Because of the Thanksgiving holiday in the U.S., TI E2E design support forum responses may be delayed the week of Nov. 21. Thank you for your patience.

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: GPCCHEM tool - error percentage implications

Part Number: BQ34Z100-G1
Other Parts Discussed in Thread: BQSTUDIO, GPCCHEM

Dear experts,

I am trying to characterize new LiFePO4 cells to be used in a product.

I submitted my charge/discharge data logs via online tool as described here:

https://ti.com/lit/slva725

The reports came back with recommended chemical IDs, and the deviations are quite high (9.5-9.9%):

My questions are:

(1) What does this error typically mean for SOC estimation? What kind of SOC accuracy can I expect, and where would the 10% error typically show up (closer to 100% SOC, 0%, or across the entire range)?

Since the gauge (BQ34Z100-G1) has a Coulomb counter, my understanding is that chemical ID only comes into play when the gauge needs to estimate SOC based on the measured voltage - after that, any charge put in or taken out will be taken into account. Please correct me if this is wrong...

(2) In an application where the gauge is active most of the time, can we expect that SOC accuracy will improve over time, despite the initial 10% estimation error due to Chemical ID not perfectly matching the cell? Or are we pretty much guaranteed to always have that 10% error?

(3) Assuming I can't get any better than 10% error in the SOC with these chemical IDs, what would be the next step - do I need to physically submit cells for characterization?

Thank you!

  • Hello Pavel,

    For LFP batteries if common to see such high error values. Must of the time this is in the flat region of the cell voltages since LFP batteries have an extremely flat OCV a small change in voltage could be a large change in DOD.

    The gauge uses the chem ID in relaxation and to get the starting DOD points before an active charge or discharge to know the initial DOD value. If you cycle the batteries through the flat portion of the voltage and never relax in the flat region, most likely the DOD error would be much lower since you would relax in a voltage region that has a larger DOD vs OCV curve.

    Can you share your data submitted to the tool so we can review it? Usually if you get over 5% max DOD error we recommend getting the cells characterized, but as I mentioned above if you are never resting in the flat region it will not be as much of an issue for impedance track. Even with the cells characterized the voltage is so flat it may have high error.

    Sincerely,

    Wyatt Keller

  • Hi Wyatt,

    Thank you!

    Here are the files I submitted and reports I received from TI:

    Cell A data.zipCell A report.zipCell B report.zipCell B data.zip

    For the second cell, the recommended ID is 6106... in bqStudio, this corresponds to a NiMH cell.

  • Hello Pavel,

    Thanks for the data, it looks like there could be some changes in the data submitted to help the tool better match the chem ID. I would make sure you follow the guide, especially section 3.2 LiFePO4/LFP/Lithium-Iron Phosphate for the GPCCHEM tool.

    I would only include one discharge in the submitted data, and make sure you get the 5 hour relaxation periods before and after the C/10 discharge.

    https://www.ti.com/lit/pdf/SLVA725A

    Sincerely,

    Wyatt Keller

  • Hi Wyatt,

    I did follow the guide - the manufacturer recommends charging at C/2, and this is a 4000mAh cell, so the charging was done at 2A. The discharge was at -400mA (C/10). Relax times were 5 hours, both after charging and after discharging.

    I tried re-submitting the data with only one charge/discharge cycle, but the same exact result came back.

    When you say that the chemical ID is only needed to estimate DOD during relax - does it mean every time the battery goes in relax, or only when the gauge is powered up? Because in our application, the gauge is ON most of the time.

    And let's say the initial DOD estimate was inaccurate because the battery was in relax somewhere in the flat region of the curve ... then we start charging or discharging the battery, and it reaches full charge or discharge, where gauging is accurate ... will SOC reported by the gauge self-correct at that time? Or will it remain inaccurate because we started in the "difficult" region?

    Thank you!

  • Hello Pavel,

    Sorry I did not realize your time steps are not the same throughout the data logging. The rests do appear to be long enough.

    It is when the gauge enters relax and starts looking for the dV/dt conditions for an OCV measurement. This is around 4uV/s. If the gauge is active it will still enter relaxation and take an OCV to correct the DOD when the voltage is relaxed.

    The gauge will correct itself when it takes another OCV, if the gauge took an OCV in the flat region (which we have protections for to prevent this) then the error may remain until a new OCV depending on the Impedance Track settings.

    Sincerely,

    Wyatt Keller

  • Thank you, Wyatt.

    We will get started with the chemical IDs returned by the GPCCHEM tool and do some testing to evaluate gauging performance.

    A couple more questions on this:

    (1) What is the process for getting new cells characterized by TI? Is there a step-by-step document similar to GPCCHEM?

    (2) If I understand correctly, when cells get characterized by TI, a new chemical ID and a "chemistry" configuration block (to be loaded onto the gauge chip) are created. Does the "chemistry" block get saved as part of the "golden image" (df.fs file), or is another step required to deploy that new "chemistry" to the device? This affects products that are already in the field: we can update the gauge image (df.fs) during firmware update, but I'd like to know if something else is needed.

  • Hello Pavel,

    There is not a step by step guide, we would need to take it to private messages and I will explain.

    Yes the DF.FS contains all of the chem ID information you would need to do a field update.

    Sincerely,

    Wyatt Keller