't be accurate midway through the learning cycle. 

    Sincerely,

    Wyatt Keller

  • 0 in reply to Wyatt Keller
    Prodigy 20 points

    Hi Wyatt,

    Thank You for the recommendations.

    I had one of the cells extracted from the pack and performed a chem id cycle with it. I was using a 100 mR shunt on the bq34z100EVM for the current measurements.

    LiFePO4_chem_id_cycle_2021_10_06.zip

    LiFePO4_chem_id_cycle_2021_10_06-report.zip

    The reported DoD deviation decreased significantly, but the best matching LiFePO4 chem ID has still a max DoD error of 11.79%.

    Is there anything indicative in the log file that would explain why a <5% DoD error was not achieved?

  • 0 in reply to Priit Tänav
    TI__Guru** 105590 points

    Hello Priit,

    Your OCV curve does not look like most, there are 2 dips that are not usually present is other chem IDs I've seen. Do you know what is causing these dips in voltage?

    Sincerely,

    Wyatt Keller