Samsung 30Q INR18650-30Q 3000mah 15a cell

[ridethelightning]

i had exactly the same problem.
for a friend, i initially was going to build a 20s9p pack, but later we decided to make 22s, so ordered another 18cells.
strangely, the first batch was the NEW cells with 3,44V, but the 18 cells we purchased later (1 week maybe) had about 3,66V, so it must have been earlier batch

sorry if this is obviously spelt out here, but just to clarify, were the 2 batches behaving like this with the same number of cycles on them? i mean, if the first batch had even a few more cycles, this could also explain the difference in capacity, just looking at all possibilities here.

 

[Allex]

That's some harsh abuse! Any info on pack design, IR, capacity? As I'm sure you know, if there is a weak point in the pack, those cells would get hotter and experience more decay. I'm curious why the weak ones are weaker

Pack design is a simple layout 6rows by 18 rows of cell each cell is connected to the next serie, so all of them have equal current distribution.
I have now disconnected all the cells and check the drift. A few of the cells went from 4.15(fully charged) to 3.5V in one week. half of the pack is ok and did not loose any charge at all. So it looks like some of them disliked the abuse and some if them did not bother at all

 

[madin88]

sorry if this is obviously spelt out here, but just to clarify, were the 2 batches behaving like this with the same number of cycles on them? i mean, if the first batch had even a few more cycles, this could also explain the difference in capacity, just looking at all possibilities here.

both batches had been factory new. zero cycles
what i did was discharing the older batch (with the higher voltage) to the same voltage as the newer batch came.
that was 2s9p with 2A discharge current. No abuse here done

 

[ridethelightning]

thanks maddin in this case it looks like the best way to go is to NOT mix the 2 types if possible.

.

 

[MitchJi]

Model S cells end at near 75C or less and get 3.1Ah on a 10A full draw, in my testing with a taped on thermocouple. No airflow. I consider 75C too hot.
I expected the 30Q to be _MUCH_ cooler on 10A, as in ~50C at the end of a 10A draw.

Have you doned any testing with charge rates with the Model S cells?

 

[ridethelightning]

sorry if this is obviously spelt out here, but just to clarify, were the 2 batches behaving like this with the same number of cycles on them? i mean, if the first batch had even a few more cycles, this could also explain the difference in capacity, just looking at all possibilities here.

both batches had been factory new. zero cycles
what i did was discharing the older batch (with the higher voltage) to the same voltage as the newer batch came.
that was 2s9p with 2A discharge current. No abuse here done

Ok so my friend did some discharge tests of the 2 types of cell with a balance charger.
2 of each type were tested for capacity and internal resistance.

this is what he found-

[... in a nutshell there appears to be a 45mAh difference. (5amp discharge) cells charged to full 4.2v, but the initial storage voltage from supplier is shown here to denote the 2 types.
Cell. V.------------------------Capacity mAh. -----------IR mohm
1, 3.432 ------------------- 2777 ------------------------27
2. 3.435 -------------------- 2775------------------------31

3. 3.655 ---------------------2730------------------------29
4. 3.655------------------------2732-----------------------30

IR is essentially equal and the difference in capacity is aprox 1.6%
I would guess that if the lower cells were spread amongst the P groups it would be fine]

so i thought id ask around for other opinions here, if it would effect the packs performance much, or if better to keep the 2 types separate?
cheers
RTL

 

[madin88]

ride, thanks for the tests. good job!

as expected, it shows the new batch is a tiny bit better which is good on the one hand, but bad on the other.
50mAh difference could mean that during discharge the new batch still has 3,5V while the older batch already has something like 3,2 or lower, and the deeper the cells get discharged, the higher the voltage difference will become.

Same thing during charge, only the other way round: the new batch will have 4,2V while the older already would have 4,25V or something like that because it only can store 50mAh less, so it will be fully charged quicker.
If you have 10p lets say, the difference is 0,5Ah. This would mean for instance that Adaptto BMS with it's 50mA balancing current would need 10hours!!! to equalize the pack.
This is something you don't want to have on a new pack, so better don't mix those batches.

It will be absolutely no problem if you spread the batches EQUALLY among the p-groups, but i would recommend to first match the voltage, because 0,22V is not that small difference and the balancing currents could be quite high.

 

[kdog]

hey maddin
i could be in the same boat as RTL.
i was suggesting that it would be ok to mix up the two batches to RTL but i wasnt sure.
say you had 8p with one cell 50mA less. all P groups were the same etc. my feeling is that this would be fine and that there would be just some minor current flows to and from the weaker cell during ch/dsch.
is this correct?
and the cell wouldnt be stressed as it would just be supported by the others?
cheers

 

[madin88]

say you had 8p with one cell 50mA less. all P groups were the same etc. my feeling is that this would be fine and that there would be just some minor current flows to and from the weaker cell during ch/dsch.
is this correct?
and the cell wouldnt be stressed as it would just be supported by the others?
cheers

yes, exactly
but i would match the voltage before welding such p-group together because 0,22V differnce could lead to high current flow.

 

[Allex]

What if you have some cells that gives you 2000mAh and others 1800mAh... is it better to mix them in one P group and let the better ones support the weakers cells or is it better to separate them in two different groups so one series would have only 2000 cells and another only 1800?

 

[ridethelightning]

my2c its better to mix them evenly across the p groups, but id still suspect some odd shinanigans, especially after the cells age a few hundred cycles..

 

[Allex]

If a good cell need to support the weaker one, would't that mean that the greater cell will see more stress than it needs to see, which potentually means faster degrading of it? I though that by separating the bad ones completley, at least you actually know which ones they are and can swap them out if needed?

But of course, with this strategy you will have a Stable BUT disbalanced pack. But knowing the nature and history if the build you can probably live with a few rows that are always a bit disbalanced. I am still not sure which method is best here...

 

[madin88]

If a good cell need to support the weaker one, would't that mean that the greater cell will see more stress than it needs to see, which potentually means faster degrading of it?

in case with the Samsung 30Q "old vs new" batches, the difference is so small that it should not be a problem.
You should keep in mind that the higher you discharge the cells, the more the stress will be because of the voltage cliff.
Avoid to come there and your pack will stay balanced very very well and BMS never needs to work.

I though that by separating the bad ones completley, at least you actually know which ones they are and can swap them out if needed?

describe "bad ones"
40-50mA difference could be after a few cycles and if you swap an entire group of cells out, the remaining cells will be than worse while the new cells can hold more charge. you would not make anything better.
swapping one cell out from 10p won't be much noticable, but an entire group definitely.

If you have the problem with the different 30Q batches, i would recommend to bring down the voltage of the old batch close to the new one (some mV differnce don't matter), and spread them EQUALLY over the groups.

 

[Allex]

What I am talking about is that some cells have 50 Cycles on them and some are brand new. How would you do in this case?
I also have cells that have 50 Cycles and with a self discharge at about 0,002V per two weeks while the good ones with 50 cycles have 0.000V self discharge during the same time. The idea is now to build a pack using this "nice" mix of cells!
Of course, ideally, one would use new cells for the whole pack, but in case this is not possible, how would you proceed given the above circumstances?

 

[Punx0r]

I would mix them up so (as far as possible) each parallel group has identical capacity and self-discharge. Otherwise your pack will always be limited by the weakest link in the chain (the weakest cell group).

 

[Allex]

But don't you think that by mixing them up we really do not help the pack to be better in terms of performance, what we will se is overall evenly declined performance?

 

[Punx0r]

Say you have a 10S10P pack (100 cells) and of these:

90 are "good" at 3000mAh
The other 10 are "bad" at 2950mAh

If you put all the bad cells together you have 9 cell groups @ 30,000mAh and one at 29,500mAh. The BMS will cut out when 29,500mAh has been drained from the pack (useable capacity) because the "bad" cell group will hit low voltage cut-off while the remaining 9 cell groups are comfortably above it.

If you mix up the cells so there is one "bad" cell in each group, then all 10 cell groups will have 29,950mAh (9 x 3000 + 1 x 2950) and you will have 29,950mAh useable capacity. At LVC all cell groups will be equally discharged to the same voltage.

So by mixing up the cells you gain 450mAh of usable capacity. If the "bad" cells are also of higher internal resistance then the load on them is supported by the "good" cells. If they were in a cell group of their own then they would get run harder (and hotter) than the other groups and require more balancing.

 

[Allex]

Yeah thats make sence, thx!

 

[Cephalotus]

Updated\
Seems the Model S cell has some sort of self healing characteristic. Very interesting.
Any significant pause in cycle testing results in improved capacity.
All of the same pauses are done on the 30Q, and it only results in the same or reduced capacity.

Any updates on those cycle tests?

 

[goatman]

Interesting. I have a Leaf, and I don't know exactly what the charging algorithm is, but it's definitely faster at low SOC (I use 6.6 kW level 2; my car doesn't have the DC quick charge option). It's nice how that works out, since low SOC is exactly when you want to push the charge in quickly to get the most range.

The Leaf also takes this into account for regen. There's a display that shows the available regen in 5 discrete levels, not sure what that corresponds to in physical units. At full charge you can't do regen at all, but then it's not like there's some threshold voltage where all of a sudden you have full regen. You gradually have more available as the SOC drops from 100% to 50-60%.

Yes. Here is a fast charge curve for a Model S cell:

Note it goes to about 0.9A by the time it hits the CV phase at 4.2V.

So, if you're doing CC/CV charging to 4.2V, safe would be 0.9A. But supercharging allows you to use 4A, down to 2A up to 70% SOC.

I suspect the 30Q could take an epic fast charge. I'll setup an ultra fast charge routine and cycle test with it.. Ill use 6-7A at low SOC...

The 3A/2A/1A charge routing i'm currently running is already pretty damn fast. 1 hr 45 min average for 95% charge.

Okashira, I was thinking the same thing about the 30q taking a 7amp charge maybe even a little more like 8 or 9amps. did you ever do the test?