Statistics: Possible results when harvesting cells

SLS BATTERY AUCTIONS are now live at EV, Lithium, LifePo4, Inverters & More


Sep 2, 2017
As promised, here is my thread where I will post the results of my harvesting. Every so and so often the question comes up what is to be expected when reclaiming cells from old laptop batteries. The answer is always the same, it depends and if you are unlucky you will get hardly anything useful. This thread is aimed especially at people new to the business but veterans are also invited to have a look of course :)

Disclaimer: The results here should be taken as one of several possible outcomes when harvesting cells. I'm not trying to prove anything, I'm just providing an example and breaking smaller lots of batteries I get down to individual cells in terms of statistics.
At the moment I get most of them on eBay when I find a cheap offer. A cheap offer at the moment is 2 EUR or less per battery here in Germany, excluding shipping.

This is also quite interesting for me, not because I'm too bothered about success rates, but because I like some good statistics. So I will gladly write this stuff down and maybe someone will find this useful to get a clue of what can be expected when doing this sort of thing. I will also calculate the costs to see what I've effectively paid for these used cells.

I will update this first post as soon as I have new data. If you want to do this as well, feel free to grab a post where you do this. I will put links in the first post so your data can still be found in case this thread gets a bit longer. I will not put up any rules if you want to do this, choose any format you find suitable or just do it like I'm doing.

Here we go:

Batch 1, eBay:
34 batteries, 31 good, 1 noname,2 corroded

Most of these batteries were HP ones and that doesn't matter as such, but one specificthing about them was interesting:
There was a pattern emerging when I checked the voltage of the cells after I freed them from their case. These were 3S batteries, as it is very common, and the first cell had good voltage like 3V and more, the second cell was always lower than the first one and the third one was always lower than the second one. And we are talking about big differences here like 1V between each cell.I guess this is a case of failing BMS's, the battery died well before its time because the shitty BMS couldn't keep it balanced and then just killed it. First tests support this, many of these cells are very good.
But, remember the disclaimer, this doesn't mean anything, we are talking about a very small sample size here and this is not necessary the case all the time with HP batteries.

65 dead cells (0V / strange behaviour / corrosion)
123 usable cells, all of them branded cells, the noname battery had bad cells
188 cells total, 34,6% failure rate before testing,0.65 EUR per testable cell

4 cells didn't make it to testing, voltage dropped below 4V after charging
8 cells with very low capacity
111 tested sucessfully, 0.72 EUR per usable cell

14xx: 5
15xx: 3
16xx: 4
17xx: 15
18xx: 14
19xx: 11
20xx: 24
21xx: 16
22xx: 8
23xx: 6
24xx: 3
25xx: 2
Total: 219.979mAh
Average: 1.981,79mAh
Price per Ah: 0,364 EUR

Batch 2, eBay:
23 batteries, 17 good, 2 noname (surprisingly with branded cells), 1 noname and old (PowerBook G3, around year 2000, Panasonic 17670s), 1 old (again, PowerBook G3, but original, Panasonic 17670s),1 very old (Compaq1400mAh, proprietary cell format) and 1 LiPo

Started taking them apart, few surprises:
The old Compaq battery: Proprietary Panasonic cells, still working! No use for them though, so they will go to the recycling center.
3rd party and original PowerBook battery: Both have the same Panasonics, 17670 format :)
Among the 17 good, branded ones there was another old battery from Toshiba with working Sony 17670 cells.
Both newer noname batteries had branded cells in them as well, one of them was a 12-cell with Samsung ICR18650-26FU. -26F is very common, -26FU is a bit weird and you find only very little about them.

Interesting: Many of the 17670s from the 3rd party PowerBook battery died over night while most of the same batteries from the original battery are working fine. Obviously this might be a coincidence, but it is interesting nevertheless.

8 dead 18650s (0V / strange behaviour / corrosion)
12 dead 17670s (0V / strange behaviour / corrosion)

63 testable 18650s
21 testable 17670s

18650 failure rate: 12.9% before testing (8 of 70)
Not counting the 17670s for the costs, full costs distributed across the 18650s only
0.49 EUR per testable 18650 without Sanyos
0.29 EUR per testable cell with Sanyos, will test them separately to see which can be used

4 cells with very low capacity
6 hot Panasonics, very low capacity as well
53 usable cells for now, 0.58 EUR per usable cell

20 red Sanyos, cold while charging
16 tested ok, 4 had very low capacity
No heat during discharge, will check for heat during recharge

24 red Sanyos, hot while charging
Put them through a cycle, some stopped getting hot, some got warm (so less heat than before), some got hot as before
Will continue to cycle them

Capacities for the 53 tested 18650s:
14xx: 2
15xx: 9
16xx: 8
17xx: 4
18xx: 1
19xx: 1
20xx: 9
21xx: 3
22xx: 10
23xx: 3
24xx: 3
Total: 102.957mAh
Average: 1.942,58mAh
Price per Ah: 0,299 EUR

Capacities for the 20 Sanyos that made it through testing:
14xx: 1
15xx: 2
16xx: 1
17xx: 0
18xx: 3
19xx: 7
20xx: 4
21xx: 2
Total: 38.063mAh + 102.957mAh = 141.020mAh
Average: 1.931,78mAh
Price per Ah: 0,218 EUR

Capacities for the 21 17670s that made it through testing: Most of them were rated around 1200mAh and had between 900 and 1200mAh left

Batch 3+4, eBay, came together from same seller:
45 batteries, 40 good, 4 noname, 1 LiPo
Some Lenovo batteries had a mix of four 18650s and four cells in a proprietary format, looks weird...
The Nonames had Sanyo cells, apart from one and that one had bad cells, so no working noname cells from that batch either.

50 dead cells (0V / strange behaviour / corrosion)
146 usable cells, all of them branded cells
106 red Sanyos, who knows...
22 non Sanyo heaters, also who knows...
315 cells total, 15,9% failure rate before testing,0.74 EUR per testable cell without Sanyos and the other heaters, 0.38 EUR per testable cell with Sanyos

58 cells didn't make it to testing or have very low capacity
88 tested between 1500 and 2600mAh

Needless to say that this is a pretty horrible result. Many bad cells, many Sanyos and usable cells at over 1 EUR per cell.

Gerard Achternaam

New member
Sep 5, 2017
Might as well drop my stats here. I have just started testing and charging and do some extrapolation at the end. Plaintext dump from Soulver:

bought = 40 kg = 40 kg
monies = 45 eur = 45.00 EUR
doa = 223 cells = 223
higher_than_2volt = 154 cells = 154
higher_than_1volt = 102 cells = 102
higher_than_0volt = 163 cells = 163
total = doa + higher_than_2volt + higher_than_1volt + higher_than_0volt = 642

avg_cellweight = 44.5 gram = 44.5 grams
testable = total - doa = 419
failure_rate = doa / total as % = 34.74%
cells_per_kg = total / bought = 16.05 per kilogram
price_per_testable_cell = monies / testable = 0.11

cap_tested = 30 = 30
higher_than_2000 = 9 = 9
higher_than_1800 = 6 = 6
higher_than_1700 = 3 = 3
below1700 = 12 = 12
charged = 97 = 97
hot = 23 = 23
vdrop = 26 = 26
total_tested = cap_tested + charged + hot + vdrop = 176
hot / total_tested as % = 13.07%
vdrop / total_tested as % = 14.77%
dropout = (hot + vdrop) / total_tested as % = 27.84%
price_per_alive_cell = monies / (testable - dropout) = 0.15
useable_cell_rate = higher_than_2000 / cap_tested = 0.3
price_per_usable_cell = price_per_alive_cell / useable_cell_rate = 0.50


Sep 2, 2017
I initially kept cells with more than 1500mAh, but since this was a very consistent batch in terms of minimum capacity I kept these five 1400s as well. These were the lowest apart from only eight 800-1000mAh cells I chucked away. These are perfectly usable cells, perhaps not for bigger applications, but certainly for smaller stuff and for testing. I have a separate box where I keep those, they will go in there.

Also I wanted the statistics to be as detailed as possible. If someone wants to apply a higher threshold then he can easily do that and take the lower capacity cells out of the equation.
I will consider adding "Price per Ah over average" or over a certain value though.


Sep 13, 2017
Here is my Status on salvaged cells that was not dumped due to: 0 Volts, less than 1500 mAh tested capacity, getting hot during charge (a lot of Sanyo cells, that I may put up for sale), and finally a few with a bump in them indicating that the pack has been mechanical abused.

Today my database has 263 cells totalling 564 Ah. That is 2164 mAh average.

A long way to go to 14s80p :-/



Sep 2, 2017
Preliminary results for Batch 2 are in, I've yet to deal with 50 Sanyo and Panasonic cells that got hot during their first charge


Sep 2, 2017
Pretty horrible results for 3+4

This is the end of my little experiment for now. I have enough cells and if I need more I'll think about buying new ones.