Since I've already put in the effort learning for myself, here is a quick guide of how I recover 18650's from laptop battery packs. This is also as a good chance to get some feedback on my method. In ~1k cells I have yet to have any cells catastrophically fail on me. If I'm doing something right I'm keen to share and if I’m just lucky I’m keen to prevent an incident. Elsewhere on the Internet it seems that there is a lot of fear and understandable caution when recovering cells. The most basic of instructions I read online are to generally throw out cells below some nominal voltage, I think this is both wasteful and not safe enough. Many of the bad cells I have identified were salvaged above 3.5V or higher and holding charge. With a variety of cell chemistries available in the 18650 format and the uncertainty of recovered cells past lives (bad cells can be charged too) I think it is better to be informed and actually fully test the cells rather than just rely on general rules.
I have tried to present evidence based methods, so readers can be reasonably confident that they recover safe and reliable cells without having to just trust me. Where I cannot find any information or have not taken the time to I have made a few assumptions, (Usually made obvious with sentences containing the phrases with "I believe", or "presumably"). The links to the references are at the bottom.
Recovery procedure:
[1] Choosing good 18650 battery packs [2] Remove cells from battery pack. [3] Test cells resting voltage and write on cell. [4] Fully charge cells to 4.2V and let them set for 2+ weeks. [5] Cycle batteries Charge/Discharge/Charge [6] Measure internal resistance
[1] Choosing good 18650 battery packs
Used laptop battery packs containing 18650 cells are often quite cheap and sometimes free if you look. It's not that they are a bargain -- used cells are just not worth that much, most of the value is actually added from the hours spent recovering cells. Often it may be better to just buy the cells depending on your application. However, with 18650 cells in high demand when buying small quantities there are a number of fakes and deceptively advertised cells. Name brand laptop battery packs offer a source of reliable genuine cells on a budget. Given the multitude of cheap used laptop battery packs available, here are some things I think about when deciding to pick up a bunch of batteries.
[1.1] What cells are in the pack?
Newer cells are of higher capacities and have had less time to age.
Li-ion technology has advanced greatly since the late 1990's and newer cells are likely to be safer and of a higher quality.
I wouldn't bother opening super old packs unless you don't have many options or are just bored.
You can usually find some information on the manufacture date, cell type and mAh rating on most battery packs.
Many packs also have small vent holes under the sticker,
through these vents you can usually see the colour and some of the cell markings.
This information is usually enough to quickly tell if the pack will be worth ripping apart.
Also If you are digging through a bucket of packs, some packs have a button to display the batteries
remaining capacity on LED’s.
This is a quick way to get some good cells.
If you press this button and any lights come on you likely have a good/fresh pack.
If none come on however, this is not an indication the pack is bad.
If batteries are unused new old stock is also something to consider,
this does not guarantee that they are good cells they may have been sitting around for years
self discharging while new improved cells have come out.
New old stock is often seen a source of reliable cells and sells for a large premium online.
However, I'm not sure if this premium is really deserved.
I've seen some pretty crappy "new" batteries, which are packed like and look like new stock but with cells in pretty poor condition.
[1.2] Number of cells in the pack.
Opening larger packs is a much better use of your time, opening a 9 cell pack takes about the same amount of time to open as a 3 cell
yet it offers 3 times more cells.
Constructed with multiple paralleled cells, per-cell charge and discharge amperages should be lower in these extended battery packs.
As treating cells gently is known to increase their life.
I believe these larger packs probably contain healthier cells.
[1.3] Pack construction
Opening multiple packs of the same style is often easier than trying a random assortment of packs as you learn the best way to get inside.
Identical packs also offer identical cells which are much easier to match in your final end use.
If a pack is too time consuming to open, don't bother wasting your time getting more of that style.
[2] Remove cells from battery pack.
Removing the cells from the battery packs takes some manual work as the plastic laptop battery packs are ultrasonically welded or glued together. It takes a lot of prying. A flat screwdriver and some needle nose pliers have been all I have needed to open every pack. The pliers are very useful in removing the welded battery interconnects. You are bound to scuff up a few cells and they may need some heat shrink or electrical tape. If you physically damage any cells, mark them for disposal.
[3] Test cells resting voltage and write on cell.
When removing the cells from the battery packs it is a good thing to measure their salvaged voltage. For my cells I measured their salvaged voltage and binned them into 0-0.25V, 0.25V-1.75V, 1.75V-3.25V, 3.25V-4.2V. In my experience good cells can identified in every bin. However, I do still maintain a little caution with the cells which have been resting at very low voltages. It takes a little extra work and some brain power to recover low voltage (<3v) cells, but testing is better than guessing especially when assessing safety. It is not uncommon to read recommendations to throw away all cells below 3V or even higher. Without some cheap testing equipment this may be a seem reasonable but voltage is not the whole story.
[3.1] 0-0.25V Cells
Cells sitting at 0V most likely need to be disposed of, even the ones that can be recovered.
Cells may be at 0V for a few reasons, over discharge, Internal damage or CID trip.
Measuring the continuity and resistance across the cell is a good way to see what's happening with these cells.
If the cells are open circuit it is likely that the CID [Link CID] has tripped.
It is possible to reset the CID and obtain a functioning cell with a good voltage.
However as the CID was likely triggered during exposure to extreme heat, the cell is potentially damaged and may be unsafe to use.
I have no information on if a manually reset CID would perform its function again reliably,
right now I am asuming it is not worth the risk.
If there is continuity across the cell, the cell has probably been fully discharged or fully self-discharged.
This over discharge can occur for many reasons and some cells may accept charge and be recovered while some may not.
Even if a cell which was sitting at 0V can be recovered, its remaining capacity is usually poor.
These cells may be damaged internally and at risk of failure.
Cells stored at a very low state of charge or deeply discharged may experience corrosion of conductor
resulting in a loss of capacity and growth of impedance (internal resistance will also grow).
I have disposed of these out of specification cells even if they appear to to work as they are so extremely discharged.
[3.2] 0.25V-1.75V
Cells have some remaining voltage and some may be recovered safely whilst others not.
As laptop battery BMS systems should make the battery inoperable when serious under voltage cells are detected
the maximum number of extreme over discharge cycles in a laptop battery should only be once (I think).
Thus it is my guess that their low voltage is not necessarily a sign of repeated abuse.
With a good intelligent charger some of these cells may be restored close to their nominal capacity, many may not.
Despite their recovery these “good” cells still may have acquired some internal damage from over discharge,
and they may be more likely to fail in the future.
Below 2.5V Li-ion cells are vulnerable to internal damage, this damage is often proportional to the number of over
discharge cycles and the time spent at this voltage.
Tests of some cells show that over discharges to 1.0V or below does not always affect all battery chemistry significantly.
With longer times spent at low voltages (>1 week) copper dendrites can grow which result in elevated self discharge rates.
This is bad for safety (risk of internal short circuit) and limits the use of these cells to applications with short cycle times.
Given low voltage conditions copper dendrite growth presents a risk of internal short circuits,
this risk can be indicated by a high self discharge rate.
Low voltage cells with high self discharge and internal resistance values deviating from their nominal levels are not worth taking the risk.
As all manufacturers use subtly different chemistry some cells are more resilient than others.
I am very cautious about reusing these cells and only do so if their internal resistance, self discharge and total capacity remain acceptable.
Additionally It takes no effort to go onto Google and check if others have found that if the exact cell is robust at low voltages.
I am undecided if it is worth putting in the extra effort to recover these cells.
Many do retain ~90% or more of their original capacity with no signs of age.
Currently my recovered cells from this voltage range below 1.5V are set aside unused as 2nd quality cells,
I am staying on the side of caution for now.
[3.3] 1.75V-3.25V
Li-Ion cells discharged to lower than ~2.5V are still candidates for internal damage and will likely have reduced capacity.
It is important to have an intelligent charger for low voltage cells and be especially wary of signs of cell ageing.
With these over discharged cells I presume the most likely reason for the particularly low voltage is from self discharging for a long time.
Monthly 1% of the full capacity of a cell may be removed by self discharge and potentially more again removed by the BMS system.
Above 2.5V cells are close enough to the normal charge cycle that they are good candidates for recovery with,
depending on their history, minimal damage.
[3.4] 3.25V-4.2V
There is not much to immediately do with 3.25V+ cells.
They are are sitting at a good voltage for storage and unless just pulled from use must not be self discharging very quickly.
Always be prepared for weird cells but most of these should be good.
[3.5] Note
Pay attention to battery chemistry.
Here is a short intro to get you started (It is super basic but does not tell you much more than that different chemistries exist).
There are a few odd cells out there in the 18650 format with different charge/discharge cuttoff voltages etc..
If you have many identical cells it is certainly worth looking their specific chemistry to make sure you are doing things correctly.
[4] Fully charge cells to 4.2V and let them set for 2+ weeks.
Charging all the cells puts them into a healthy voltage range and hopefully stops any further low SOC damage from occurring. In the case of my cells they were fully charged to 4.2V at a current of 500mA. This charging current is reasonably low providing an optimal and realistic mAh rating for battery banks. There is much written on the Internet about charging Li-ion batteries and I have glossed what happens. If you get a good analysing charger (This site is an awesome resource for 18650 chargers) it should take care of all of the hard work for you. While charging, cells which got hot were marked as bad, this is likely a sign of too high internal resistance. These cells were then left to sit for further investigation and eventual disposal. Successfully charged cells were then left to sit for 2+ weeks to self discharge. Cells were stored indoors ~15-25°C for 2-6 weeks before re-testing. After 2+ weeks cell voltages were remeasured cells below 4.15V were put to the side. At two weeks a cell sitting just below 4.15V has lost ~5% of its capacity. This pretty much ensures the cells will be good for any energy reasonable storage application. 5% in 2 weeks is quite a conservative cut-off and the cells I have put to the side for now are probably all good. However, with most cells passing this test eliminating a small number of anomalous cells is probably good for safety and reliability. I will further refine this value as I test more cells.
The commonly accepted self discharge rates for Li-ion cells are given for two regimes: immediately after charging, weeks/months after charging. A typical cell looses 4-5% in the first 24 hours and 1-2% per subsequent month. Potentially more if connected to the Battery/BMS system. These common values seem pretty conservative for new cells with most new cells out performing this. Typical self discharge rates for 18650's depend on both battery chemistry and condition. The self discharging link contains an in depth analysis of 18650 self discharging including destructive testing of the cells. In this study sets of three new 18650 cells from two different manufacturers were fully charged and left for 14 days. At 14 days these three sets of brand new cells had discharged to 4.16V and 4.17V.
[5] Cycle batteries Charge/Discharge/Charge
Again at 500mA per cell my cells were tested for capacity using OPUS BT C3100 V2.2 and LiitoKala Engineer Lii-500 chargers. These chargers drain cells to ~2.9V. One thing to note is that using a charger which records charging current without a Charge/Discharge/Charge function is not as accurate. Internal resistance may result in a higher charge reading and most importantly recovered cells may be in varying states of charge. Chargers with a proper Charge/Discharge/Charge loop will make battery measurements directly comparible as they will all experience the same conditions/voltages.
[6] Measure internal resistance
There are a number of ways to measure battery internal resistance. It is important to note that this is a very small resistance value and unless you correct for the resistance of your probes, device, and have a good connection quality you will get strange results. Not wanting to put in too much effort for my cells I just made sure the charger I bought was reasonably accurate at measuring this (most chargers are not). For my cells internal resistance was calculated as the average of two values from two different positions in my OPUS BT C3100 V2.2 charger, minus the approximate resistance of the charger ~30mΩ. For an analysis of the OPUS BT C3100 V2.2 accuracy when measuring internal resistance see the link. Internal resistance is a good measure of battery condition and needs to be performed, especially if you are trying to confirm cell safety. This value is extremely measurement sensitive and whilst extreme internal resistances are markers of bad cells it is quite variable. Secondly remembering Ohms law V=IR, the voltage drop across a resistor is proportional to the current, the more demanding your application the more this internal resistance will matter. In high draw applications cells in batteries with badly matched internal resistances will need much more aggressive balancing and may fail much earlier.
Tom's electronics pages / tom@mmto.org