• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 2 answers
  • Subscribers 63 subscribers
  • Views 226 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

BQ40Z50-R3-DEVICE-FW: Custom Cell Support Process and Questions

Eric Castle
Intellectual 710 points

Part Number: BQ40Z50-R3-DEVICE-FW

Hello,

I am looking to use a custom cell with the BQ40Z50-R3 and wanted to make sure that I understand the process for this. As far as I can tell, I should do the following:

  1. Run the procedure outlined in “slva725a.pdf” (run a cycle on the cell, format the data correctly, upload it to TI, and the automated system responds with the best match; program the gauge with this).
  2. Run the procedure outlined in “sluubd0.pdf” (run a cycle on the cell at room temperature, then cold temperature, format the data correctly, upload it to TI, and the automated system responds with values to adjust in the gauge to handle low-temperature behavior; update the gauge accordingly).
  3. Run the procedure described in “slua848.pdf” to do a learning cycle on the full pack (again, running a cycle that includes some rest periods, while commanding the gauge at certain points).

I had some questions about the first two steps here:

  • Is it ok to run Steps 1 and 2 using a bare cell on a Maccor? Step 2 recommends the final product (which will not be available yet) or attempting to replicate the thermal conditions better (putting the cell in a case to mimic the thermal properties of the final battery).
  • Low temperature performance is important for my project (all the way down to -30C). Is it possible to run it at multiple temperatures and combine the optimization results? If not, how should I select the low temperature to test at?
  • What if there is not a good match for step 1 above?
  • Is there any way to refine the gauge’s behavior if it does not meet the customer’s accuracy requirements?
  • If this process does not work well for my application, what is the method to get a cell added to the list of cells that the gauge supports?
  • Can I adjust the gauge to be more pessimistic about SOC? (i.e., always try to err on reporting less energy is available, rather than over-estimating?)

Thanks,

-Eric Castle

  • 0
    TI__Genius 10370 points

    Eric,

    The procedure is correct for what you have. I dont know the slua numbers however the steps you have are correct

    1) Run a Room (C/10 Rate) discharge. The steps would be Charge, Rest, Discharge, Rest. You can run this on a bare cell using a Maccor. We will need Time, Voltage, Current, and Temp all logged. The thermistor needs to be taped directly to the cell for a good thermal connection

    2) Next run a Room and Cold (C/5 Rate) which will be used to tune for your temps. There is no tool to "merge" cold runs. I suggest collecting the data at the temps you want (-30, -20, -10, 0) (our algorithm is based on 0C). We can assist you in determining the best values to use depending on the results. 

    3) Program the result into a device and run a test to check accuracy. There are certain parameters you can use to make sure we are pessimistic for sure, but we will need alot more detail on the application to know which knob to turn. 

    If you are unable at step 3 to get it to look good (which I expect will be hard due to the -30) we might need to look into creating a custom chemID and i wll link you with our ChemID tech to do so. That process takes about 1 month (worse case) to complete once we receive cells.

    Thanks,

    Eric Vos

  • 0 in reply to Vos
    Intellectual 710 points

    Hi Eric,

    Thank you for you reply - I really appreciate your assistance. A couple of additional questions that have come up:

    1) Do the rest periods need to be exactly as long as specified, or could they be longer? (potentially several times as long) The series are long enough that it's hard to run them in a single day, and so may need to sit overnight at some points. Is this ok? (or do we need to fully automate these tests somehow?)

    2) Have you found it useful to run multiple cells, average their results, and then feed that into the characterization tool? It seems like that might give better results. 

    Thanks,

    -Eric Castle

  • 0 in reply to Eric Castle
    TI__Genius 10370 points

    Eric,

    1) Longer rest periods do not hurt us. You could always cut the log file if needed. The only issue would be if the log gets too large to manage (which would be alot)

    2) Yes i do find it helpful for sure. Law of averages is always better in engineering. This is better to make sure the cell you have isnt an outlier for some reason. I do generally expect very similar results, but always good to check

    Thanks,

    Eric Vos