bq27350 questions on learning cycle, RUP_DIS, Update status and the FC bit

Jenny,

I assume you have selected the correct chem ID already

1. make sure the FC bit is set after charge to full

2. When start the cycling to learn Ra, make sure RUP_DIS is cleared to start with. This bit can be cleared by FW once a well relaxed cell it detected (may take 30min to 2hr), or enable IT using a well relaxed cell to force the RUP_DIS cleared.If RUP_DIS is set during the cycle, it may indicate a large R update due to the learned value and the baseline R has large difference. Re-do the learning cycle may be needed. Qmax may be learned if two OCV have been read before and after the charge or discharge cycle with capacity change greater than 37% for normal operation (>90% for learning cycle).

3. Update Status bit set to 0x04 means IT is enabled. it will change to 0x06 once Qmax and R is learned. Once the golden image is generated, you should change the Update Status to 0x02 to indicate that Qmax and R has been learned (bit 1 is set) but IT is disabled (bit 2 is cleared). Once the golden image with the proper calibration data has been programmed in production line (no need to cycle every pack) the last step after cell is attached is to enable IT to make the update status = 0x06.

4. FC bit has to be set during the learning cycle. For normal operation, it is ok to not reach full charge. 

Let me know if you have more questions.

Thanks

Ming

Hi Ming,

Thanks very much for your answer.  Your answers to 3 and 4 have helped me a lot.  I have a further question on answer 2 though.

To confrim, yes the correct Chem ID has been selected.  I did as you suggested, charged the pack until the FC bit was set.  I left it over night, and in the morning the RUP_DIS bit had cleared.  I then discharged the battery and at the end of the discharge ( + 5 hours relaxation) I read the data flash values and saw that the RUP_DIS bit had been set again.  There were no Qmax or Ra table updates.

BTW, the RUP_DIS bit was clear during first learning cycle.  It was only when Update Status changed to 06 did I then start seeing the RUP_DIS bit change to a 1 after discharge.  I was wondering if perhaps this was normal? e.g. to reduce the amount of writes to flash maybe? or maybe my Qmax and Ra values are still not accurate enough.  In any case, to re-do the learning cycle, do I have to set the Update Status back to 04, or will a Qmax update occur regardless of whether the RUP_DIS bit is set or not?

Thanks and regards,

Jenny

1. what is the discharge rate?

2. is there any interruption during the discharge?

3. can you attached the log data?

4. to re-do learning cycle, I suggest to reload the original FW and load chemistry to start from fresh state.

Thanks

Ming

Hi Ming

1. The battery powers a 1W LED giving an average discharge rate of about 380mA, but varies from about 350 mA to 450 mA throughout the discharge.

2. The discharge is continuous.

3. I have attached a spread sheet which shows the data flash values after each charge and discharge cycle I have performed so far.

4. OK.  I have a feeling that my original Design Capacity value of 8800 mAhr is probably too far out judging by the Full Charge Capacity and Full Available Capacity values I have been seeing.  Perhaps I should change it to 8000 mAhr?

Thanks and regards,

Jenny

Hi,

I'm afraid that after several more cycles that I am still seeing no Ra table updates and I have more questions than answers.

1. I read somewhere that for the Ra tables to update, the discharge current must be greater than C/10.  Is that true?  My battery pack is a 4P, 8800mAhr battery pack but my discharge current is around 400mA which is less than C/20.  Does this mean the Ra tables will never update in this product?  Is there a fuel guage that can cope with these (relatively) low currents?  Can I trick the fuel gauge into thinking the current is higher than what it is by changing the Design Capacity register to say, 3500mAhr?

2. I have noticed that the VOK bit has not cleared after a full charge + relaxation time.  I have left the battery pack on the charger overnight with zero current going through it at a voltage of around 4114mV.  This should have provided plenty of relaxation time, but the next morning when I read the data RAM values, the VOK bit is still set.  Does this mean that an OCV measurement was never taken?  Why would this happen?  Would this prevent the Ra tables from updating?

3. I have noticed that the Ave I Last Run register has not updated since Update Status changed to 06 dispite the fact that I have charged and discharged the battery 6 times, and with slightly different discharge currents.  Does this have anything to do with 1) and 2) above?

Please help!

Thanks and regards,

Jenny

Jenny,

I checked the excel file you attached and confused by what is learned Ra value. When update status changed to 0x06, the Ra has changed a lot. This is step 7 at end of charge, Why Ra change from step 6 to step7 during the charge?

I'd like to see a data log for learning cycle if you have any. I am afraid what's happening is that the Ra learned is not valid so each time a new discharge starts, the actual Ra is far away from the Ra stored in data flash which cause the RUP_DIS being set to prevent Ra update.

I did not see anything else standing out. When you did the learning cycle, what is the discharge rate?

Thanks

Ming

Hi Ming,

I'm not sure why Ra changed after the charge.  This confused me as well.  Perhaps the fuel gauge re-scaled the Ra values after the new Qmax was learned?

So I assume that when you say the Ra value is not valid, you mean that the actual Ra value is greater than 3 x or less than 0.3 x the Ra value stored in flash?  I assume that the RUP_DIS bit stays set then for the rest of the entire discharge?

Here's what I did during the learning cycle: I charged the battery to 4.1 Volts not 4.2 Volts.  Since I have set my Taper Voltage set to 200mV, this ensured that the FC bit got set after the charge.  For the discharge, I used the application to discharge the battery which is a 1W LED discharging somewhere between 350mA - 450mA current.  Since this is an 8.8Ahr battery, the discharge rate is slightly less than C/20.

I am using the PCB from our application, not the evaluation kit to do the learning cycle thats why I don't have a data log of the learning cycle.  I will have to solder extra wires to the board to be able to log data as it goes.  But I can do this.  Do you think that the way I did the learning cycle with the lower charge voltage and low discharge currents caused the invalid Ra values?

Thanks and regards,

Jenny

Jenny,

first of all, you can use TI EVM to do the learning and it is preferred. The EVM has to be properly calibrated like the system board you are using. Data log is very  helpful.

The discharge current is kind low and the recommend rate is C/5. You still need charge to 4.2V to do the learning cycle as it is to learn the battery from 4.2V to 2.9V so the Ra and DOD is correlated. Then you can charge the cell up to 4.1V after learning cycle to learn the Qmax for 4.1V charge termination voltage. I suggest to keep the taper voltage to 100mV and taper current to C/10 or less but higher than the charger tape current.

Thanks

Ming

Hi Ming,

Something weird is happening.  The battery voltage is actually INCREASING in value when I attach a discharge load to the battery.  I noticed this as I was playing around trying to set up the data log.  The higher the load, the more the battery voltage is increasing.  I thought this might be a calibration problem so I recalibrated the fuel gauge using the EVSW a few times but this did not seem to work.  I also played around with the CC Offset and Board Offset parameters by manually changing them in the EVSW Calibration Tab.  Changing these values also did not seem to make any difference and my voltage value still increased under a current load.

What parameters would cause this to happen?  I have attached a log file which shows the voltages and currents with and without the load.  I have also included the .gg file for your information.

I suspect that this is the reason my Ra tables are not updating as this would indicate a negative impedance.

Thanks and regards,

Jenny

Jenny,

this may be caused by ADC ground selection. I have to check FW team to find out what ground is being used in bq27350 code. However, this should not prevent the Ra update. The gg file looks ok to me. When you run the learning cycle, did you log the RAM data? If not, can you re-run the learning cycle with C/5 discharge rate and enough relaxation time and log the dataRAM as well as auto export gg files? At this point we are just guessing around without the data log.

Thanks

Ming

Hi Ming,

I think you are right about the ADC ground selection.  I think this is what is causing the voltage offset under load.  I have taken a log of a battery pack under a 5 Ohm discharge load and measured the actual cell voltage using a multimeter measuring at the cell connections to the board.  I see an offset of around 44mV under load, and 0mV when there is no load.  Results in the attachment.

The Ra tables did update during the learning cycle, just not afterwards.  Anyway, I charged the battery to about 60% capacity, relaxed, then discharged.  Then I found the Ra tables did update.  (see Battery 1 tab in worksheet) I suspect that this ADC ground issue is worse at higher cell voltages because the cell impedance is less.  Also, since I have 4 cells in parallel, this would also accentuate the issue. 

I am still working on getting the data log on a new battery, but suspect this ground issue is causing my problems.  Is there any fix for this?

Thanks and regards,

Jenny

OK, thanks for your help Ming.

Jenny

OK, thanks for your help Ming.

Jenny