Behaviour of cells in parallel take 2

Announcement - Help us fight the BOTS! Please report all spam including stuff in your inbox!

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,581
OK here is the charging curve of this test.
Again the same batteries in the same slots.

image_tpnasr.jpg


One thing to note this is a charge at 4A.
The thing that stands out of course is the low IR cells soak up the juice very early and once happy fade away.
Cell 1 which was depleted pretty badly during the discharge as it took the biggest hit for the duration was also right up there in the feeding troth.
Remind me to move the ambient temp sensor away from the charger................ :cool:

image_fwvgzo.jpg


Full chart of D/C cycle


image_ojxlmq.jpg


Can you make a test with low Cap & high IR and high Cap cell with low IR?
I will see what I can find in my plethora of cells to run this test.


Also I think I will go with the icharger I have, to increase the dischargeto 4Aas the ZH-YU limits me to 2.6A.

Wolf
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,581
So this weekend I got busy with some more charting.
Hooked my icharger to the test board and went to town.
I can now easily do a discharge and charge trace. :)
Don't know why I didn't think of using the icharger for this sooner but I think Iwas using it to build and analyse my packs. Well nevertheless i have 2 of them now an icharger x6 and x8

All tests below where done with the icharger x6 C/D/C at 1A, 4.2V to 2.75V

Just for a baseline I decided to run 4 cells of the same kind through the tester.
Here are the results.
Very smooth and even discharge all the way to the end. The same with the charge cycle.
These arePanasonic NCR18650B which I believe is the cell of cells. If I had a bunch of these I would be very happy.
All tested ? 2800mAh a while back so I didn't retest. I did record the IR though and you can see they are pretty close.

image_jzjylr.jpg


image_izthzt.jpg


So then after a long search I found some low capacity high IR cells. You have to remember that if anything is out of the IR range at my house it goes away to the recycler. So I was lucky to hang onto some Sony US18650S cells. Why I had these I have no idea I guess I was curious and kept them.
As you can see the difference is quite remarkable. The Panasonics from the getgo carried most of the load and there was no crossover at all.
Temperature also got right to the 30 C on the heavy lifters.

image_tchizb.jpg


image_cvnxjd.jpg


I ran the test twice just for the fun of it. The second time the Panasonics got a little warmer but so did the ambient.

image_orklgu.jpg


Then lastly I ran a test with 4 cells with similar mAh results but different IR.
Relatively even results with some crossover action happening.

image_jmmbtb.jpg


image_bfzedf.jpg


So there you have it.
I know what my conclusions are.

What are yours?

Wolf
 

Generic

Member
Joined
Oct 23, 2018
Messages
217
Everyone assumes that if you have 4 cells at 4 Amps, that each cell has a 1 Amp load on it. Thanks for proving otherwise!

Without looking back, did you test two 3.7V cells in parallel with two 3.6V cells, all with equal capacities and IR? Just curious to see if the 3.6/3.7V makes a difference or not, as they have radically different discharge curves.
 

OffGridInTheCity

Active member
Joined
Dec 15, 2018
Messages
1,746
Generic said:
Without looking back, did you test two 3.7V cells in parallel with two 3.6V cells, all with equal capacities and IR?
If 4 cells are in parallel, they can't be different voltages as in 2 @ 3.7v and 2 @ 3.6v (as I know you know :)). I'm not sure I'm understanding this question/though - could you clarify?
 

Generic

Member
Joined
Oct 23, 2018
Messages
217
Nominal voltages. Like the Panasonic CGR18650E is a 3.6V nominal cell vs. a Samsung ICR18650-26D is a 3.7V cell. Chemistry could be the same, but the 3.6V cell has the bulk of its energy at 3.5-3.6V and the 3.7V cell has the bulk of its energy at 3.7-3.8V. I always theorized that at 3.5V, the 3.6V cell would be doing most of the work, while the 3.7V cell would be pretty depleted and not contributing much energy.
 

floydR

Member
Joined
Aug 23, 2017
Messages
1,312
18650's come in two basic nominal voltage 3.7 and 3.6 , I think Generic was seeing if anyone has done a parallel test with a mix of the two cells.

Later floyd
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,581
Generic,

Those cells I do believe I have in stock I will find some with reasonable parameters and charge according to spec and run a parallel test.
I am on the hunt forCGR18650E and Samsung ICR18650-26D cells.

Till then

Wolf
 

daromer

Moderator
Joined
Oct 8, 2016
Messages
5,659
Different nominal cells Will look like above where they deliver different current att different voltages
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,581
Generic said:
Everyone assumes that if you have 4 cells at 4 Amps, that each cell has a 1 Amp load on it. Thanks for proving otherwise!

Generic,

The other thing to consider is also the charging of the cells, as it is also influenced the same way. Just because you are charging cells in parallel does not mean they all absorb the same amount of energy at the same rate.
Wolf
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,581
Generic said:
Nominal voltages. Like the Panasonic CGR18650E is a 3.6V nominal cell vs. a Samsung ICR18650-26D is a 3.7V cell. Chemistry could be the same, but the 3.6V cell has the bulk of its energy at 3.5-3.6V and the 3.7V cell has the bulk of its energy at 3.7-3.8V. I always theorized that at 3.5V, the 3.6V cell would be doing most of the work, while the 3.7V cell would be pretty depleted and not contributing much energy.

Looky-Looky I found some. I was some surprised.Not!


image_srsuis.jpg


I will get the 4 lowest IR and test them in my new Vapecell S4 fast charger 3A
Just another off the shelf mediocre charger/tester. IR is slightly better. Cant trust it at low V though.
I can't wait to get my 2 MegaCellChargers and see what they will be like.

image_dodlpo.jpg


Wolf
 

Generic

Member
Joined
Oct 23, 2018
Messages
217
I actually picked those two at random from the cell database, but I was looking for 2 fairly common cells with similar capacity and different nominal voltages. I have no idea if they have similar IR ratings from the factory, so hopefully they do. I'm looking forward to your test results!
 
Joined
Mar 7, 2018
Messages
1,055
Don't forget to consider this :

"With all batteries, SoC affects the internal resistance. Li-ion has higher resistance at full charge and at end of discharge with a big flat low resistance area in the middle."

https://batteryuniversity.com/learn/article/rising_internal_resistance

With second life cells these variations in IR vs SoC may create other issues that single reading IR values miss...

Like the data and testing as it is this sort of work, experimentation and research that uncovers the reality... ++1
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,581
Generic said:
I actually picked those two at random from the cell database, but I was looking for 2 fairly common cells with similar capacity and different nominal voltages. I have no idea if they have similar IR ratings from the factory, so hopefully they do. I'm looking forward to your test results!

Well here are the results of these cells tested.
Quite normal results of 2 reasonably decent batteries. I do not believe the Nom V had much influence on anything.
Wolf

image_mmrdjs.jpg


image_ybtipg.jpg
 

Generic

Member
Joined
Oct 23, 2018
Messages
217
Thank you Wolf!! During the discharge, right around 3.6V, cell 1 and 3 (3.6V cells) started to contribute more current than cells 2 and 4 (3.7V cells). However, the increase in current was not as dramatic as I expected. The 3.7V cells were around 0.9 Amps and the 3.6V cells were around 1.15 Amps. Likewise, on the recharge, the 3.6V cells took about 1.25 Amps each vs. the 3.7V cells 0.75 Amps each initially, before slowly reversing as more charge was absorbed.

So, nominal voltage makes a small difference, but not enough to cause concern when assembling packs, as long as there is about a 50-50 distribution within the packs.
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
Worth emphasis: even if you match the IR and capacity well, unlikecells in packs can still end up quite imbalanced by the end of a discharge due to chemistry mismatches - which can lead to very different voltage profiles under load, e.g. see the graph below. This means that some cells will be under much more stress than others near end of discharge, so they will degrade much faster than others (and this isa positive feedback loop, which ends upknocking down the weakest cells in a domino effect, till the capacity of the entire pack is brought down).

E.g. below note that at 3.6V the LG has about 44% left but the Sanyo only has about 14%. so even iftheir voltagesare matched their SOCs may be way off (but preciselyhow they balance under load will depend on many factors. e.g. IR, temp, etc)

image_hswnjs.jpg
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,581
gauss163 said:
Worth emphasis: even if you match the IR and capacity well, unlikecells in packs can still end up quite imbalanced by the end of a discharge due to chemistry mismatches - which can lead to very different voltage profiles under load..............
@gauss163

That is very true hence we try to encourage powerwall users to charge to ~4.0V and discharge to a maximum minimum of ~3.4 as the cells tend to wander at that pointeven if they are the same manufacturer, same part number, same production date,same capacity, very close IR etc.
I also encourage IR differences no greater than 15m? and capacity differences of no more than 400mA and if possible even less than that in a pack build.
No sure if you have looked at my Harvested cell analysis sheet with over 6000 cells recorded. Out of those 6000 used and tested cells I was able to get
1200 cells that passed my stringent tests of 40m?to 65m? and 2250mAh to 2650mAh with a minimum85% SOH. I also placed same manufacturer and part numbercells next to each other in the pack so that they would react similarly in a group.
If you have looked atall my tracesit shows how cells behave with different neighbors having a smalldifferenceof capacity and IR and large differences of IR and capacity and a combination of both. I have not seen another study like that anywhere but I may be mistaken. Discharge curves are going to be different for each cell especially different chemistry such as ICR chemistry is going to discharge different than INR NCA IMR. That we do know.Again I do not personally advocate mixing high drain and low drain cell in a pack.

The 2 batteries shown in your chart are more than likely 2 different chemistries I can find the LG18650HE4 and it is a high drain low IR cell. Designed for power tool packs.
http://www.batteryspace.com/prod-specs/10679.pdf

I have to guess at the Sanyo though as it is just (Red) no identifiers of the manufacturer part number and yes I did look at that chart and this website is already in my favorites anyway so an educated guess is that it is a ICR chemistry made for laptops, with a much higher IR. My guess it is a UR18650FM as the discharge curve is strikingly close to the one displayed above.

So kind of comparing apples and oranges and I have done those test also coming to the conclusion that it really isn't the best practice to mix chemistries.


Wolf
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
Wolf said:
[...] So kind of comparing apples and oranges

But that's exactly my point - when you mix very different chemistries in a pack they are less likely to remain comparably balanced during discharge due to their very different voltage profiles.

Re: your remark on energy vs power cells: this doesn't really matter here since the same voltage profile differences can still occur when both cells are energy cells (I simply picked the first examples I found from the list that show the voltage profile difference). It really has more to do with major chemistry differences (e.g.LiCo vs. NMC) that radically alter voltage profiles.
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,581
All I can say very true I completely agree.
Your point is well taken.

My research shows that after ~3.5V all cells no matter what Li-Ion chemistry it is will drop of causing the other stronger cellsto take over if there is a demand, creating stress on them. It isquite obvious what to do. Do not discharge below 3.5V
As far as the discharge curve is concerned if IR and capacity are relatively close the cells will not be stressed till they get to the magic voltage which keeps popping up in my tests ~3.5V
Unfortunately the discharge curve of 1 cell compared to the other cannot give you the true relationship of that discharge curve when both of these cells are in parallel.
It rudimentarily changes that discharge curveas each cell will contribute according to its ability and will hybridise that discharge curve as the voltage of the cells will be very close.
I do findLiCoO2 andLiNiCoAlO2 chemistrysomewhataccommodating to each other. For some reason the seem to play well. Almost kind of likeLiFePO4 withPb.
I don't know why but they do.
Not so much withLiNiMnCoO2 andLiMn2O4. Must be themanganese that's all I can think of.

So much more research needs to be done as far as used cellsis concerned. What can,cannot and or should notbe done.

Now mind you many powerwall are running quite well, some without much of a regard to many of these factors.
Mine is running quite well. At ~3.99 V all my packs start to congeal and come very close together to generally within 0.02 volts. At full charge the packs are at 4.04 to 4.05. None of my packs ever get to the point of the batrium BMS to step in and need to balance, YET, my packs have not been running that long only 10 weeks we will see what happens in a year.
The maximum delta I have seen so far is 0.06 V at ~3.5 V at discharge.
You see most of the research that is done is onnew cells,not used cells, of whichwe have noidea how many cycles they have been through.
All we can do with the equipment that is readily available and reasonably priced is come up with a method to at least get an idea that this cell is worthy.
After the testing we have some idea of the SOH of that cell.
Wolf
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
Wolf said:
[...] Unfortunately the discharge curve of 1 cell compared to the other cannot give you the true relationship of that discharge curve when both of these cells are in parallel [...]

Yes, such discharge curves are only a starting point, but they already suffice to show how big chemistry mismatches can greatly disturb balancing. To get more precise results you'd need to do some dynamic modelling (as do fuel gauges), necessarily incorporating all the pertinent parameters (IR, temp, etc), which is quite hairy. Instead, it is usually much easier to run some tests and attempt to glean some intuition from the results - as you have done. Even then it is often difficult to make sense of the results - even for a few cells, let alone 10 times as many. But don't give up, it seems you've already gained some insight.
 

OffGridInTheCity

Active member
Joined
Dec 15, 2018
Messages
1,746
Wolf said:
Now mind you many powerwall are running quite well, some without much of a regard to many of these factors.
Mine is running quite well. At ~3.99 V all my packs start to congeal and come very close together to generally within 0.02 volts. At full charge the packs are at 4.04 to 4.05. None of my packs ever get to the point of the batrium BMS to step in and need to balance, YET, my packs have not been running that long only 10 weeks we will see what happens in a year.
The maximum delta I have seen so far is 0.06 V at ~3.5 V at discharge.
Yes- I'm seeing similar results with my2 oldest packs at 720 daily cycles w/average 40% DOD in 4.0v -> 3.5v range - they've neverneeded an explicitbalance other than to bring new batteries online.These 2 oldest packs (14s120p - e.g. 260ah packs)are a randommix (both parallel and serial)of 'new' 36v 20cellebike packs made up of[size=small][size=small]LG-MF1-18650 and ICR18650-22Pbefore I was aware of any of these topics. [/size][/size]

Since then, I'veregularly addedbatteries of greenNCR18650A, greyLGABB41865, and blue LGDAS31865. BUT, even these are high quality (e.g. new'ish) compared to dead laptop batteries. 1 batteryat 400 cycles, 1 at 300 cycles, and thelatest couple at 100 cycles - I find that all 84 packs play nicely
with a 3.54v bottom and a max high of 4.0v (same as you). I've let it go up to 4.15 a few times (40? times out of 720 cycles)but not much / had no noticeable affect.

I've only had to turn on balance when I add a new battery. Have gone as long as 9 months with no balance so far - but during that time,one pack diddrift from 0.06v max diff to a 0.09v max diff. I added 3 new cells to it and it's been find since... so maybe a bit under powered? It was an early pack before I actually tried to balance ah / pack. One of several background questions I live with.

Wolf said:
The maximum delta I have seen so far is 0.06 V at ~3.5 V at discharge.
You see most of the research that is done is onnew cells,not used cells,of whichwe have noidea how many cycles the have been through.
I've never tried to balancebeyond 0.03v diff when I add a new battery because thingswill vary during the cycle from a 0.03v hi/low pack to 0.06v high/low pack from day to day no matter what I've tried - and of course anything <100mv is perfectly fine for my operation. I've tried 'individual packcharging' (tweak individual pack voltages) and'add / remove cells here and there' (for suspect ah compatibility) butthere seems to be some amount of fuzziness in the system - cells or longmons or reported measurements-and a 30mv -> 60mv gap up/down each day in no way takes away from the use I need from the battery bank or seem to have any long term affect.

*I'm amazed at the stability of the battery bank (without balance) - even though individual packs varya bitover time and the daily max difference between packs goes up/down a bit each day. In spite of this, the battery bankremain stable!

I did avoid using those MOLI cells that you tested - showing quite a different discharge curve from the NCR18650As.

I did have todiscard 5 packswith Green Sony G5/G7 as they would not balance with other packs - the hi/low voltage would swing 150-200mv wilder than the others (don't remember exactly) but I put that down primarily to their 200-250 IR? rather than their discharge curve but I don't know. One of the ongoing mysteries I carry from this journey.

gauss163 said:
Yes, such discharge curves are only a starting point, but they already suffice to show how big chemistry mismatches can greatly disturb balancing. To get more precise results you'd need to do some dynamic modelling (as do fuel gauges), necessarily incorporating all the pertinent parameters (IR, temp, etc), which is quite hairy. Instead, it is usually much easier to run some tests and attempt to glean some intuition from the results - as you have done. Even then it is often difficult to make sense of the results - even for a few cells, let alone 10 times as many. But don't give up, it seems you've already gained some insight.
Yes sir. I'm amazed that I've had such a good experience so far (700 cycles) and no problems with the newest packs. But of course when you add time (1,000s of more cycles) I'munsurewhat to expect to maintain the battery bank.

As packs degrade..... will they degrade more or less 'similarly'? There are several maintenance scenarios I can imagine
- a slow process of tweak/repairing packs one by one as they start failing to keep balance
- replacing failed packs with new / less-AH packs to match the remaining packs so the battery bank continues... BUT I expand the DOD % (or the size of the battery bank)to keep supplying the daily power needed.
- all or nothing
- other issues.

The continuedoperation looms as each cycle accumulates and has becomemore front/center in my thinking. Time will tell :)
 
Top