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.

BQ27427: State of Charge at beginning of Charging is Overestimated

Part Number: BQ27427
Other Parts Discussed in Thread: BQSTUDIO, BQ27Z746, GPCCHEM

Tool/software:

Hello,
we have a design, that makes use of a BQ27427 with a lithium-ion polymer battery from Renata (ICP521630PJ-05), connected to a microcontroller (nRF53) that runs Zephyr-OS (and makes use of their fuel gauge driver).

When I fully charge the battery and then slowly discharge the battery (over the course of a day), the state of charge drops slowly until 5%. When the microcontroller detects the SOC of 5%, it disconnects the battery and the gauge using a PMIC (PCA9420BS). The gauge is not put into SHUTDOWN mode.

When starting to charge the battery, the PMIC connects to the gauge and battery again. The first SOC, that we get from the gauge is 6%, which seems to be very reasonable. But the next values seem to be way to high, after a few seconds, the next reported SOC is already 59%.

Indeed, stop charging, the reported SOC drops slowly and a last reported SOC is then 43% and the device brown outs.

The Charging Voltage of the battery is 4.2V, so we picked 1202 as Chem ID.

What can we do, to adapt the gauge SOC estimation better to our battery?

best regards

Torsten

  • Hello Torsten,

    Today is a national holiday, and we will work on your ticket in the upcoming business days.

    Thank you,
    Alan

  • Hello Torsten,

    Have you completed the learning cycle on the gauge? I will link a document that talks about this: https://www.tij.co.jp/jp/lit/an/slua903/slua903.pdf

    Regards,

    Adrian

  • Hi Adrian,

    I have to admit, that I'm totally unsure and confused as to what we have to do, to increase the accuracy of the gauge. The data sheet does not even contain the word "learning" and promotes the BQ27427 as a battery fuel gauge "that requires minimal user-configuration and system microcontroller firmware development, leading to quick system bring-up.".

    The Zephyr driver does not even have any mean to apply a golden image, which mean, we would also have to change the driver or even write our own.

    Can you confirm please, that we have to go through that golden image creation cycle to get decent accuracy with the BQ27427 gauge?

    best regards

    Torsten

  • Hello Torsten,

    Yes, there are configurations that need to occur for the gauge to be accurate with SOC calculations. I would recommend reading over this document to see the parameters that need to be changed: https://www.tij.co.jp/jp/lit/an/slua903/slua903.pdf

    Regards,

    Adrian

  • Hello Adrian,

    I got my hand on a Windows Laptop and installed bqStudio on it. Now, I'm asked to select a target. The BQ27427 is not in the list of supported targets. Does it mean, we have to develop our own measurement software to run the learning cycle? Or is there an alternative device that we should pick, that is very similar to the BQ27427?

    Edit: Looks like support for BQ27427 was added to the untested Test version of bqStudio :-/

    Edit: I've order the required hardware now. What is still unclear to me, is how to apply the resulting `golden image` later. If I got it right, the golden image will describes some volatile Data Memory. The BQ27427 Technical Reference Manual mentions the term `gold image` two times: One (7.1.1) mentioning that there are some Data Memory locations that will become part of a golden image and a second paragraph (7.4.3.1) describes the ability to update the resistance data as the only reason, why they are displayed.

    What I would need to know is, how this golden images are formatted and where they have to be written to.

    best regards

    Torsten

  • Hello Torsten,

    You should use the Test version of Bqstudio.

    Regards,

    Adrian

  • Hi Adrian,

    when picking the ChemId in bqStudio (Test-Version), there are no batteries from Renata. An attempt to update the ChemID data base, by running ChemicalUpdater failed with the error message, that no installation of bqStudio could be found. Can this become an issue for us? (I would pick ChemID 1202 otherwise).

    Could you please elaborate as how we would later have to apply this golden image to the BQ27427?

    best regards

    Torsten

  • Hello Torsten,

    The BQ27427 only has 3 pre-programmed ChemIDs that can be applied on the gauge. If you have a specific ChemID you would want to use you will need to choose a flash based gauge like the bq27z746.

    Regards,

    Adrian

  • HI Adrian,

    I think, I managed to start a learning cycle. After 24h the battery was discarded. Unfortunately, the software stopped to do anything, it just showed the last measured voltage. I left the system in that state for > 10h but there were no changes (see screenshot).

    Even as the status diagram looks like, we were going through a full relax charge relax cycle, the state jumps very quickly (with in a few seconds) to the discharge state.

    Is there something obvious, I'm doing wrong or that I'm missing?

    We picked the BQ27427 due to the advertised easy integration and the "Ultra low power consumption". We got a little bit surprised by all the hassle we are experiencing now we have to make a hardware decision at the end of next week. Is the some kind of list of batteries that are compatible with the BQ27427?

    I there any chance that we could get (paid) engineering support from TI to find a suitable fuel gauge that works reasonable accurate with our battery?

    best regards

    Torsten

  • Now, I managed to start from the beginning. The first relax cycle finished and I started the Charge cycle after an hour, the battery was fully charged, but does not reached the nominal Charge Voltage of 4200mV, but only 4181mV. So bqStudio still thought, the device would be charging. I left the system in that state for > 3 hours without any change. The Cycle State showed: "Charge Mode - Connect charger now." for the entire time.

  • Hello Torsten,

    We don't have a battery list that is compatible with ChemIDs that is already preprogrammed onto the gauge. You can use the GPCCHEM tool to see which battery chemistries will be a close match to your battery. However, for now I would continue using ChemID 1202 and try to complete the learning cycle. Afterwards, we can assess the SOC accuracy. 

    I would follow the procedure but not use the BQStudio plug in. You can do the learning cycle process without that.

    Regards,

    Adrian

  • The description of the "Learning Cycle" starts with 3.1. "Initial Qmax Update Criteria". According to the documentation (SLUA903), we should send a GAUGE_EN command. I can not find any way to send such a command nor does the reference manual mentioning such a command. Is there an alternative command, we have to issue? I want to prevent us from loosing another day just to find out at the end of the discharge / charge cycle, that the learning was not successful.

    Please keep in mind, that I am (as a software engineer) not an expert in battery gauging (nor are our hardware engineers).

    Edit: I discharged the battery, disconnected charger and load and waited. I waited for ~6 hours now, the voltages was at 3137mV for over 60 minutes. Now, it looks like the voltage is even dropping (3136mV). Still, the RUP_DIS bit is still set (I assume, that the RUP_DIS flag is the equivalent of of the RDIS flag described in SLUA903. Is the battery now relaxed or not?

    Sorry, but I need somehow _precise_ instructions to perform the required "Learning Cycle".

    A desperate customer

  • Pleas find also a screen shot showing the current status attached:

  • Hello Adrian,

    is there any chance, that we could do the learning cycle with the BQ27427 and BQStudio?

    best regards

    Torsten