The Li-Ion battery in my IBM Thinkpad X31 laptop has been getting a bit long in the tooth. Rated at a 4.4 AH capacity new, /proc/acpi/battery/BAT0/info told me that it was only holding 1.8 AH now that it was a few years old. (This is normal for Li-Ion batteries, which degrade over time, even without multiple charge-discharge cycles.)
A brand new IBM (or Lenovo now) battery costs over a hundred dollars, but by shopping around I was able to find a “compatible” battery for as low as $50. It was only rated at 4.4AH, but that is relatively close to the 2nd generation 4.8AH batteries that IBM/Lenovo sell for twice as much. I started to wonder if it might be cheaper to buy OEM li-ion cells and simply replace the cells (keeping the case, and charge/discharge electronics). The first step would be to determine what type of Li-Ion cells I’d need to buy, so I decided to open up my old battery.
As you can see, the standard X31 battery has six cells, in three parallel groups of 2. Cells are nominally 3.6volts, so this adds up to 3 x 3.6 or 10.8 volts. In the photo I have removed the shrink wrapped packaging from one cell to view the markings. Note the relatively complicated PCB along the back side of the cells that handles charging and discharging. If you zoom into the photo, you can see that the controller PCB is connected to each end of every pair of cells (orange and black wires to the far ends, silver metal tab connections to the middle two). This allows the controller to charge each parallel pair of cells at a different rate. The controller PCB is also connected to a thermocouple that is resting between the two middle cells. This gives the controller a temperature reading on the cells during charging and discharging. If the PCB detects that the temperature is too high, it can lower the charging rate, or shut down the power draw (and laptop). Also note the heat fuse (small white block in series with the power line between the two leftmost cells) that is designed to open the circuit if the charge/discharge controller for some reason fails to maintain a safe temperature. All of these safeguards are designed to keep your laptop battery from igniting, and will be very important to maintain in any “re-manufactured” batteries.
I actually determined what type of li-ion cells were used by measuring them and then looking for li-ion cells of a similar size. They are about 2.5″ high by 11/16″ diameter, or very close to the 64.9mm x 18.3m diameter size of a 18650 style cell that I found on www.batteryspace.com.
My battery is rated at 4.4AH, or 4400mAH. As it has three sets of parallel cells, each set of two cells must have a 4400mAH capacity (because they are in series, you add the voltage, not the amperage), so each cell must have a 2200mAH capacity.
From a mAH per dollar standpoint, batteryspace.com’s 2000mAH cells are the best value, but I decided that since the total cost difference was only six dollars, I could afford purchasing the 2200mAH cells. These are slightly lower than the 2400 mAH capacity of the newcells that Ibm/Lenovo now use, and I’ll end up with a 4.4AH battery (just like the original part number, before IBM/Lenovo upgraded it to 4.8AH).
Just the cells cost $34.20, but I chose to purchase them with solder tabs attached (an extra $1.50) because I figure the people at batteryspace.com are better than me at attaching tabs, plus having extra tabs to work with (those that come on the cells, plus those I salvage from the original cells) will make my life easier. (Besides, the $7 of shipping is the largest extra expense….if the 2600mAH cells weren’t almost twice as expensive as the 2200mAH cells I’d have gotten them just get a 5.2AH capacity battery!) The total cost was $43.42. A week later my batteries arrived. (Thanks UPS!)
Important safety note!
You should never replace li-ion cells with cells that have a lower capacity rating, or charge/discharge rate rating. The electronics in laptop batteries are programmed to prevent the cells from overheating and catching fire, and do a very good job as long as the replacement cells in the battery are of equal or higher rating than the original cells.
Here I have two Li-Ion cells ready to solder together into a parallel pack. The technique that I found to work best was to insert a coil of small electronic solder between the tab and the bottom of the next battery, making sure that the solder never overlapped, so that their was always only one thickness of solder thick at any particular place. Then, I’d hold down the end of the tab with a screwdriver (to make sure it didn’t bounce back up while the solder was still hot) and heat the tab with heavy downward pressure from a heavy duity (stained glass) soldering iron for a few seconds. I’d press down until the solder visiblely melted (usually accompanied by a puff of resin smoke) and the tab settled down. Then I’d remove the iron while keeping pressure with the screwdriver.
Important safety note: At the factory, tabs are welded to li-ion cells very quickly. Li-Ion cells will catch fire explosively if they are overheated. It is very important that you do not leave a soldering iron on the cells long enough to raise their internal temperatures to dangerous levels. Heat the tab, melt the solder, and then remove the iron.
You will note that I used one set of tabs to attach the two cells, and the other set of tabs was positioned 90 degrees off axis, to be used to connect to other packs of cells. Note that I’d use tape to hold the cells closely together while soldering them, but would remove the tape later to make them fit into the battery holder.
Here is my first pack, soldered to the charging electronics. Note that I have temporarily covered the exposed positive ends of these cells with clear plastic tape.This tape insulation is very important! It keeps this pack of cells from accidentally shorting through the pack of cells that I will add next. The process involves laying the cells on top of each other while soldering, and it is very dangerous to short out li-ion cells. (All of the charging electronics and fuses in the battery are designed just to keep that from happening.) Note that other things on your workbench can short out li-ion cells. For example, a coil of solder conducts electricity very well, at least until it melts. So do screwdrivers and pliers, and they are less likely to melt. Li-Ion cells are usually stored and shipped 40% full to prolong their life, and the ones I had certainly did spark a few times. A small spark here or there won’t hurt anything, but if you short a cell for longer than a moment, it will begin to overheat and you could get into a runaway situation where it eventually catches fire explosively. Don’t do that!
Once I’ve safely insulated the ends of these cells, I add the 2nd pack and solder them together.
After the tabs are soldered together, you can unfold the 2nd pack and lay it down flat.
When soldering existing metal tabs from the charging electronics to the cells, I found it best to coat the tabs with solder first, and then attach them to the cells. Below you can see a pre-existing tab that I pre-soldered before attempting to attach it to the back of the cells.
Note that the first and second pack have a good space between them (by design) in the battery holder, which leaves plenty of room for the solder tabs to be folded between the two packs of cells. However, the second and third packs must be mounted much closer together. Because I did not have a custom manufactured set of solder tabs, I found it easier to just butt the batteries up against each other and join the solder tabs above the cells.
Now, I have all three packs connected in serial, and to the charging electronics. All that remains is to find some very sticky double-sided tape to affix everything to the interior of the battery case. I suggest you use carpet-tape (designed for tacking down carpet), which appears to be about as strong as the original tape I removed from the case when disassembling it.
Once the tape is in place, all that remains is to snap the two halves of the battery case together, being careful to get all of the tabs in the right slots. Everything snaps together with a satisfying “clicking” noise, and the battery is as good as new. If you do not remove the manufacturers sticker (as I did) and are careful to use a plastic tool and not leave scrape marks when prying the battery apart (as, once again, I did) the battery will look practically like new.
But how does our existing charging charging circuitry deal with the brand new li-ion cells?
When I first plugged the battery into my laptop it read as completely discharged (even though I KNOW that the cells were shipped partially charged) possibly because the electronics had been unsoldered and lost power completely. I expect they default to a completely uncharged state if they ever loose power completely. An hour later the battery reported that it was almost halfway charged. (My laptop batteries normally take around two and a half hours to charge, getting the first 90% of the charge in two hours, and the remaining 10% a bit more slowly as the charging electronics “top up” the cells.)
At this point I went to bed, so of course I unplugged my laptop because I didn’t want a newly constructed battery charging unattended. (I know what I’m doing, but it doesn’t hurt to be careful…)
The next morning I plugged the laptop in and after 45 minutes it had reached 99% charged. I believe that this was because it had run up against the “last full capacity: 16880 mWh” setting that was stored by the charging electronics. It stayed at 99% charging for well over an hour, charging at a reported rate of 3109 mW with a voltage that ranged near 12.5 volts ( 12487 mV). Most Li-Ion cells reach full capacity (and regulated chargers usually stop charging) at 4.2 volts (4.2 * 3 is 12.6 volts), so it looks like the charging circuit got the cells very close to a fully charged state, even though they had more capacity than the electronics were expecting.
Once the battery hit 100 % (fully charged) I checked the /proc/acpi/battery/BAT0/info file. It reported that the “last full capacity:” was still 16860 mWh (I was hoping for something closer to 44000 mWh). The interesting thing is that the battery used to report a “design capacity: 44000 mWh”, but now it reports “design capacity: 47520 mWh” which is an improvement in the correct direction.
I ran the battery completely down (which took 2 hours and six minutes) by using the laptop normally (but with full screen brightness) for an hour, and then running a script to use 100% of the CPU for the next hour and six minutes. This is about the same battery life I would expect for a brand new 4.4AH battery (I expect 2.5 hours of power when not running the CPU and back-light at 100%) so it looks like the cell transplant was successful.
However, the power monitoring electronics in the battery took a while to get used to the new cells. On the first discharge the ACPI based battery gauge dropped rapidly in the first 45 minutes, and then stuck at 6% remaining for the next hour and ten minutes. However, the laptop BIOS (which controls the color of the power LED) had a better idea about how much life I had left, as the green power led didn’t turn orange and start flashing until near the 1 & 1/2 hour mark. Near the end of the discharge period I noticed that the reported “last full capacity” had raised to 20230 mWh.
Over the next several discharge/charge cycles the “last full capacity” slowly raised in approximately 4000mWH increments.
original: 16860 mWH
1st: 20230 mWH
2nd: 24220 mWH
3rd: 29040 mWH
4th: 34750 mWh
5th: 41660 mWh
It appears that the electronics are unwilling to raise their view of the cell’s capacity by more than 10% of the design capacity per cycle. The battery has now stabilized with a reported capacity just over 4.1 AH. This is 0.3 AH less than the new capacity of the Li-Ion cells, but it’s what I would expect as the charge/discharge electronics are slightly conservative and you can get a few extra minutes of power out of the battery after it reaches “0%”.
Knowing what I do now, I could repeat the entire procedure in 2 to 3 hours, so the $7 savings over buying a $50 4.4AH “compatible” battery isn’t really justified. However, if you spent $65 for the 2600 mAH cells, you could have a battery that has 8% more capacity than the $100 to $120 official IBM/Lenovo 4.8 AH battery packs (and 15% more capacity, or 22 extra minutes, than my $43 battery) with the same amount of work.
18 month update: Current battery status.
Some charge control circuits will “shut down” if the existing cells are removed (similar to inkjet cartridges that refuse to be re-filled) so you may want to solder your new cells into a pack, connect it to your circuit in parallel with the old cells, and only then cut off the old cells. Think of this as Indian Jones sliding the statue off the pedestal while at the same time replacing it with a bag of sand…hopefully with better results.
The charge control circuit in my (after-market) X31 battery did not have this problem, and didn’t mind not having cells attached for an hour or two (it started right back up), but others have reported that some battery charge circuits will detect a complete removal of cells and stop functioning.