BMS free zone battery management

Post by **acmotor** » Thu, 11 Jun 2009, 23:56

In another time, another place, another thread the topic of not using a BMS was raised. that other place

Ok, so thinking caps on.... can the charger / motor controller be intelligent enough to 'manage' the cells of a battery pack for over and under voltage and perform some degree of balancing without the need for individual BMS on each cell ?

Don't choke..... it was a fair question.

Remember, 'can't' is not a word, just a state of mind.

Some possible inputs / options.....
You can have total control over charging current and voltage and also modulation of the current. You can have plenty of info on ambient temperature, pack charge and discharge history etc. You can measure pack impedance at any time.

No ostrich solutions of pre-balanced / matched cells thanks.

Detecting a faulty cell is an essential function.
Fault may be physical damage to cell or wiring.

System will most likely be quite chemistry dependent.

Don't reply to this post if you think it can't be done. !

Website · Post by **Simon** » Fri, 12 Jun 2009, 00:07

How about just having individual chargers? Guess that would be as costly as a good BMS when your talking 45+ cells!?
And a Paktrakr or similar monitoring each cell?

Post by **Tritium_James** » Fri, 12 Jun 2009, 00:16

Well it's certainly possible with flooded lead acid! You overvoltage everything, the higher cells gas, and this burns off the excess power while the lower cells come up to voltage. When everthing's gassing, your pack is balanced and you stop.

Don't try this with lithium... (sorry, I'm going to be negative about that one!)

Post by **Tritium_James** » Fri, 12 Jun 2009, 00:19

On a more serious note, it would be theoretically possible with a lithium pack to detect a failed cell, monitoring only the total pack voltage, if it failed suddenly. Gradual droops in voltage cannot (I think) be distinguished from the droop due to change in state-of-charge/temperature/load, even if you know the temperature and load.

edit: i spel gud.

Post by **acmotor** » Fri, 12 Jun 2009, 00:48

Simon, individual chargers and a paktraker per cell don't count.
For this exercise, you can only access the total pack end terminals, not individual cells.

TJ, there was a yellow top sealed lead acid test done to 800 cycles on a 144V string some years back in the US with no BMS. It used a pulsed overcharge to examine the impedance.(and knowledge of Ah supplied before and after the pulse overcharge). That seemed to work. As far as I know, nobody gets 800 deep cycles out of yellowtops in practice.

A mathematical model for the battery pack with voltage, current, temperature, internal impedance history compared to what you have happening in front of you would be good.

For instance, what happens to the internal impedence of a TS in the 3.6 to 4.2V zone ?? Coulomb, you have the test rig !

Post by **Tritium_James** » Fri, 12 Jun 2009, 01:18

Internal impedance at what frequency? It probably does matter...

Post by **acmotor** » Fri, 12 Jun 2009, 02:27

Well that is just it you see. If you can find (or already know) of some neat measurements that relate to battery condition, they may be useful in making the ultimate battery managment system !

Post by **Squiggles** » Fri, 12 Jun 2009, 04:21

There is probably some super scientific way of injecting pulses of different frequency signals and looking for reflected waves from changes in impedance or something. The reflected wave could tell you the distance to the point....this could also be a lot of hooey but it is how cable faults are identified and I know that is very accurate.

Post by **acmotor** » Fri, 12 Jun 2009, 05:09

Ummm, so whilst the original idea was one wire each end of the battery pack, it may be an idea to have a connection to the centre of the pack as one way of looking for a balance between halves and also to allow signals to be injected at each end wrt centre and look for some reflection profile of cell impedances ? Keep thinking !

Post by **Richo** » Fri, 12 Jun 2009, 05:19

Mmmm.

Some square pulses of variying Current using a Fourier transform may produce a "3-D Snapshot" of the battery pack performance.
Hard to say if it would show early signs of failure.

Even if there was a way too see variations on the pack
from the end terminals and you could tell the pack is unbalanced
I don't see a way of balancing from the end terminals without
the use of a per cell BMS.

Lets say a high freq pulse was put into the end terminals.
The lower the frequency the more chance the pulse would get
to the end battery without degredation.
As the frequency is increased the pulse will be distorted
to the point that the last battery doesn't notice it.
Therby adjusting the frequency each battery could sort of be
charged individually.

Nah that'd never work

Plus the RF emmisions would never pass CE/C-tick.

Post by **acmotor** » Fri, 12 Jun 2009, 06:41

Ok, so are we getting a feel for monitoring versus correcting ?
Are the tasks quite separate ?
In the first instance, can we just monitor for OV, UV and cell failure ?

What are our options for getting a 'picture' of the pack ?
Perhaps just to say there is a problem rather than actually locate it to start with.

Post by **Richo** » Fri, 12 Jun 2009, 07:13

Just monitoring OV and UV at the end terminals could lead to missing a faulty cell.
ie pack becomes unbalanced so some cells OV some UV but the overall is in spec.

Post by **acmotor** » Fri, 12 Jun 2009, 18:30

Richo,

Quite right, in its simplest form.

The idea of this thread is to think more about battery health from angles other than the BMS on each cell position.

e.g. Toyota have stuck with NiMH because (? for one reason, don't go there) they can manage a pack without individual BMS per cell.

Zebra Nickel salt don't run BMS per cell.
Some Lead Acid operate without BMS per cell as do some Flooded nicad.

Now some NON BMS per cell (and even NON overall BMS) battery systems just work to start with and die an early death. Others (as above) work either by nature of their chemistry or by intelligent management.

What I am fishing for is.... are we not able to run without BMS per cell (long term) with Lithium simply because of the chemistry or because we haven't been cleaver with an overall BMS design.

So, how to detect an unbalanced or faulty in a string without connecting to each cell ? (remember, you are already connected via the string).

Thinking caps gents !

Post by **fuzzy-hair-man** » Fri, 12 Jun 2009, 19:21

acmotor wrote:So, how to detect an unbalanced or faulty in a string without connecting to each cell ? (remember, you are already connected via the string).

Thinking caps gents !

I'm mostly out of my depth in this discussion but it occurred to me that as you (acmotor) said we are connected via the string, which will also conduct heat from the cells, perhaps you can sense whats going on with individual cells down the string by putting a temperature sensor on the pack terminals, you might be able to calculate using thermal conductivity which cell(s) have just started creating heat, if you knew how much current you were putting in and the heat this would ordinarily create, you might be able to calculate how far in from either end the warm cell might be? The heat signals might be too small or confused by the time you get to really large strings but if your breaking the pack up anyway? perhaps the non BMS can work with the 48V modules?

If you were confused by a signal, ie it could be one cell failing or the whole module starting to warm, then you might be able to limit current out / in as a cautious reaction if then you got heat signals consistent with the whole module warming up then you know you're OK? or you watch things over time as current / and temperature change and then attempt to use these to figure out what cell(s) are creating the heat and how much? we said in a previous topic that the best heat signal probably came from the terminals...

Which actually has me thinking again, what about a a sort of physically implemented BMS? temperature switch sort of like a bimetalic strip connecting the cells as it gets hot it breaks a circuit(probably separate), OK it's getting more BMSy but it is still pretty simple and a way to get a temperature sensor on each cell to let you know when one gets dangerously hot? if you aren't able to shunt power about or bypass a cell, detecting and alerting the driver about failing or unbalanced cells is the best you can do right? so you don't necessarily need to let them know which cell went bad (although which module would be nice).

So charging if you aren't able to shunt charge around doesn't that mean you have to detect when some cell in the pack has had enough and then default back to a level of charge that would not damage or at least limit that cells damage whilst the other cells are getting topped up? could the temperature be used to say, OK one cells had enough we have to start trickle charging (or something similar)?

Half baked ideas so....

Post by **Richo** » Thu, 25 Jun 2009, 03:38

electrochemical impedance spectroscopy

page 48/71 is the relevant topic.

Still won't help you balance the pack though.

Post by **Electrocycle** » Thu, 25 Jun 2009, 03:57

How 'bout putting all the cells in parallel so you only have one BMS module, then running a big boost converter

Post by **Richo** » Thu, 25 Jun 2009, 04:23

A 3.2V to 144V 100kW converter would be nasty

Post by **Electrocycle** » Thu, 25 Jun 2009, 04:38

just a bit!

The main power cables would have to be fairly thick.

Post by **Richo** » Thu, 25 Jun 2009, 05:13

Well at least the 9000Ah 3.2V TS cell would be an option then

Post by **coulomb** » Thu, 25 Jun 2009, 05:16

Richo wrote: A 3.2V to 144V 100kW converter would be nasty

I bet Tritium could do it. I've heard they can convert half a volt to any reasonable voltage with reasonable efficiency, though I'm not sure of the power level. Obviously, you don't use the usual circuit for this application.

You could also use just one huge cell, like this one:

http://www.thunder-sky.com/pdf/TS-LFP9000.pdf

9000 Ah at 3.2V = 29 kWH.

Post by **coulomb** » Thu, 25 Jun 2009, 05:16

Richo wrote: Well at least the 9000Ah 3.2V TS cell would be an option then

Oops, beat me to it.

Post by **Richo** » Thu, 25 Jun 2009, 05:52

Post by **acmotor** » Thu, 25 Jun 2009, 06:23

I think there is a fine if the text in your post is shorter than the text in your signature below !!!

.... after all, a bus would go nicely with a pack of 9000Ah TS !

Electrocycle. Putting many cells in parallel would solve one problem.... you wouldn't need to look for a faulty cell, it would just vaporise if it went s/c internally !

Post by **Tritium_James** » Thu, 25 Jun 2009, 13:30

coulomb wrote:I bet Tritium could do it. I've heard they can convert half a volt to any reasonable voltage with reasonable efficiency, though I'm not sure of the power level. Obviously, you don't use the usual circuit for this application.

Heh, that was a bit different power level to EVs. That was 0.5V to 12V boost, at about 10W output. We managed ~60% efficient, which is pretty good for starting from 0.5V, but not too spectacular when running a car from batteries! Going from 3V would be better.

OK, to convince myself it's not do-able, here's the back-of-the-envelope calculation to see what's feasible. The power device (MOSFET, IGBT, etc) must be rated to the input (low voltage) current, but the voltage of the output plus a bit. Say a 75kW rated converter, for a reasonable performance EV. You're using our motor controller, so 450V max voltage (using 600V silicon). Input current @ 3.2V = 23.5kA. So you've got silicon rated to process 10.5 MW, actually handling 75kW. Not very good device utilisation, and a 10.5MW IGBT isn't going to be cheap

Also have fun finding a fuse and contactor for 25kA...

Post by **Electrocycle** » Thu, 25 Jun 2009, 14:59

acmotor wrote: Electrocycle. Putting many cells in parallel would solve one problem.... you wouldn't need to look for a faulty cell, it would just vaporise if it went s/c internally !

It'd be fine. The cells would all be too scared to go S/C