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: Issues learning the impedance track algorithm

Part Number: BQ34Z100-G1
Other Parts Discussed in Thread: BQ34Z100

Hello,

We've been using the BQ34z100 for a while now as a system side pack gauge for a while now and are trying to create a golden image using the battery studio tool on our real application. The gauging/calibration seems to work fine in the field (Qmax guess, voltage, current etc without IT enabled) however the impedance track algorithm needs to be updated. During the new calibration I'm running into issues getting the update status/MaxError to change after several cycle attempts.

Specs:

  • LiFePo4 A123 ANR26650M1-B 16 series, 4 parallel, ~1000Wh/20Ah
  • 28A charge/4A discharge

I've attached a calibration cycle attempt and the used register values. It would be nice if I could get suggestions what's going wrong in the attempt and if any registers require updating.
In the attempt I did the following:

  • Load registers (attached)
  • Enable IT
  • Charge pack to full (with CC 1C + 5Hrs CV)
  • Relax Pack (~2Hrs)
  • Discharge pack (0.24C ~4Hrs)
  • Relax pack (~7Hrs)

Thanks in advance!

DataMem.gg.csv

85776.Log.log

  • Hello Vincent,

    Let us take a look through your files and get back to you with some suggestions.

    Sincerely,

    Wyatt Keller

  • Hello Vincent,

    I have a few questions:

    1. How did you select your chemID? 

    2. Based on the time stamps it looks like you created the gg.csv file at the end of the IVT experiment, is this correct? If so, we need the gg.csv file at the beginning of the test. 

    3. Could you try repeating the learning cycle based on the Impedance Track applications note? You could also try running a learning cycle on an individual cell to identify if the learning cycle is successful.

  • Hi Jessica,

    1) The description of 440 from the battery studio matches the used cell, I verified the use of this CHEMID in earlier TI mail correspondence.

    2,3) We'll re-run the cycle according to these documents and ensure the gg.csv is exported in the beginning. I will reply again soon with these files.

    Thanks,

    Vincent

  • Hi Jessica,

    I'm back with results with a re-running the cycle a few times. We've tweaked a few settings and the set-up (such as using a lower current supply), unfortunately with the same results. Could you or your colleagues analyze our files and perhaps give some feedback on the cycle?

    Find the logs attached. Thanks in advance!

    Regards,

    Vincent18-10-20.gg.csv18-10-20.log18-10-20_after.gg.csv

  • Hi Vincent,

    We will look into this and get back to you by the end of the week.

  • Hi Vincent,

    Could you try following these steps when running a cycle?

    1. Use just one cell

    2. Discharge cell to empty

    3. Relax for 5 hours to get dv/dt (ensure 0 current)

    4. Charge to full

    5. Relax for 5 hours to get dv/dt (ensure 0 current)

    6. Discharge at C/5 to terminate voltage

    7. Relax for 5 hours to get dv/dt (ensure 0 current)

  • Hi Jessica,

    The BQ34z100 is inside an integrated solution using per string 4 packs with 4s2p cells in each pack (4 parallel strings, 16 packs total), each pack has multiple protections. I cannot use the hardware with a single cell unfortunately. Currently we parallel 4 strings which could be reduced, would you suggest using 1 string instead perhaps?

    Thanks!

    Regards,

    Vincent

  • Hi Vincent,

    Since you can't test it on single cell I recommend that you make sure to discharge the cells all the way to empty and ensure that the current reaches 0 during relaxation. From your logs it looked like the current didn't quite reach 0A,

  • Hi Jessica,

    Could you elaborate on the current going to 0A? The relaxation current is around 14mA which is within the discharge/charge/quit threshold and sleep current.

    As for the discharge, the current went below the cell terminate voltage. However since we noticed the learned status did not change from 0x04 to 0x05 during the charge relaxation, so we assumed the cycle failed and did not relax the battery afterwards. It seems to me that full charge has been detected so I would expect the status to change. Any idea on why this did not happen?

    Thanks,

    Vincent

  • Hello Vincent,

    The gauge looks for the voltage to stabilize to take a good OCV (Open Circuit Voltage) measurement, for the learning cycle when the batteries are relaxing the current should be 0mA. 

    Sincerely,

    Wyatt Keller

  • Hi Wyatt,

    I performed the cycle again, ensuring the current shows 0mA, so this cycle the current is reporting 0mA after the charging. VOK was cleared after approximately 3 hours after the charging yet the learnedstatus remains 0x04.

    I'm not really sure how to proceed at this point. Could you please look at the data provided and tell me what could be going wrong here? We would really like to get the fuel gauge up and running soon. The only thing I can think of is using FCSet=-1, however the FC bit seems to toggle at a reasonable point already.

    Thanks in advance,

    Vincent

    2811.gg.csv

    11.gg.csv

    30-10-2020.log

  • Hi,

    Update status = 4 means Impedance Track is enabled but Qmax and Ra data are not yet learned. You can search our forum as there are many threads that talk about this issue.

    Best regards,

  • Hi Nick,

    I understand the meaning of each update status. That Qmax is not yet learned is clear, I'm however puzzling why the Qmax is not updated when the OCV measurement is taken after charge termination. The fact that the OCV measurement is taken and VOK clears after FC means that the charge termination and relaxation is working well.

    I suppose the only reason can be that the Qmax update must be disqualified (difference between Qmax cell 0 and Qmax passed Q too large?) or an offset error is occuring. I cannot find the exact conditions required for these disqualifications and if this is really what's happening, so cannot judge any further. I see in the cycle that after programming at the start of the cycle the initial DOD0 is slightly low.

    Any help in this regard would be appreciated. I've gone through quite some posts so far without a clear answer. The below post mentions a similar issue but no solution:

    e2e.ti.com/.../911562


    Regards,

    Vincent

  • Hi Vincent,

    Check out this thread: 

    Let me know of more question,