Hello,
We're currently designing a battery pack with the bq78pl114S12 / bq78pl116. (Future redesigns with bq78pl116)
The pack consists of 4 LiFePo4 cells in serie. The cells are "LP44XXXXX" 70AH-cells from LISHEN.
As we want to run the pack with higher currents than 320A we used a 100 microohms shunt, but configured 4000 microohms in the tmap-file. With this adjustment all current-values are divided by 40. Later we multiply the values again by 40 in a external microcontroller.
As for the cemistry we're using the ID 413 as those cells have nearly the same capacity as our cells divided by 40 (70000mAh / 40 = 1750mAh, 413-Cells have 1700mAh)
Now to my Problem:
I've done the nessesary configurations and measurements for a .aux-file generation and sucedded in generating the file.
I've set all the parameters as they should be in the final application and loaded them onto the bq together with the .tmap and the .aux file. I've also done the temperature and current calibration
At the first charge/discharge cyle everything worked fine and the RSOC was calculated correctly. But with increasing cycles the RSOC started do drift upwards and showed 100% long before the pack was full and stayed there longer each cyle.
My question now is, why the RSOC drifts and how I can prevent that drift?
1.) Is it happening because of the wrong chemistry? (I've also made tests wit the chemistry 411 and had the same problem) If yes, can we send in some cells to have a proper chemistry-file generated? If yes where?
2.) Or does it happen because of an error in the.aux file generation?
I've attached the log-file for the .aux-file genaration, the.aux file and a log file witch showes the RSOC-Drift. I've also added the ppcsv file for both logs (its the same)
Thanks in advance
Cedric
1411.LogsRSOCdrift.zip
Cedric,
The sense resistor and hardware gain will need to be configured to support the current measurement range that you need. You may not be able to configure the IPScale to pass with this setup and you can use the uC to scale the data reported by the SMBus.
There are a few things that can cause RSOC drift.
+ The PL114 has a problem where RM and RSOC may not report correctly due to a delay after entry in the Charge state from a fully discharged state. This will normally correct itself during or at the end of charge through. This was fixed in the PL116 firmware.
+ The cell chemistry can cause the problem as well. If the cell does not match the ChemID, then the simulation will be off. If the ChemID is close, then the device should be able to trim its' internal parameters after several Charge-rest-discharge-rest cycles. You mentioned that you are using LiFePO4 cells, so the rest period could be up to 12 hours to allow time to make good estimates.
+ The rest time during the aux file generation also need to be extended when using LiFePO4 cells.
One other parameter to consider when using these cells is the minimum cell voltage. The PL114 and PL116's measurement systems are not guaranteed to work below a 2.75V cell voltage, so make sure that your parameters are setup properly to meet this requirement.
Best regards
Tom
Tom,
Thank you for your answer.
+ I was able to resolve the drift problem to some degree. When cycling the pack with 8h rest time the RSOC still drifts a little, but adjusts itself in some cycles after the OCV Idle Qualifier has elapsed. The RSOC still drifts when cycling with a short rest-time afterwards.
I'm currently doing the measurements for the aux-file generation with long rest-time and hope it will solve the problem completly.
+ To ensure that the minimum cell voltage is always higher than 2.75V, I've set the CUV Threshold to 2.8V (It has been so before)
Regarding the .chem files I'd have another question. Does it only contend a SOC look-up table depending on OCV and Temperature or does it also contend information about the size (Ah) of the cell? Or even more? As the file is coded I can't find out myself.
+ As described we are using an uC to scale the current measurements to the correct value, but this factor may still vary during the development. If a chem file contends information about the size of a cell or it's impedance values we would need a separate .chem file not only for each diffrent cell, but also for each diffrent uC-Scale-factor. Correct?
+ What would you need in order to generate a .chem file for our cells? And what would it cost?
Hi Cedric,
I have browsed Lishen's web site and I can't find any mention of a 70 AHr cell, nor could it find the part number LP44147172. I would be quite interested in such a cell. Can you provide me with any more information?
Thanks
Bob White
Hello Tom,
I did find out, that my "Transition to Idle Current" - Value was wrong and the bq78pl114 never went into the idle-state.
After changing the value, the bq78pl114 did go into idle-state and now I'm experiencing the same problems described in
http://e2e.ti.com/support/power_management/battery_management/f/180/t/134325.aspx
This convinces me that my ChemID is definitely wrong and I need one made for the cell.
Thus my Question from above about the contents of a chem file is now even more important:
Cedric Gasser Regarding the .chem files I'd have another question. Does it only contend a SOC look-up table depending on OCV and Temperature or does it also contend information about the size (Ah) of the cell? Or even more? As the file is coded I can't find out myself. + As described we are using an uC to scale the current measurements to the correct value, but this factor may still vary during the development. If a chem file contends information about the size of a cell or it's impedance values we would need a separate .chem file not only for each diffrent cell, but also for each diffrent uC-Scale-factor. Correct? + What would you need in order to generate a .chem file for our cells? And what would it cost?
@ Bob White
We are currently developing this cell together with Lishen and it is not yet for sale and therefore not on the website.
The ChemID could be your problem. It contains OCV and Ra data for the cell. If these do not match your battery, then the simulation estimates will be incorrect.
We can characterize your battery and create a ChemID for it. The service is free and you will need to provide 2 or 3 cells and the battery datasheet. The process normally takes about 6 weeks to complete.
We are also reworking our procedure for aux file data collection for LiFePO4 cells. These cells require much longer rest periods than the Li-Ion cells that the procedures were developed to support. We are probably going to recommend 7 to 12 hour rest periods between discharge and charge and charge and discharge.
Best regards,
Thank you for the quick Answer. I've already made a Phone-call-back request on the TI Customer Support Center to clear the details with shipment etc. Or can you support me with a direct Phone-number to clear this details?
Additionally i have the folloing points regarding a .chem file
+ We could support detailed measurements with diffrent charge and discharge currents at diffrent temperatures and also impedance measurements of our cells. Could you create a .chem file out of detailed mesurements? Could this reduce your workload and the time needed to create the file?
+ Would it be possible to adjust the .chem-file to our Current-scale-factor. As far as I've understood it the only thing to do would be to multiply the Ra values of the cell by the current-scale-factor.
Our cell characterization models the Ra-table based on several current measurements. The device uses this table and measured current to determine the FCC and remaining capacity based on Qmax. We will need to characterize the cells to setup this data.
RegardsTom