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.

BQ78350-R1: CEDV Data Collection for Gauging Parameter Calculator (GPC)

Part Number: BQ78350-R1
Other Parts Discussed in Thread: BQ78350, GPCCEDV, BQ76940

Hi,

I've done the 6 tests low/high rate + temperature required for CEDV and I've received the resulting report detailed below. From this report I've 2 questions:

1. It says that "if EDVV bit is set to 1... ". Which is this bit? I think that it's not present for BQ78350... So, have I to multiply EMF and EDVR0 values?

2. The reported value of EDVR0=94086, exceeds the maximum which is 65535. Which value I should set here?

Thanks,

Marco.

Detailed report:

GPC CEDV tool, rev=53
Configuration used in present fit
ProcessingType=1
NumCellSeries=1
CellTermV=3000
LearnSOC%=7
FitMaxSOC%=9
FitMinSOC%=3
ChemType=1
ElapsedTimeColumn=0
VoltageColumn=1
TemperatureColumn=2
CurrentColumn=3

CEDV parameters resulting from the fit. If EDVV bit is set to 1, EMF and EDVR0 have to be multiplied by the number of serial cells when written to data flash

EMF 51954
EDVC0 101
EDVC1 0
EDVR1 1
EDVR0 94086
EDVT0 5773
EDVTC 9
VOC75 3935
VOC50 3808
VOC25 3752


Recommended SOC deviation tolerance at EDV2 point is < 5% for low temperature and <3% for room and high temperature

Deviations for this set of parameters are given below for each file

file SOC error, % pass
roomtemp_lowrate.csv -1.85752477153744 1
roomtemp_highrate.csv -0.0948588443248655 1
hightemp_lowrate.csv -1.66730129255415 1
hightemp_highrate.csv -0.379089430102143 1
lowtemp_lowrate.csv -1.68069894084773 1
lowtemp_highrate.csv 1.65026127901974 1

Deviations are within recomended range. CEDV parameters are suitable for programming the gauge

CEDV parameters resulting from the fit. If EDVV bit is set to 1, EMF and EDVR0 have to be multiplied by the number of serial cells when written to data flash

EMF 51954
EDVC0 101
EDVC1 0
EDVR1 1
EDVR0 94086
EDVT0 5773
EDVTC 9
VOC75 3935
VOC50 3808
VOC25 3752

  • Hi again,

    Enclosed you can find the "package" uploaded in GPC Tool online and the report returned.

    4578.GPCPackaged.zip3021.GPCPackaged-report.zip

  • Hello,

    I have fixed the point number 2 of my previous question. It didn't introduced the proper "Number Of Series Cells = 14". Point 1 is still open.

    Meanwhile, I've continued my testing without multiplying EMF and EDVR0 value by the number of series cells. So, I've introduced the parameters as they has been reported. Afterwards, with the battery pack fully charged, I've done a "reset" and I've executed a discharging test monitoring all registers, specially RSOC and ASOC. In this test I saw that EDV2 reached when all the 14 cells where around 3,765V (note that nominal cell value is 3,7V). Contsequently, RSOC and ASOC were updated from 38% to 7%. 

    Note that I've set the values of EDV 0, 1 and 2 Hold Time to 20s, in order to be completely sure that any kind of noise in cells measurement could affect for premature EDV2 detection. 

    Do you have any idea about that is going wrong?

    Enclosed you can find he reported CEDV parameters: 5282.GPCPackaged-report.zip

    And here you are the log file of discharging test executed afterwards with a graphic result focussed on RSOC: W005_Test#001_EvalParams2o_Package.log

     

  • Hi Marco,

    The EDVV bit does not apply for the BQ78350. This parameter is present in some other TI battery gauges, but you can ignore it for this device.

    I am looking at the log file, and it looks like the PendingEDV column is showing 3665mV - this is what the gauge is watching for to trigger EDV2. It seems a little high since EDV2 should correspond to 7% SOC. Around 100 seconds before the EDV2 is triggered, it does look like the Cell 10 voltage dips by around 100mV for around 20 seconds. It doesn't seem to get as low as 3665mV, but it is pretty close. I wonder why only this voltage is dipping - do you have the recommended 10uF and 1k filter components on the VC10X and VC5X pins?

    I also recommend setting the EDV_EXT_CELL bit in CEDV Gauging Configuration. This will use the average cell voltage for triggering EDV2. It may also be good to set the VFLT_EN bit in the same register. There is also an Overload Current parameter that can be used to ignore large short pulses.

    I looked at the report from the GPCCEDV tool and it looks like the error is pretty large. Can you send the log files so I can take a quick look at what might be causing this?

    Best regards,

    Matt

  • Hi Matt,

    I'm going revise on the hardware about the recommended 10uF and 1k filter components on the VC10X and VC5X pins.

    Meanwhile, enclosed you can find the files used to generate the report. How could I reduce the "deviations for this set of parameters given" by the report? There is a deviation of 9,48% in lowtemp_highrate test. I've tried to reduce this deviation by repeating the test, and I've obtained similar deviation result.

    GPCPackageFromBqStudio.zip

  • Hi Marco,

    I see some issues in the log files. One issue is that the current turns off and on during the log (it appears the under-voltage protection is triggering which is opening the DSG FET). Another issue is the the logs are not all reaching the cell termination voltage you defined (3000 mV). They are stopping much earlier. 

    You want to make sure your protections are configured a little looser while collecting these log files. The log files are showing the characteristics of the battery cells themselves, so the gauge is not even required - it is just helpful to log the voltage, current, and temperature. 

    I also notice the External Average cell voltage is not showing correctly in the log files. You may need to calibrate this parameter to get the correct reading.

    Best regards,

    Matt

  • Hi Matt,

    Thank you so much for your answers.

    First of all, as you say, the values of 10uF and 1k filter components on the VC10X and VC5X pins were wrong. I've corrected them. 

    Moreover, I want to change the cell connection for BQ76940 following the recomendations of table 26 of bq76940.pdf datasheet. So, I've changed prematurelly AFE Cell Map in Data Memory according to the future change. Contsequently, after writting it to data memory I've lost the SMBus communication. I suppose that the device is in permanent fail. Is there any way to come back to previous configuration? As I'm not able to communicate, also I cannot program a new software.

    Regarding, the log files. After these HW changes I will repeate the 6 log files following your advices. 

    Thanks!

  • Hi Marco,

    When the AFE Cell Map is programmed incorrectly, it can be challenging to recover the device. Some of the cells configured will be at 0V which will trigger a voltage based Shutdown. You need to wake the device again with a rising edge on the TS1 pin, but then quickly change the AFE Cell Map register before it shuts down again.

    Regards,

    Matt

  • Hi Matt,

    I've repeated all 6 log files high/low temperature + high/low rate with the proper RC filter in VC10X and VC5X + the recommended 14 cell configuration of table 26 of bq76940 datasheet. Moreover, I've relaxed the protections in these 6 log files, setting for example UV=2,7V or OTD=65ºC, and I've calibrated External Average Cell Voltage as you recommended.

    The reported parameters show a bigger deviation results, with the next note: "Accuracy Error: Deviation is so high that it is most likely due to anomaly in the data. Please check that data files have recomended format, units and test schedule".

    I've checked all 6 data files, and I don't see any anomaly in the data. 

    Could you give any idea about the reason of this deviation?

    Furthermore, I've one question: If I enable EDV_EXT_CELL = 1, the calculated EDV0,1,2 by CEDV algorithm will be based only on the External Average Cell Voltage?

    In other words, if the External Average Cell Voltage is around 3,5V (which could means 7% of battery) but the lowest cell voltage is near of UV=3V. How it does affect to RSOC calculation? Would I see a jump from 7% to 0%? In this case, it seems better approach using the lowest cell voltage (not EDV_EXT_CELL=1).

    For example, from the 6 log files that I've collected I've seen that cell7 reaches to 3V when the ExtAvgCellVolt=3,3V. It seems that cell7 has less capacity than others.. but it can occurs obviosusly because production process or aging. I've also checked the cell 7 balance during the charging process, and it's totally ok (maximum deviation of 20-30mV respect to the others).

    Enclosed you can find the 6 log files and the reported parameters. 

    Package_Original6LogFiles.zip

    1351.GPCPackaged-report.zip

  • Hi Marco,

    Can you send the zip file that you submitted to the GPCCEDV tool? I don't see anything obviously wrong in the 6 log files, so I want to check that the files submitted do not have any errors.

    Thanks,

    Matt

  • Hi Matt,

    Here you are the requested zip file submitted.

    Thanks,

    Marco1513.GPCPackaged.zip

  • There was no particular issues with the data-files, but fit can be sometimes improved by trying different fitting range.

    For this data-set I got the best results with these ranges:

    LearnSOC%=7

    FitMaxSOC%=10
    FitMinSOC%=6

    I attach the report file below.

    5700.GPC_report.zip



  • Hi Yevgen,

    Thank you for your response.

    I've checked your reported file and I've seen that the parameter value EDVR1=78467 is out of range. The max value acceptable is 65535. Are these parameters still valid? 

    Anyway, I've run a test with these parameters setting EDVR1=65535. During the 1rst discharge there's a RSOC jump from 12% to 7% (calculated EDV2=3676mV). This EDV2 point is above the "knee" of the discharge curve and it affects to the new calculation of FCC which becomes updated to 1,96Ah. The nominal values of these 14 cells that I'm using are 3,7V and 2,1Ah.  Later, the calculated EDV1=3653mV and EDV0=3598mV. So, the batteries remains at 0% too much time until UV occurs at 3V.

    Questions:

    1. After introducing the parameters, adjusting "Dod at EDV2" and executing a reset. Is it required a complete cycle from "fully charged" to "fully discharged" to update FCC and obtain in the next discharges a better RSOC% calculation? Is it mandatory for a good RSOC%? I'm wondering if this process should be added in the future production process to send batteries with good RSOC% estimation.

    2. How I could improve the RSOC% results explained before? 

    Enclosed you can find data graph of the discharge and charge process.

    Thanks,

    Marco.

  • Marco,

    you are correct that the value shoudl be limited to 2^16-1, this normally happens naturally but this battery is has unusuall high impedance.
    Anyway, I was able to re-fit the values with limiting the max, and get a good fit. Try using attached parameters.

    To you other question - CEDV only learns at EDV2 threshold, so it is expected to have a jump at first discharge cycle, or if temperature or rate conditions have

    changed. Some customers to implement learning cycles in production to learn FCC to reduce initial jump, but as I mentioned, change of temperature conditions will change a jump as well, so learning is only justified if yo also like to use learned FCC for some quality  control purposes. 
    Unfortunately it is not normal behavior for CEDV, and you would have to go to an IT gauge to make adjustment of FCC happen earlier.

    Regards,

    Yevgen

    GPC_report2.zip

  • Hi Yevgen,

    Thanks for your response.

    I've tested the new values and they are quite acceptable. But I've seen that the calculated EDV0 = 2943mV which is lower than UV = 3000mV. Is there any problem if EDV0 < UV ??

  • Hi Marco,

    Normally the under-voltage threshold is significantly lower than EDV0 (the default is 2500 mV). It is okay if it is higher, but the protection will trigger before the SOC reads 0%. Since FCC learning takes place when EDV2 is triggered, it will still learn the new capacity as the battery ages.

    Best regards,

    Matt