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.

BQ34Z651: Incorrect Full Charge Capacity and Remaining Capacity

Part Number: BQ34Z651
Other Parts Discussed in Thread: BQ34Z653

Hi Everyone, 

I am having a persistent issue whereby the gas gauge often incorrectly changes the capacity of the battery, rendering all information about the battery's state of charge useless.

In the graph below shows the results of running 5 charge/discharge cycles of my 84600mWh battery over the course of a few days.

Raw Data:   Copy of BM2_cycle_test.xls

The cycle looked like this:

1. Charge until taper current is reached

2. Rest for 30mins

3. Discharge until the low voltage threshold is reached

4. Rest for 30mins

5. Repeat 5 times

While the Voltage plot mirrors this cycle nicely the Relative State of charge plot does not reflect this at all, with significant clipping below what it perceives to be 0% and it never reaches full capacity because of that offset. Additionally the maximum SOC reached each cycle decreases, indicating that the the full charge capacity is being recalculated incorrectly every cycle. Following the experiment the Full Charge Capacity of the battery is 163770mWh, much larger than what it should be.

The current and voltage measurements are calibrated to within +/-10 mA/V just before the experiment and those calibrations are still accurate following the experiment. 

This is an issue I have seen many time during the development of this battery, while this is the highest full-charge capacity measurement that I have seen, I have seen values as low as 17000mWh. The problem also persists between battery packs so it is not device specific.

This problem is wreaking havoc on all state of charge measurements such as relative state of charge and time to empty and is rendering these completely useless.

Is this a problem that other people have come across and may have a solution to?

Thanks, David

    • Hi,
    • A suggustion is that you should relax 2 hours after charge and 5 hours after discharge to make sure the battery voltage stable. 
    • 30 minutes relax is short and you can relax longer.
    • Another thing is to make sure the chem id is correct.

  • Hi Terry,

    I am happy to make it relax for longer, my only worry is that that is not at all representative of how it will get used in it's parent system, in fact no rest period would be more accurate (although it will probably rest once it is fully charged). If the accuracy if these measurements is dependent on long resting periods then I have bigger problems.

    I selected the chem ID as 2125 as recommended by TI's Gauging Parameter Calculator, it there a command that I can send to verify that I didn't load an incorrect one by accident?

    Thanks,

    David

  • Hi Terry,

    I repeated the experiment after re-programming the chemistry ID to one that is correct for the HG2 cells used and increasing the resting period to 2hrs at both charging and discharging peaks.

    It may be worth noting that each of the cells in my battery consists of 4 HG2 cells in parallel, a previous discussion I had with TI indicated that this should not be an issue.

    The results of this last test are as follows:

    It can be seen from these graphs that the same issue that I was having before is continuing. In the second chart you can see the battery re-calculating it's full charge capacity every time that the battery is fully charged. At ~2600 minutes it decides that the capacity is zero, I believe this is because the capacity it calculated exceeds the maximum design capacity for the BQ34Z651 device.

    Do you or anyone else have any insight into why this might be? 

    I have attached both the data that I generated these plots with and the ROM file that I am using for your reference.

    Thanks,

    David

    651_2cell_HG2.zip

    Battery Test 2.xlsx

    • The max battery capacity for bq34z651 is 29000mAh, please check your battery is not over 29000mAh.
    • Have you send IT enable command before doing the learning cycle?
    • www.ti.com.cn/.../slua597.pdf
    • Here is a link that shows how to perform a successfull learning cylce.

  • Hi Terry,

    Thanks for supplying that document, I learnt a lot from it.

    The capacity of my battery is 86400 mAh so I am well under the limit, I have a couple of follow up questions:

    1. The document you sent through details how you can use the value in UPDATE_STATUS to debug the learning cycle. However the values it describes (0,1,2) seem to differ for the descriptions in the Technical reference manual (http://www.ti.com/lit/ug/sluu986/sluu986.pdf) and also what I am seeing in the device register (0x04). Do you have another document that will help me transpose these different numbers?

    2. I'm able to start the learning just fine and I see the FC flag come on at the top of charge but the VOK flag never (it's been 15 hours now) goes away. So I do not believe that the battery performs the Qmax update. The document you sent me seems to indicate that stopping charging too soon or too late after the FC flag turns on can be bad. If there a certain time period that I should be waiting?

    3. Due to the integral nature of the coulomb counting, any small inaccuracy in the current measurement results is quite a marked deviation of the full charge capacity from what it should actually be. Is there an established methodology for improving the accuracy of the current measurements?

    Thanks,

    David

  • Hi Terry,

    Have you had a chance to look at my follow up questions?

    Thanks,

    David

    • Hi,
    • The TRM in your link is BQ34Z653, not BQ34Z651, please check you have use the right document.
    • The make the Coulomb counter accurate, a good chosen value of sense resister and the good PCB layout is very important to achieve this.

  • Hi Terry,

    I am aware that that document is not for the right part but so far the parts have been sufficiently similar for it not to be an issue. TI hasn't made the 651 TRM available online, are you able to provide me with that? Also that still doesn't account for the mismatch between the values that are in the UPDATE_STATUS register and those that are in the learning cycle document, so are you able to provide a learning cycle document that has the correct values for this family of parts? For example UPDATE_STATUS on my device is currently 0x04, which is outside the range of values described in the document you provided so I can gain do insight into what is going on.

    I am happy with the board layout and kelvin sampling scheme. but I have ordered some better sense resistors to see if that makes a difference.

    I'm able to start the learning just fine and I see the FC flag come on at the top of charge but the VOK flag never (it's been 15 hours now) goes away. So I do not believe that the battery performs the Qmax update. The document you sent me seems to indicate that stopping charging too soon or too late after the FC flag turns on can be bad. If there a certain time period that I should be waiting?

    Thanks,

    David

  • Hi Terry,

    Can you or someone else please help me solve this issue, it is starting to affect my company's ability to deliver our product.

    Thanks,

    David