• State TI Thinks Resolved
  • Locked Locked
  • Replies 11 replies
  • Answers 3 answers
  • Subscribers 64 subscribers
  • Views 509 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-G1: BQ27421-G1A

William Widjaja1
Prodigy 70 points
Part Number: BQ27421-G1
Other Parts Discussed in Thread: GPCCHEM, BQSTUDIO, BQ27421EVM-G1A

Hi TI E2E forum,

I am using the BQ27421 Fuel gauge, and I have noticed that there is a SOC "jump" from the SOC before turning my device off to after turning it back on. The biggest jumps is from a SOC of 70% -> 2%. The SOC stays at 2% whilst the battery is being heavily charged (C/2), or lightly discharged (C/20).  

I am not controlling the battery current when the device is turned on, so it may be charging or discharging. 

Also worth noting is that I haven’t seen the SOC jumps from a full charge battery at rest (SOC = 100%) going into discharge. I think this is because the SOC cannot be any higher that 100% during the FG initialisation (or OCV measurement). Likewise I didn’t observe SOC jumps from a fully discharged battery (SOC= 0%) going into a charging state, we think that this is because the SOC cannot go lower than 0% during the initialisation going into a charging state. 

Continuous charge, rest and discharge cycles as described in learning documents are achievable with the Fuel gauge configuration, and the true SOC errors I have observed are less than +/-5%. Note that I haven't observed SOC jumping is observed when performing these ideal cycles. 

Can you please advise where I should look to manage the SOC jumps? 

Best regards,

William 

  • 0
    TI__Guru 52115 points

    Have you used the GPCCHEM to check whether you are using a proper chemistry id?

    Also, could you provide your gg file and a bqStudio log file so that I can take a look?

    Andy

  • 0 in reply to Andy Liu (BMS)
    Prodigy 70 points

    Hi Andy,

    The learning file outputs we configure to the fuel gauge are below:

    Qmax Cell 0 = 16384

    R_a0 0 = 94
    R_a0 1 = 94
    R_a0 2 = 95
    R_a0 3 = 100
    R_a0 4 = 73
    R_a0 5 = 70
    R_a0 6 = 84
    R_a0 7 = 98
    R_a0 8 = 95
    R_a0 9 = 92
    R_a0 10 = 114
    R_a0 11 = 128
    R_a0 12 = 219
    R_a0 13 = 447
    R_a0 14 = 713

    V at Chg Term = 4190
    Avg I Last Run = -200
    Avg P Last Run = -200
    Delta Voltage = 2

    Deadband = 1

    Please let me know if there are other parameters.

    I dont have the bqStudio log file at the moment.  I will follow up and see if it is available

  • 0 in reply to William Widjaja1
    Prodigy 70 points

    Hi Andy,

    I have some additional questions:
    • In our system there will be cases where the FG will lose power, and the host will need to reconfigure it to our specific configuration. We realize we need to do this before there is significant load on the battery. Is it right for us to configure the chip and then poll the OCVTAKEN for a marker of when we can apply load? Obviously we would like to start-up as soon as possible for our application.
    • We also notice that there is a short period after configuring it when some readings, like the state of charge, are 0%. We see that waiting a second or so seems appropriate. Is there any marker that we can poll to indicate that the readings are ready?

    Thanks,

    William 

  • 0 in reply to William Widjaja1
    TI__Guru 52115 points

    I would need to see your log files in order to have a better understanding of your issue. Can you provide them? Thanks.

    Andy

  • 0 in reply to Andy Liu (BMS)
    Prodigy 70 points

    Hi Andy,

    Please find the last log files, log error files and Data flash file I could find. 

    The log files were generated using the EV2300 connected to the BQ27421EVM-G1A with a battery simulator inorder to meet the 1uV/s requirement for learning. The Rest times around 2hrs, and the charge rate was C/2, discharge was C/6.

    learning-20-6-18-175mAh-3-4V-terminateV.gg.csv0068.29-6-18-t6-bat-det-10k-175mah-smu-automate.logPlease Note that we did perform another learning cycle in our actual device with the final PCB traces, real battery additional battery circuitry (no EV2300 or EVM connected) which generated the 20 or so learning parameters posted earlier.

  • 0 in reply to William Widjaja1
    TI__Mastermind 24870 points

    Hello,

    It does not look like Qmax updated. Can you make sure that QMAX_UP and RES_UP are updated after the learning cycle? These bits should be set.

  • 0 in reply to Kang Kang
    Prodigy 70 points

    Hi Kang,

    Thanks for that, I can try and rerun the learning that we performed in our device using the EV2300 today or latest tomorrow and get you the data logs and flash. 

  • 0 in reply to William Widjaja1
    Prodigy 70 points

    Hi, 

    Apologies for the late response, here is the data flash exported from BQstudio after a successful learning cycle on our specific device. The full register log is also attached at the fastest sample rate;10-6-20-learning-attempt5-P2-successful.log14-6-20-P2-FG-learning-parameters-attempt5-successful.gg.csv

  • 0 in reply to William Widjaja1
    TI__Mastermind 24870 points

    Hello William,

    The learning cycle looks successful. The SOC accuracy on the second cycle looks ok.

  • 0 in reply to Kang Kang
    Prodigy 70 points

    Thanks for the confirmation. DOes the Chem ID need a better match here?

    Are you also able to provide answer to following:

    • In our system there will be cases where the Fuel Gauge will lose power, and the host will need to reconfigure it to our specific configuration. We realize we need to do this before there is significant load on the battery. Is it right for us to configure the chip and then poll the OCVTAKEN for a marker of when we can apply load? Obviously we would like to start-up as soon as possible for our application.
    • We also notice that there is a short period after configuring it when some readings, like the state of charge, are 0%. We see that waiting a second or so seems appropriate. Is there any marker that we can poll to indicate that the readings are ready?

    Thanks,

    William 

  • 0 in reply to William Widjaja1
    TI__Mastermind 24870 points

    Hello William,

    That is up to the GPCCHEM report for chemID match.

    In order to close this thread, please move your new questions to a different thread each time you have new questions.

    1. You should look at the ITPOR bit and not OCVTAKEN bit. Please refer to EVM user's guide or TRM for the power modes and configuration mode. There's also an INITCOMP bit you can look at.

    2. You do not want to send i2c transactions more than 2 times a second for this device. We only update once a second. You need to follow the EVM user's guide on BAT_DET.

    Please read the documentation on the EVM user's guide and follow the steps for proper configuration.