Cell matching with Capacity & Int resistance caracteristic?

[Doctorbass]

After assembling many M1 cells, i had different experiences and results about succes to get a good battery pack.

Now, i need to find the REAL way to match cells. I am not sure about capacity matching and need to clarify that with your opinions.

First, i think there are at least 4 parameters to take in acount about cells matching:

-measured internal resistor
-measured capacity
-the age of the cell
-the past event that happened with the cell (over dis/char, reversed..)


Like you know, most of my M1 cells come from defect 36v battery pack so it's not difficult to get good and bad cells.
When i get the pack, i first measure each voltage from those to identify wich one have low voltage. I dont know what is the past of these pack, so it's the only way to get info about those.

I often get at least 2 or 3 bad cells out of 10 on each pack. I consider bad when voltage are under 2.5v and other are like 3.2-3.3V. I suppose that they could have go lower tham 2v because after it occure, the voltage usually rise a bit.

I am confused about something strange: the measured capacity of the "bad" cell is good most of the time !!? .. i measure around 2250mA. I see no capacity difference between bad and good cells

Now, if i decide to match my battery pack by their capacity... I am not sure that i would use the bad cell for that...

On the other hand, i ovserved that the internal resistor have relationship with capacity. the lower is the capacity, the higher is the internal resistance...

Doc

 

[BiGH]

I am confused about something strange: the measured capacity of the "bad" cell is good most of the time !!? .. i measure around 2250mA. I see no capacity difference between bad and good cells

well what makes it a bad cell if it has still got 2250 mA? voltage level on full charge?

I hate to say it, but i don't think you can limit yourself to just one method - must do them ALL.

if a cell fails one test then it should get relegated into a backup pack.

You could use the extra (mediocre cells) as a booster pack

keep up the great work!

 

[Ypedal]

Un-ballanced pack !

What i think may be happening is that during shipping/storage etc.. while the packs were un-used. some cells loose their peak voltage and become unballanced, then during use the lowest cell cuts off before the others resulting in the pack being returned udner warrantee..

Big question here : How many mah does the BMS allow for ballancing ? is it possible that it would require multiple cycles to re-ballance a pack using the BMS ? wich most contractors won't bother with on the job site.

 

[qdlaty]

My way of matching cells before assembly big pack is to measure their capacity and internal resistance and compare them with tables and curves I made for different types of cells (producers, nominal capacity etc.) so if I'm not able to match them perfectly identical by both - capacity and Ri I'm doing some compensation - to use cells with higher Ri I must chose them to also have higher capacity.

This idea is based on research where I get dependency between SOC and Ri - the higher SOC is the lower Ri I measure.

After that I've very stable BP that not only is well balanced but also behaves very solid during high discharge currents.

 

[Ypedal]

gd : What the hell is that in your avatar ? One honking M1 cell power backup for a small town ? or ?

And welcome to the forum !!! 8)

 

[qdlaty]

Ypedal
Thanks for warm welcome.
The "thing" in my avatar is me, and if you are asking what is this box I'm holding on it's just a 54kWh Battery Pack that you could use to drive around many cities.

 

[Doctorbass]

Ypedal
Thanks for warm welcome.
The "thing" in my avatar is me, and if you are asking what is this box I'm holding on it's just a 54kWh Battery Pack that you could use to drive around many cities.

Impressive!!

And welcome too ! I appreciate your coments about capacity, Ri and ROC to match cells.


Have you pics inside the pack!?
I know that i'm not the only one that would see it!!
:wink:

Is it made from LiFe, 18650 , A123...NiMh?
Doc

 

[xyster]

Ypedal
Thanks for warm welcome.
The "thing" in my avatar is me, and if you are asking what is this box I'm holding on it's just a 54kWh Battery Pack that you could use to drive around many cities.

Looks like a coffin.

 

[qdlaty]

This "coffin" is made of regular Li-ion batteries, in NCM technology.
Unfortunately, it was made for private company so I can not show its inside.

Anyway, I will be glad to help anyone with the little experience I have about managing Li-ion batteries. So, feel free to ask.

 

[TylerDurden]

Looks like a coffin.

It looks like the big slab in "2001: a Space Oddesey", to me.

Imagine a buch of EV apes jumping around it!

 

[Doctorbass]

qdlaty,

The pack we are talking about seems to be similar to the one with the tesla roadster right?:

"The
pack operates at a nominal 375 volts, stores about 53 kilowatt hours of electric energy, and delivers up to 200 kilowatts of electric power" (6831 cells)...


(The Tesla Roadster Battery System Tesla Motors
August 16, 2006 By Gene Berdichevsky, Kurt Kelty, JB Straubel and Erik Toomre)


http://www.teslamotors.com/display_data/TeslaRoadsterBatterySystem.pdf

Doc

 

[Doctorbass]

?

 

[TylerDurden]

And to me it have a very similar case to the 15kW car audio amplifier made by Rockford Forsgate

BOOM, baby!


Welcome, qdlaty!

I am confident your "little experience with LiPo" is as little as the coffin/slab/amp/pack...

 

[Ypedal]

I will be glad to help anyone with the little experience I have about managing Li-ion batteries. So, feel free to ask.

Be careful what you ask for .lol...

 

[qdlaty]

OK,
just a bit of information about this black baby. I've made three of them (Fat Betty, Hightower and Wardrobe were the codenames).
The nominal voltage is 259 and full charged 294V
it is build of 200Ah cells that can kick constant current of 1000A and I don't even want to know (or test) their peak capabilities.
The application is big city car (van size) and we are setting up production line to start 2 Jan next year.
That's all.

For my experience, I've master degree in electrochemistry (specialized in polymer electrolytes for lithium ion batteries) and pending PhD on that subject and 3 years hands-on experience in building Battery Packs and BMSes for EVs, electric boats, UPS etc.
So, whenever I can help, you can count on me.

That's definitely all.

As far as I remember the topic is about matching cells lets go back to it.

 

[vanilla ice]

I'd definitely kill myself trying to build one of those. I make enough sparks and smoke playing around with what I got.

 

[Malcolm]

Hi qdlaty, it's a small world

I'm planning on building a small A123 pack for ebike use. I'd like to use a 12s 4p configuration, which means I can't easily use the original DeWalt charger. What's your view regarding the need for balancing A123 cells during charging? From what others have reported it seems that only occasional balancing is needed if a pack is well balanced to start with and cells are not discharged beyond around 2.0V. Does that match up with your own experience?

 

[qdlaty]

With A123 products I've only experience in building buffer battery for solar installation. Cells were only discharged with small currents so I did not notice unbalance in time.
For all I've done with small cells (PHET, A123 etc.) the most important thing is proper connection. The cells itself have internal resistance at the level of 40-50 miliOhms but you may make your pack unequal by wrong welding.

My suggestion is to prepare way of connecting every single cell in identical way.
Second thing is balancing, usually is done by comparing actual cells voltages between lowest and highest and dependently on the threshold you set highest cells are connected in parallel with resistors (it's how passive balancing works). But the problem with LiFP and LiTiO is very flat discharge curve for very wide State of Charge level, so balancing will only work during end of charge or discharge, so the efficiency of chosen circuit must as very high.

But it general my conclusion of using LiFP cells (no matter of the producer) is that small capacities can vary very little in time so keeping passive balancing at the end of dis/charge that will take off the the single cell 10% of charging current will do the work.

 

[Malcolm]

Thanks qdlaty,
That's a good point you make about ensuring equal connection resistance within the pack.

 

[qdlaty]

To continue that topic, about equal connection.
I made Battery Pack of 1806 pieces 18650 cells connected 43S42P.
The 42P blocks were made by spot-welding cells to big nickel plates with custom shaped holes (between every pair of welding spots you must have hole, space in metal, otherwise you get electric arc) and I noticed that with such big area I get unequal connection quality (you can not undo the welding, so if one cell is made wrong, whole block is a waste) so it was very hard to chose equal quality blocks.
Next I've noticed inequality in temperature distribution on the plate - the further from block tabs were cells the warmer the area around welding spots was, so that gave me the idea of resistance problem.
So, the next construction is a bit different - every cell has its own connecting plate and everything is separately connected to block's current collector. This also gave me the possibility of easier PTC fuse installation - instead using ring-shaped PTC dedicated to spot welding I only use fuses that are crimped or welded directly on wires. It makes the work easier and blocks are far better performing under heavy load.

The next thing in building big blocks of small cells is temperature distribution. In the 42P pack I measured even 5 deg C temperature difference between lowest and highest value.
The tip for solving that is to close block in some material that will quickly distribute heat around whole block. Unfortunately it means some kind of metal made block and that is heavy and costly. One thing I've found is a block made of some kind of foam mixture of graphite and wax. This is excellent heat conductor but causes the construction costs nearly doubled.

I'm working on third solution because I not going to increase either the weight or cost of the system but I need to have stable temperature conditions.

 

[fechter]

To continue that topic, about equal connection.
I made Battery Pack of 1806 pieces 18650 cells connected 43S42P.

Holy Crap! that's a lot of batteries.

I'm interested in the spot welding process. Does the welder use two electrodes? What is the electrode material? How many amps does it take to weld?

To even the temperature, I'd recommend circulating air between the batteries. This is how Honda and Toyota deal with the problem in hybrid cars. Blowers don't require much power and are cheap.

If the plates had holes that line up with the space between batteries, air could be circulated through the pack. Even circulating air around the existing pack will speed up the heat transfer across large distances.

Is the current pack like potted in the carbon/wax stuff?

 

[Doctorbass]

Any news about the battery pack you are working on qdlaty ?

Doc

 

[Doctorbass]

here is the excel preview of the results for the future 370x 18650V cells matching.

I made 2 graphic representing the cell distribution for capacity and Internal Resistance.

This is very strange, that they both have two peak. For the capacity, it's around 2840mAh and 2965mAh and for the RI it's 32 and 23 miliohm. :|

ALL these cells was returned to the service center because:

-deffective new
-defective aged
-aged

I had would to conclude that the reason because the two peak could be because of the two method i used to measure them. first for the 80 first cells i only used 2 wire to dishcarge the cell with the tester. then i developped a new method to test them without having problem dur to eire resistance that interfered with the RI measurement value for the other next cells.

maybe the old method (2 wire) had it's own peak with higher RI and lower Capacity and that the new 4 wire method ameliorated the resusts creating a second peak iin the graph?...

BUT NOT!! i've re-tried to built a graph with ONLY the new method and you know what?! I got two peak again!!! around the same value of 2840mAh and 2965mAh and for the RI it's 32 and 23 miliohm. :|

it seems that these li-mn cells could have this caracteristic... most of them tend to have a C around 2840 and 2965... maybe not...

it's confusing!!

Doc

-