• Locked Locked
  • Replies 4 replies
  • Subscribers 64 subscribers
  • Views 641 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

bq27421 - proper manufacturing process?

Igor Orlovich
Prodigy 170 points

What is a suggested process for bringing a bq27421 design to production? It appears there are many ways to get there.

1. Minimal one would be to just set the four things (Design capacity, termination voltage, design energy and taper rate) and then just let the product go through various cycles in normal use. We are finding this may not necessarily expose the gauge to all the conditions it needs  to learn.

2. Perform controlled learning on the bench, and then use Qmax value and Ra tables from that to load to every gauge  by their host processors on ITPOR 

3. Perform controlled  learning on the bench, generate golden image and have host processor load all of it on ITPOR

Comments and thoughts are welcomed!

  • Intellectual 455 points
    Hi Igor,

    I am also very interested in getting a proper answer to your question.

    We are also using a bq27421 gauge in our system, and we are about to go in production, but I'm still confused about what would be the best (and easiest) sttrategy to achieve sufficient gauge accuracy.

    The bq27421 uses the IT-Lite algorithm and a learning cycle is apparently not necessary during developpment, as It will be done during field operation. But as Igor mentioned, I don't see how the learning cycle conditions could ever be met. For example, how would the 1uV/s voltage stability be met in order to get OCV measurement, since the battery would most likely never "relax" long enough ?

    Is there any command or register to has to be set in order to enable the IT-Lite algorithm ? In the bq27421, there is no IT_ENABLE command, nor QEN bit. It looks like setting bit 0 and 1 in the update_status register will allow to detect if learning cycle was done successfully (these bits should be cleared after a successful learning cycle). Does setting these bits enable the IT or is it always activated ?

    An other point I'm confused about is that in our board, the mininum current consumption is about 1.5mA. Is that low enough for the battery to relax and meet the 1uV/s condition ? It is mentioned in TI documentation that the average current should be 0mA to get the battery "relaxed" enough, but I suppose that in most practical cases, this is not achievable.

    One last thing, what would be the significant difference for the bq27421, between loading a complete golden image or just loading the learned QMAX and RA table ?

    I would really appriciate some clarifications about these points.


    Thanks in advance for any help !


    Ril
  • in reply to Ril Dank
    Prodigy 170 points
    Hi Ril,
    I never got a proper answer despite talking to TI directly and posting here. As you can see it's been open since March. We put a lot of effort into the design, but found this chip to be not very well documented and somewhat flaky. So this is probably not the answer you are looking for, but we ended up designing bq27421 out in favor of a much simpler chip - LC709203FQH. It's perfect for our application and requires very little configuration. A bit of a shame, considering how much effort we spent on it.
  • in reply to Igor Orlovich
    Intellectual 455 points

    Thank you for your post Igor. It is indeed not the answer I was looking for :/ We are not yet ready to change our design and use an other gauge, so I really hope TI (or anyone else) can bring some clarification about our questions.

  • in reply to Ril Dank
    TI__Guru* 76230 points

    Hi Ril.

    The question boils down to how much accuracy you are expecting the bq27421 to provide for your application. 

    For improved accuracy, a learning cycle is recommended so a golden gmfs file can be generated and programmed on multiple gauges. That way if there is no possiblity of qmax getting learned in the field, then at least the resistances will continuosly be  updated. Also you can set op config B register to 4f. bit 6  although reserved actually means RanoQmax which means you can have resistance updated without having a  qmax update during a learning cycle. On this gauge, inorder to do learning, the update status should be set 03. Once learning cylce is complete, if successful, the update status would have gone from 03 to 00. IT is always enabled on this gauge. There is no command to disable or enable IT algorithm.

    You need to ensure that the chemistry is a match with the default ids programmed on the gauge otherwise you will have greater than 5% error in reported SOC.

    if your chem id is a match and your accuracy requirements is flexible, you can simply program those 4 parameters as stated in the quick start guide and then cycle the battery to evaluate for accuracy. If ever power is cycled on the chip, the host should check the ITPOR bit to see if it is set. if it is, those 4 parameters need to be reloaded on the gauge. 

    You can include the resistance and qmax parameters as values that need to be stored in nonvolatile memory of your mcu periodically and  reloaded on the gauge if power is cycled on it so that the gauge can continue to be relatively accurate if the battery has been in the field for some time given that those parameters would have updated.

    Let me know if any of the explanations above needs clarification

    thanks

    Onyx