LiFePO4 configuration for MAX17320

jcole97 

I saw that you were able to provide an INI file in this similar post here: LiFePo4 configuration for MAX17301 

Are you able to help me create a configuration file with a model for the type of battery we are going to use in our product?

Thanks!

jcole97 

I saw that you were able to provide an INI file in this similar post here: LiFePo4 configuration for MAX17301 

Are you able to help me create a configuration file with a model for the type of battery we are going to use in our product?

Thanks!

Hello,

I can help create a starting INI file for you.  Can you tell me the Rated Capacity for your cells and the configuration of your application (series and paralllel)?

Thank you for the quick response! 

We are using 3.2V 3800mAh cells with 3 series rows of 11 parallel cells (33 LiFePO4 cells total).  That would make the total rated capacity of the pack 41.8Ah.  

Thanks.  I also will need what size sense resistor you are planning to use.  For a 41.8AH battery, I would recommend a 2mOhm sense resistor and how many thermistors you plan on using

Assuming 1 Thermistor and a 2mOhm sense resistor... attached is a starting INI file for youMAX17320_LiFePO4_3S_41AH_1Therm_2mOhm_061824.zip

Thank you very much!

Your assumptions align with our design choices.  I'll try out this configuration soon!

I tried out the configuration you provided, but I am not seeing 'RepSOC', 'RepCAP', 'TTF', or 'TTE' values which are accurate.  The Fuel Gauge appears to be grossly underestimating the capacity of the battery (only reports 10% the expected capacity).  Are these values expected to be accurate right away?  Or does it take several cycles of charging/discharging before they become accurate?

I was also hoping that you could shed some light on how you chose values for the 'nQRTable20', 'nQRTable30', 'nRComp0', and 'nTempCo' registers since the datasheet does not provide any details on the format for these registers.

Thank you!

Can you provide a datalog from the EVKIT GUI showing the inaccuracy?  If the inititial voltage measurement is in the very flat portion of the LIFeP04 discharge curve, there could be some initial error that needs a few cycles for the fuel gauge to get syncronized with the battery.  It is recommended to charge to full, and allow the battery to relax for 45 minutes.  Then discharge to the empty voltage and again allow the battery to relax for 45 minutes.  After 2 cycles like this, fuel gauge should be ready to go.  This can also be accomplished with partial charge/discharge cycles.

Attached is a couple of Datalogs from when I was testing yesterday as well as the INI I used.  Please note that I had to go through the config which you provided and change several parameters since we only have a 3300 mAh 3S1P battery currently, which is connected to the Dev Kit with the 5mOhm resistor.  

MAX17320_Datalog_INI.zip

I also wanted to reiterate my second question from before.  How did you chose values for the 'nQRTable20', 'nQRTable30', 'nRComp0', and 'nTempCo' registers since the datasheet does not provide any details on the format for these registers?

I'll take a look at the data, but so as not to forget your other question again... all of the model values and these settings are picked based on the most common LiFePO4 cells we have tested.  These are not intended to be fully explained/understood by the customer, so they are not explained in much detail in the datasheet.  However, these settings are important for the performance of the battery to control the voltage fuel gauge behavior, empty compemensation over various temperature and loads, etc.

A quick look shows FullCapNom is 416mAH, compared to 3300 for the FullCapRep/DesignCap values. 

The initial RepCap value comes through a chain of calculations, but is directly impacted from the initial FullCapNom and the initial voltage reading.  If the FullCapNom is too low, then the RepCap is going to be too low as well.

The INI file looks correct, but somehow in the datalog, nFullCapNom shows as 0.

I'd recommend loading the INI file again and confirming if the nFullCapNom is written correctly and then that FullCapNom is also initialized to that same 3000mAH value.