Friday, February 27, 2015

Scope Creep - -

Kai had added a comment asking the question:

From what I understand, you also want to include cell-balancing for LiPos into this BMS. Don't you think this makes the whole project overly complex? There are so many dedicated devices for this task and only very few people are using LiPos on a boat or in an off-grid location anyways...BTW, I'm observing your projects for quite a while now. You are really doing a fantastic job, thank you! I'm in an off-grid situation with solar, batteries, an inverter and a 24v generator. A combination of your alternator regulator, bms and possibly also the mppt solar controller would be perfect for me!
All the best from Germany,

Good question, and as I started to type the reply it become obvious I could not keep my thoughts to a few lines.  So - the topic of Scope Creep. 

All projects have it.  There is an old, but perhaps not PC, saying in the Tech industry that at some point in time Marketing needs to come into the lab - gas all the Engineers and take what is setting on the bench; else nothing will every get to market.  My original DC generator started out as a simple regulator, expanded to include engine start/stop, then throttle control.  And hey, how about a remote panel as well!  So, what about this Cell balancing I poke at..

I think one will find we are rather close to a cross-over between traditional FLA batteries and LiFeP04.  Much depends on import duties and taxes, as well as if the LiFeP04 cells live up to their projected life expectancy, but given their wider usable range - as well as better charge efficiency, once might find they are very close to a Return On Investment cross over point with FLA..  This is a long way to say;  I think we will see more LiFeP04 deployments over the coming years.

But I also think there will continue to be lots of traditional lead-based storage out there as well.  Hence my general goal for all these projects:  Multi-chemestry, and 12..48v deployments.

But what about the Cell Balancing:  There seems to be a bit of dynamics in the LiFeP04 world to simply monitor, or have active balancing.  Will take time for that to settle.  My idea is to have as an option the ability to either monitor for imbalance, and/or provide for some type of active balancing capability.  This could range from simple monitoring using an existing modules such as the: modules. ( I have shared a few Emails with him.)  to doing an full monitor/balance.  

My work plan is to settle on a concept for the cell monitoring /balancing, as that will drive what hardware interface is needed in the main module.  Then develop the main module  (Battery MONITORING module) to work in conjunction with the MPPT controller and a future DC Generator  / alternator regulator.  Once all that is completed, can go back and add what every approach for the balancing / monitoring.  But in any case, it will be an add on option, and it is a 2nd tier work effort.  Just need to get a vision of direction today!

 (And again Thank you Kia for your questions:  Is nice to know some folks are out there)


  1. You are absolutely right. I also think that LiFeP04 will play a significant role in marine-, rv- and off-grid-applications at some point in the future. We have just invested in a new set of Trojan T-105s. But I hope it will be my last set of FLA batteries. So it is a wise decision to keep the design open for these future developments. But I had the (wrong) impression that you were planning to include all these features in v1 of the BMS (hence the name – “battery management system”, not “battery monitor”).
    I think it’s best to wait and see in which direction the development goes. There’s also a trend to make the battery packs more “intelligent” by integrating all the balancing- and protection-logic into the pack itself, so your BMS may not need (and may even not be able) to take care of that. As far as I understand, the TI-chip you mentioned in your last post is meant for this exact purpose.
    A general remark to your “engineer vs. marketing” problem: I think the best solution is to make the thing modular. This way, you can get devices out and still add some hardware features later with additional modules. Very helpful in this regard will be the CAN-bus and the protocol you are developing because it will allow the modules to talk to each other.
    Something else to tell your marketing-alter-ego: You are not on a market full of competitors, but in an open source environment. There are other people out there (me included) that are willing to help and to develop additional modules once an interface is defined. I am thinking of a nice graphical display with some information about the system... We will see. In a few weeks, I will bury a new DC- line into the ground between the shed with the batteries and the main house (right now, there’s only an AC line). I have just decided to add a CAT-5 cable for potential CAN-communication with a potential BMS in the shed…
    Nice to talk to you,
    keep up the good work,

    1. Hi, sorry for the delay - been focusing on other tasks. Perhaps I need to use BmS vs BMS :-) But yes, in the end I want to enable a wide range of chemicals (Lead, Li, etc), and voltage (12..48) with all the projects I am doing. And more it seems for the BmS/BMS having an add-on cell-balancing capability is a nice way to get there.

      Open Source and the HUI - This is one area I am keeping my eye on, and HUI is a space I really am not that keen on. My hope is by using a CAN standard, someone else will challenge the HUI aspect. Have been looking at the SignalK effort - it might be one part of the solution, but seems to still be building the basics at this point.

      And back to the TI chip: As I look at it more, it might not scale well for house batteries after-all. The note about using a 10x max scaling got me diggin a bit more - their SOC is based on model matching around measuring internal battery resistance. And in the end that might be why the scaling limit suggestions...

      So, more research to do. I have always liked the TI sensing chips for their accuracy, and ability to oversample on their own (off-loading the CPU). But I also need some type of Coulomb counter ability (if I want to use that approach for SOC). Perhaps there not a one-IC solution. . . .

      Thanks again for the thoughts - and hey: If you are running a CAT5, maybe also consider a more traditional twisted pair CAN wire as well? Still to research is the long-distance capability of using CAT-5 cable vs. 120 ohm twisted pair for a CAN bus. Would hate to have ya finding out a problem after you filled the trench in . . . . (Do you have any detailed insight into CAN & CAT5??)