Most popular systems are based on 12V or 24V. The new BMS allows up to 8s cells (based on ISL94202) and covers these systems in one PCB instead of previously two completely different designs with either 3-5s or 6-10s chips.
The previous designs needed two PCBs, which is more expensive to build. Now the power and control part are included into one single PCB.
The bq76930 and -40 ICs have some issues when all cell channels are used, as the balancing and voltage measurements are not synchronized inside the chip, resulting in wrong measurement values.
Good connection of heat sink from the bottom is now possible
The firmware support for the IC is still in progress. I’m currently also playing around with Zephyr OS instead of mbed OS. The ultra-low-power MCU STM32L452, which can be used instead of the STM32F072 in the new BMS, is not supported by mbed and I didn’t get the custom board definition to work. In Zephyr OS, everything works fine and I can talk to the ISL94202.
If you have any questions or feedback, feel free to post here or in the github issues!
Very nice design Martin!
A few years ago I’ve made one BMS with bq76930 and bq78350, but board is quite big and there are some thermal issue. So I will try to build this one. I have one free set of 8- 15Ah LiFePO4 cells for testing
As first step I’ve opened design files via Kicad and there are some minor issues with schematics - some symbols are missing (ISL94202, TPS560430X, UEXT connector) and there is no bms-8s50-ic-rescue library with the project. I’ve checked also your LibreSolar symbol lib from GitHub, but current version is quite old. Of course there is full pdf version of schematics, but it will be nice to have a complete schematics in Kicad also.
Thanks for letting me know. I didn’t realize I forgot to commit the symbol cache, as on my computer it obviously doesn’t complain. With the correct cache file, the rescue-files are not necessary anymore.
Updated this in the BMS repository. In addition to that, I also updated the general LibreSolar symbol library. You are right, it was quite outdated…
I’ve updated LibreSolar symbol lib and now every thing is fine. Thanks.
Also checked LibreSolar footprints lib and there is no missing parts regarding this design.
I’d like to ask you about the issues you mentioned above with the bq76930 (and -40) ICs. Is it only when all 10 cell channels are used that you find incorrect measurement values? If one only used 8 cells, do you know whether the synchronisation of measurement and balancing would be ok?
Yes, up to 8 cells are fine, because the cells at the boundaries of the internal 5-cell blocks are not used. I’m using the bq76930 in my system with 8 cells aswell.
