bvgandhi via EV wrote:
Hi,
I was trying to develop a BMS on my own. While I used the Elithion website
to guide me through the ASIC selection for my slave board I am not sure as
to how do I select my temperature and voltage sensor for an efficient
circuit design. I am a mechanical engineer by vocation so I would really
appreciate an electrical engineering perspective.
This is a problem I've put a lot of effort into as well. I've done a
number of BMS designs over the years, both for myself and for clients.
Nothing is perfect, but I like mine better than the crappy Chinese
versions, or overly complex auto company ones.
It's not a simple problem. It needs to be reliable, as you're going to
duplicate the circuit perhaps 100 times. It also has to keep working for
years, in the harsh automotive environment. This is what knocks out the
cheap ones.
It also needs to be fail-safe. Failure modes cannot lead to dangerous
conditions that murder batteries, or even worse, start fires! This is a
hidden danger with many poorly designed BMS. This is what gives BMS a
bad rep, and leads many to go with *no* BMS rather than a bad one.
It also needs to be affordable. People simply won't pay for reliability
and safety -- when they don't know what they're doing, they go for
"cheap" as their #1 criteria.
I know it is fashionable to use computers for everything. But I think
this is a case where KISS (Keep It Simple Stupid) is desirable.
Computers need programming; and virtually every program has bugs.
Computers also need power, which can run down the cells from just
sitting. And they "crash" in unpredictable ways. So I would rather have
the circuit-per-cell not depends on a microcomputer.
Fundamentally, there are only a few things that a BMS must do:
1. Generate an alarm if a cell gets low.
2. Reduce power or shut down the controller if a cell gets too low.
3. Generate an alarm if a cell gets too high.
4. Reduce power or shut down the charger if a cell gets too high.
If this is all you want, I've designed several dirt-simple BMS. These
were inspired by Cedric Lynch's minimalist BMS. Basically, you have a
tiny circuit across each cell. It contains a half dozen parts that cost
about 50 cents total, to perform as follows:
1. A zener diode to shunt excess charging current when cell voltage
exceeds its threshold. The shunt current is low, so the cells had better
be well balanced to begin with, and it takes a long time to balance at
the end of a charge cycle.
2. An LED that lights to show when the zener is shunting current.
3. An optocoupler, whose output turns on when shunting current. The
output side of all these optos are in parallel, so when any one turns
on, it activates your dash warning light and cuts back your charger.
4. Another optocoupler that is normally on when the cell voltage is
above your "low" threshold. Their outputs are all in series, so any one
turning off breaks the chain, to activate your dash warning light and
cut back your controller.
If you also want it to Balance (and not just monitor), then the BMS
needs a way to individually charge and/or discharge a single cell, to
bring it back into balance with the rest. This adds:
5. A way to switch a load across an individual cell, to slow down its
charging. This is the cheapest, and so most popular approach.
6. A way to charge an individual cell, so speed up its charging. This is
harder, but more efficient and versatile (lets you commission a new
pack, or balance a mismatched pack).
Then, there are nice-to-have features if you're more technically
inclined and want to know what is going on "under the hood".
7. A way to read the voltage of each cell. This provides a maintenance
check, to insure that things are working as intended.
8. A way to measure the temperature of each cell, or at least groups of
cells, so you don't try to use them if the temperature is out of spec.
My own conclusion is that the best approach to provide these features is
to have one relay per cell. The relay routes the voltage of that cell to
a central controller, which can measure its voltage, charge, and
discharge the cell as appropriate.
Then you can put a lot more effort into making the central controller
accurate, reliable, and fail-safe.
This is what I've done with my Battery Balancer
<http://sunrise-ev.com/balancer.htm>. It's a home-made prototype that I
built for myself as a research tool to explore the idea. It's therefore
expensive, as I don't have the resources to perfect or commercialize it
(and I doubt there is any market). But it's a starting point if others
want to help!
--
Problems that go away by themselves will be back with friends.
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com
_______________________________________________
UNSUBSCRIBE: http://www.evdl.org/help/index.html#usub
http://lists.evdl.org/listinfo.cgi/ev-evdl.org
Please discuss EV drag racing at NEDRA (http://groups.yahoo.com/group/NEDRA)