Central or Distributed?

Now that I have made some progress on the MPPT controller, I have been thinking ahead towards the BMS.  At 1st I had in mind to simple create a Battery MONITORING System, as opposed to a Battery Management System - with perhaps the biggest delta being if one is able to recognize, and actively rebalance batteries who's cells have drifted apart.  (Perhaps most commonly seen in LiFeP04 deployments).

This extra capability requires some attachment to each cell in a battery, and there are two ways to do it:

1. Have a master-controller, and then on each cell place a module to measure and report back status.
2. Still have a master-controller, but instead of communicating with individual cell monitoring/management devices, have a centralized one co-exist with the master-controller. 

As always there are ++ and -- to each approach.  The Distributed approach keeps the cost of the master-controller down when there is no need to actively monitor / manage at a cell level.  Perhaps only some extra logic is needed to provide an 'expansion' port.  But the distributed system might be a bit more costly, as it would require a small uC at each cell - and there are those who comment that placing something on each cell will also, over time, cause the cells to go out of balance due to slight variations in the cell level boards...

The centrally located one is perhaps a bit cleaner, as only one (or two) wires need to be routed to the junction between each cell - as opposed to a small board and some communication wires for the distributed option.  And one could do an expansion board for the centrally located master as well, or just integrate the logic into the master and perhaps make it a depopulation option....


I think this comes down to two approaches for this project.  A Master controller with CAN, overall battery voltage and  current monitoring and then either:

• A common Cell module which can monitor voltage as well as temperature - communicating back likely over a LIN serial network.  --OR--
• An expansion board which attached to the top of the master controller and support 6 batteries.  Communicating perhaps via the SPI bus, and only able to monitor voltages of each cell (no temperatures).  Should have the ability to add up to 4 expansion boards, for 24 cells (48v battery support being a common goal across all these projects)

Either would also include the option of enabling resistive "bleed-down" cell balancing if desired / needed.

Perhaps I will mock up a couple ideas and price them out - see if that gives any additional insight.