From EvTechWiki

Back to: Main Page MK3EB

Success consists of going from failure to failure without loss of enthusiasm. ~Winston Churchill

Failures

I have yet to create a battery management system that I actually truly would trust, hands off, to manage my battery pack. I've had two battery packs destroyed so far through problems with software, hardware, and operator error. Thusly, I've learned a lot of things NOT to do.

Here's a chart:

BMS failures

Project	Type	Reason it failed	What I learned
BasicStamp BMS	Probe only	Poorly documented, lack of basic understanding of task requirements	Too much to list.
BasicStamp BMS II	Relay shuttling	Failure to use basic safety precautions, components (ATO relays) not rated for task	High voltage DC is no joke! Also, if you don't include fuses, your wiring will become your fuses. EV capacity batteries are capable of making hundreds and even thousands of amps. I've more than once joked about having a EB node that is a smoke detector.
MK3v1	Shunt	PCB design not sufficient for 300V isolation (yes, *bang*)	Always carefully check the isoluation barrier between high voltage and logic. Also, more than one set of eyes should check over a PCB layout before it goes to fab. Also, budget for at least two and more probably three board turns for *everything*.
Tango	Modular Charger	MCU not sufficient for task, not enough time budgeted for transition to live/operating system, too many people involved, not good enough communication, *always* use a wiki!, code not commented well enough for occasionally maintained, not enough time budgeted for testing, not enough equipment budgeted for testing	Budget more time, more equipment, arrange to test things in small, easy to cope with chunks rather than as entire systems
PK3	Shunt	MCU not easy to code for in assembly, sometimes it just makes more sense to write in C, designer didn't think through all of the rammifications of change in CPU, transformer isolation without good enough detector, not enough noise immunity	Don't choose a CPU just because everyone else likes it. Don't expect to switch CPUs and still be as fast in writing assembly.
MK3v2	Shunt (w/upgrade header)	Well, it hasn't exactly failed *yet*. It got pretty close for a while because I was not putting enough energy into the project, but I think it's recovering under my ministrations.	Put a consistant effort into getting your project done ;-)