so this is related to my other question if anyone wonders. I’m building out a EV battery pack to boost charge a vehicle in real time during heavy duty cycles. The vehicle currently has no external charge capability. It runs on internal battery pack with peak voltage at 266v DC and can go as low as about 35% of that around 100v DC (it’s not linear it seems).

I have built a few prototype battery packs capable of the 260-270v DC. The challenge is cell balancing and finding a BMS capable of doing the job in different configurations. Tesla uses 24v packs and then connects them together (serial/parallel?). I have a few thousand cells currently to work with.

The largest BMS units I can find get very expensive if going to say 72s configurations. At 14s they aren’t too expensive but then how to get up to the 260-270vDC.

I’d like to get some perspectives and would definitely like to hear of anyone who has conquered this for similar EV, BikeEV, MotoEV projects. Obviously the parallel aspects of each bank would increase total KW. But how many ’s’ in each bank is my challenge. The flip side is once the pack is designed it has to be able to be charged from a J1772 charger too.

The basic question is: What is the most economical configuration of Ns cells to get both a reliable BMS and up to the 260-270v DC?

  • \$\begingroup\$ I don't think there's any one answer. Breaking it into individual packs that are 48V or less is probably a good idea, because there's safety regulations that kick in when you get much above that. Breaking it into individual packs of no more than 50 pounds or so (or whatever you want to make people lift and carry) is probably also good. Were I doing this job, I would find a way to have each pack manage its own charge, and to communicate with other packs as its doing so to help in keeping the charge balanced with a minimum of Joules burnt up in resistors. \$\endgroup\$ – TimWescott Feb 2 at 22:21
  • \$\begingroup\$ Yep, kind of along my thoughts too. Once you get over 48v, the choices get much fewer for good reliable and cheaper BMS systems. \$\endgroup\$ – EngMarc Feb 2 at 22:25
  • \$\begingroup\$ Certain features of the BMS make more sense to implement at the sub-module level (such as cell balancing, etc). But if over-voltage occurs in one submodule, I think it makes more sense to assert a common fault signal to a master control which then opens up the circuit and prevents any further charging from occuring. If you follow the typical BMS practice of having charge and discharge FET's in each sub-module (12S or 6S sub-module), those FET's need to be rated for the full system voltage, and that is kind of a big deal. \$\endgroup\$ – mkeith Feb 2 at 22:33
  • \$\begingroup\$ Yeah, that was the challenge I had in the past with an anticipated 4s72p pack and 14 packs. I couldn’t find any reliable and not unbelievably expensive BMS solutions. One could spend a few hundred thousand on R&D but that’s beyond this project scope. Ideally, I’d like to design a system that uses as much off-the-shelf products yet achieves the design goal of boost charging on demand. \$\endgroup\$ – EngMarc Feb 2 at 22:49
  • \$\begingroup\$ like these -> infineon.com/cms/en/product/power/mosfet/… \$\endgroup\$ – EngMarc Feb 2 at 22:51

Aliexpress has 32s (150V max) BMS at up to 300A (120A continuous) that uses a microcontroller for any battery chemistry and bluetooth/Android smartphone app to set the cell voltage, max current, etc. Personally I'm using it for 128V battery sections (lipo lasts far longer when max charge is 4v) and then series those through isolated output buck-boost converters. I'm only using 5 sections for 700v max output (shutoff voltage of battery is 105.6v each section/3.3v per cell) but I imagine the total is only limited by the insulation of the isolation transformer of each converter (and how many mortgages you want to finance). While total costs can get pricey (several thousand $ not including cells) that BMS is quite reasonable at ~$150. Need more current? The board uses 30x cheapish TO220 fets and can be replaced with better 100A ones (also TO220) from Digikey for about $100 (600A continuous/2500A 10s with a better heatsink, heavier power wiring and bus bar, etc assuming the cells are rated for that).

  • \$\begingroup\$ This seems more like discussion than an answer, which is not the purpose of this site. (The questions itself probably does not fit here either). Further, using multiple low-rated BMS in series would be unsafe as the switching element are then exposed to the total voltage which likely exceeds their rating. It sounds like maybe you are hiding them behind converters, but the same issue then applies to the converters - are they actually rated for exposure to the the total sum voltage? Probably not. \$\endgroup\$ – Chris Stratton Feb 27 at 18:44
  • \$\begingroup\$ Actually they are, 1700V SiC fets in them in addition to the ~2000v isolation rating of the output transformer (built it myself). Must have done something right, hasn't been a spectacular explosion yet after several hours of run time. \$\endgroup\$ – user208607 Feb 27 at 21:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.