Recycled 18650 testing - new method - more info.

[fechter]

You really need to test both capacity and self-discharge. Either one can make for a bad cell.

If you're evaluating a big pile of used cells, both tests are time consuming. I usually test capacity first since I have an automated tester for that. If they pass that test, then a self-discharge test comes after that. Some people find it easier to reverse the order.

 

[DrkAngel]

You really need to test both capacity and self-discharge. Either one can make for a bad cell.

Personally, I would label:
Low capacity cell = Poor
and
Self-discharging cell = Bad!

 

[ASK1]

Question for you guys about cell testing. I have 110 new NCR18650A cells. I charged all of them to 4.21 V and let them sit idle for ~3 weeks. I've checked the cell voltages today, here is the result:

They were charged 4 at a time (painful) over the course of a week so some of the cells had a longer sag duration (up to an additional week), so that widens the distribution a bit. Are any of these cells suspect? How long does a voltage sag test need to be to reliably detect defective cells, and what's the threshold?

Also, do I really need to be able to charge cells to +/- 0.001 V precision when I assemble my pack? I don't have that capability. But I do have a BMS.

 

[atomek1000]

Regarding only your last question the answer is simply no.
Equalizing cells before connecting them in parrarel is reccomended to eliminate high balance currents at connecting stage. But voltage difference less than 0.1V will not do much harm, probably none.

The other thing is series.
Balancing packs of parrarel cells to be connected in series is very important because a plenty of low cost bms does have pretty low balancing current and it could take ages to balance high capacity pack if it is very non balanced.
I reccomend balancing series to 0.01v before connecting them and using in ebike.
In the worst case scenario when bms cannot balance cells the differences will expand in time resulting in less range etc.

 

[DrkAngel]

Based on your results ... I would consider using the best 105 only and charge to only 4.05V.
Balance charging often!
But!

I would recommend equalizing all cells to exact same voltage before testing for self discharge.
I used to solder all cells together ... but now I combine using small neodymium magnet discs and thin steel wire. >> paper clips!
Take lowest cells (within .10V) and link together,
begin charging and when charged voltage approaches remaining cells add them, Allow to charge till charged to ~4.20V.
Remove from charger. ...
Important - Cells will not be of equal voltage yet!
Leave cells attached in parallel for a few hours.
Separate and confirm all cells of precisely equal voltage, might be ~4.15V.
Let set several days before rating self-discharge
.001V capability is desirable but not "necessary".

Panasonic NCR18650A specs - View attachment NCR18650A.pdf

Most notable might be the capacity deterioration from their test cycles.
Graph shows 80% rated capacity after 500 cycles.

Charging to only 4.05V and not discharging below 3.20-3.30V (during .5C discharge) should possibly triple usable life.
Should have same capacity after 250 cycles, better capacity at 500 cycles and much much better capacity past 1000 cycles!
Even at the lower charged voltage, might have better usable capacity after 1000 cycles than the 4.20V charged cells have after 500 cycles! ... ?

I would recommend these cells for ≤ .5C continuous and 1C for brief periods.

~3.2V under .5C discharge looks "empty" IMO, might rebound to near 3.50V at rest.

 

[riba2233]

Your test is valid only if each cell had exactly the same charging conditions, which means precise CV voltage, and same cutoff current.

 

[ASK1]

Based on your results ... I would consider using the best 105 only and charge to only 4.05V.
Balance charging often!
But!

I would recommend equalizing all cells to exact same voltage before testing for self discharge.
I used to solder all cells together ... but now I combine using small neodymium magnet discs and thin steel wire.
Take lowest cells (within .10V) and link together,
begin charging and when charged voltage approaches remaining cells add them, Allow to charge till charged to ~4.20V.
Remove from charger. ...
Important - Cells will not be of equal voltage yet!
Leave cells attached in parallel for a few hours.
Separate and confirm all cells of precisely equal voltage, might be ~4.15V.
Let set several days before rating self-discharge
.001V capability is desirable but not "necessary".

Panasonic NCR18650A specs -

Most notable might be the capacity deterioration from their test cycles.
Graph shows 80% rated capacity after 500 cycles.
View attachment 1
Charging to only 4.05V and not discharging below 3.20-3.30V (during .5C discharge) should possibly triple usable life.
Should have same capacity after 250 cycles, better capacity at 500 cycles and much much better capacity past 1000 cycles!
Even at the lower charged voltage, might have better usable capacity after 1000 cycles than the 4.20V charged cells have after 500 cycles! ... ?

Hi DrkAngel,

I have 6 extra cells (my pack calls for 104), so I can definitely remove the 6 weakest cells from the group.

Could you explain to me why you would advise to only charge to 4.05 V? I definitely am on board with charging to a bit lower voltage to preserve battery longevity (something like 20-80% cycling), so I was planning on charging to 4.10 V or thereabouts. 4.05 V though, that seems kind of low. Do you have any estimates for how the charge cutoff voltage affects the capacity of the cells (at the DoD in question)? (Cause I know you have lots of data, haha)

I can definitely charge to 0.01 V precision, but I won't be able to charge in parallel groups as you described. The tools available to me for charging are a 4-cell Foxnovo charger/capacity tester and a 54.6V 13S pack charger. So until I build the pack, I will have to charge the cells individually on this Foxnovo unit.

I love this forum, so helpful.

 

[ASK1]

Your test is valid only if each cell had exactly the same charging conditions, which means precise CV voltage, and same cutoff current.

Yeah, unfortunately I'm using a Foxnovo 18650 charger/tester that can only do CC. I charged all of the cells at 1.0 A. I know this is not the ideal way to charge.

 

[riba2233]

In that case your test is irrelevant. If cells are new and unused you don't have reason to worry, there cells are high quality made in Japan.

 

[ASK1]

In that case your test is irrelevant. If cells are new and unused you don't have reason to worry, there cells are high quality made in Japan.

So you're saying - don't test the cells, they're new and manufactured by Panasonic, so just build your pack w/o capacity or voltage sag testing? Don't I still need to group the cells based on their capacities? Need to read up on that...

 

[riba2233]

Absolutely, they are one of the best cells on the market, with really tight production tolerances. These are not hobby or used batteries. But now you have de-balanced them, so you need to balance them again. You've really wasted time unfortunately.

 

[ASK1]

Absolutely, they are one of the best cells on the market, with really tight production tolerances. These are not hobby or used batteries. But now you have de-balanced them, so you need to balance them again. You've really wasted time unfortunately.

I am well aware of what cells I have. But I did not know if they were all from the same production lot, or how long ago they were manufactured. I had assumed that the cells were unbalanced and that I would have to charge them. My bulk charger was still in the mail, and I had that time to charge them individually (as I assumed would be necessary). No, I'm not an expert. No, I don't always know what I'm doing. But I'd like to get there.

And I'm more interested in learning than having the pack ready as soon as possible, so I'm not worried about making (non-disastrous) mistakes. But yes, now that I know this I can say that it was a mistake.

 

[riba2233]

It wouldn't be a mistake if you've used precise, CC/CV charger, but as you can see your charger didn't charge them all equally, so now you have to do that.
New, brand name 18650 cells come balanced from the factory, you can just assemble them in a pack, and that's one of the good sides of using them.

 

[ASK1]

So apparently my charger is CC/CV. It's just misleading because it displays the CC setting (options are 0.50 A or 1.00 A) even when it reaches CV mode. Not thrilled that I bought it (when the manual is 2 tiny pages in broken English, you've probably f'ed up) but I'm going to at least attempt to balance my cells to 4.15 V using it and a multimeter.

Is there a single-cell 18650 charger (Li-ion capable) that anyone would recommend (ideally with multiple independent slots)?

 

[dnmun]

you can never balance a pack closer than using a set of cans that are from the same lot and date of manufacture. this is the single most important variable controlled and the Bollinger bands monitored on for any of the process steps.

 

[ASK1]

When you connect cells in series & parallel when you're initially building the pack, what's the rule of thumb for the max acceptable voltage difference? I've confirmed that my charger will charge all cells to within +/- 0.005 V.

 

[DrkAngel]

1. Cells connected in parallel should be of equal voltage ... within a few hundredths. ~ .02 -.05V.

2. Determining maximum recommended voltage differential before paralleling cells.
Based on the recommended charge voltage of 0.5C.
Apply .5C charge (1300mA = 1.3A for 2600mAh cell) to a 50% discharged cell ~3.8V and note voltage rise, likely will exceed .05V.
(Base C rate on actual capacity rather than original, 1A might exceed actual .5C)
Keep cells "equal" within this "recommended range" before combining.
Allow all cells to equalize-stabilize before measuring bank voltages.

 

[okashira]

1. Cells connected in parallel should be of equal voltage ... within a few hundredths. ~ .02 -.05V.

2. Determining maximum recommended voltage differential before paralleling cells.
Based on the recommended charge voltage of 0.5C.
Apply . .5C charge (1300mA = 1.3A) to a (2600mAh) cell and note voltage rise, likely will exceed .5V.
(Base C rate on actual capacity rather than original, 1A might exceed actual .5C)
Keep cells "equal" within this "recommended range" before combining.
Allow all cells to equalize-stabilize before measuring bank voltages.

1. I have Model S cells that were paralleled, and a month after removing them, they are still within 0.0005V or less.

2.) I did this calc for Model S cells based on max charge rate (based on supercharging data) and DCIR.
~0.2V differential between cells or less should be safe. Up to 0.3V depending on SOC (lower SOC is better)
Never parallel with greater then 0.3V difference. 0.2V or less preferred.

^^ I think this could be a good rule of thumb for other cells as well. If a cell has higher DCIR (like laptop pulls) they can't support as fast charge rate as higher drain cells.While lower DCIR cells can take a higher charge rate (usually DCIR is pretty well correlated to anode porosity, and anode porosity is the primary determinate of max safe charge rate) but it's not 1:1 thus rule of thumb.