Hello.
I trying use bq34z100 with 4-parallel NCR18650 Panasonic (2900ma) battery set, but after testing got very bad results. Please check my records, maybe you can give me recommendation for improving working with this ICs.
At first I set config for my battery specification and received golden image (file "config.gg"). Here three graphs of Remaining Capacity prediction in calibration progress when discharging (graph1).
![]()
![]()
As you can see, the last (3) curve has a very good precision, if draw a line for the end (graph2 is zooming) we get about 1.1% accuracy in overall. But shape of the graph is not very good for real use, because it have unknown step-fluctuation (with sharp jumping of % capacity about 2%, see graph1). Besides, prediction capacity is over before voltage reached minimal value. Please see graph3 with end of third calibration curve with imposed voltage curve. As I understand, prediction capacity must be related with current voltage according chemical table (for my type of battery NCR18650, wich be uploaded before calibration), but correction too juddering, even it is.
![]()
Here measurements on a less current ~1.3A, results not bad (see graph4), exept step-down at the end.
![]()
Here measurements on a larger current ~4A (see graph5), and results not very good - curve of remaining capacity prediction fall down on about 9% earlier of real capacity. And again we see a several big step-fluctuation (as I understand this is may be a fault of comparing with chemical table).
![]()
And next battery prediction testing at different from calibrated temperature on about -16C. Results very poor. (see graph 6) Curve has giant steps without any correlation with real capacity.
![]()
However, once I noticed, that counting on charging is not work correctly at the end of charge. Capacity counting just stops on certain value and not count some % of energy. Its happend while charging battery after it was fully discharged at -16C temperature, then heated by the room temperature with restoring remaining capacity to about 2000ma. See graph7.
![]()
So, as I see capacity prediction is not work correctly, but cant find reason for this.
Q1. Whether it is possible smoothing graph of remaining capacity prediction to eliminate step-correction of prediction?
Q2. Is it possible eliminate graph break at the end and strech it according minimal voltage value?
Q3. How to reach good working bq34z100 with my battery on terms different from first calibration (20C, 2A)?
Q4. Why there was a failure of capacity counting on charging?
PS: Im sure that chemical of my battery is the same as I choose in chemical selection list (NCR18650 2900ma code 0238), temperature probe has good thermal contact with battery, was placed between nearby elements and is correctly type (103AT 3435K), all elements brand-new and fresh. I attach photo of my testing setup and full log in four files of all my tests.
I will be grateful for any advice.
With best regards, Viktor.
(Please visit the site to view this file)
(Please visit the site to view this file)
(Please visit the site to view this file)
(Please visit the site to view this file)
(Please visit the site to view this file)
![]()
