Project result report: Re-celling 8-cell X6x (non-tablet) battery

[flyingfishfinger]

Hi all-
Figured this was worth at least putting in its own post. I recently decided that I would try to re-cell my 4 & 8 cell X61 batteries; both of which were degraded to the point of complete uselessness. Here's a short & quick summary of what I did and the results. Feel free to ask for details on any part of it

Equipment:
- Cypress FX2LP CY7C68013A Dev Board from Ebay
- Victor's excellent blog & project here and here
- New NCR18650GAs from Liion Wholesale Batteries. Discharge-tested to 2.8V and achieved something like 3450mAh, so they seem to be pretty good cells.
- Soldering station & lots of flux
- Old 4 & 8 cell batteries

Process:
- Extracted original firmware (you only need the EEPROM part) from both packs according to Victor's blog. Pretty flawless process, except for one gotcha and a minor hiccup (reading out the correct "magic byte" to unlock the flash worked as expected but this value is different for each battery; and a minor mismatch between the instructions and build versions)
- I accidentally bricked my 4-cell by flashing the dump from the 8-cell battery to it without saving the 4-cell firmware. Name your files properly and don't overwrite previous readouts!
- Follow blog instructions on parsing the EEPROM dump and modifying necessary parameters. I zeroed out the cycle count and set the capacity to 7000mAh.
- Unsolder triggering leg of the weird 3-pin fuse device, just in case the controller decides to try and blow it after new cells are installed
- Replace cells. I do not have a tab welder, so I made the possibly risky decision to solder directly to the cells (topic for a different discussion) but re-used all the connecting tabs from the old pack. Removing tab-welds is quite interesting, I ended up using a small screwdriver as a chisel and broke each spot, more or less preserving the strips for reuse.
- In the 8-cell pack, batteries come in two parallel pairs of two, and one unit of 4. I snipped out each of these three units separately and redid each one; then reinstalled them to the controller board. Be careful to insulate each metal piece and wire you unsolder; even the old dead cells arced a gash into one of the tabs when I accidentally touched something.
- I pretty much wrapped any exposed metal part in a layer of Kapton tape for safety
- Once re-celled, the pack won't be "alive" even if the cells are charged. To activate it, short the positive side of the cell pack to the positive output pin of the main connector
- Reflash only the EEPROM section with your modified dump
- When you first plug it in, the battery monitor / whatever app you're using to watch it will most likely be confused.
- Turn off your machine and let it charge overnight, even if the cells are full. That'll probably help it recognize the new pack

Results:
- Overall a great success. I haven't fully tried all use cases yet, but my X62 idles at around 8-9W and I ran it for over 6.5 hours and used maybe 3/4 of the capacity according to Batter Bar (!!!). Then I had to go home so I turned it off and charged it. More results later.
- Sometimes Battery Bar is confused. It can reduce the effective reported capacity, but not increase it. I also don't know what the controller uses for a low battery cutoff voltage, so I'm not sure what I should set the "new" capacity to. If 2.8V, 7000mAh should be close to correct, but if it's higher I'll have to redo my single-cell discharge test to get a more accurate value. Does anyone know what the cutoff value is?
- The cells originally all used tab welds. Since I soldered my new ones, I couldn't quite get them as flat as they used to be. The case is pretty tight, so I ended up not being able to properly close it. On one side I just decided to cut a hole in the case, the back row of 4 cells protrudes by less than 1mm. It's enough to not snap closed, but not enough to actually stick out. On the "underside" set of 4 cells, something is too tall and my assembly makes the pack bulge a bit. It sort of fits into the laptop, but I might want to try and reseat them / redo some of the internal taping.

Unfortunately I did not take any pictures during the process as my mobile with a decent camera is currently broken, and I didn't bring a regular camera to work. I can post pictures of the completed assembly later with some close-ups, but I don't really have anything from the soldering / breaking cells apart section of the project.

I do want to acquire another old / broken 4-cell to redo, since I like the shape of an X61 without the protruding 8-cell... Then I can post up pictures, hopefully.

Happy to answer any questions, and sorry for the lack of pictures again
Cheers,

Rafael

[fourthree]

Fantastic work. Any reason you couldn't generalize this to other thinkpads? (Extra credit: ultrabay battery re-celling!)

[sheinz]

Thanks for sharing your experience!

I also repacked my battery for x230. But when I did it I didn't know that it is possible to reprogram a controller. So, I did a cells "transplantation" without powering off the controller. The controller didn't lock up and the battery was working fine. Next I tried the battery gauge reset so the battery would learn a new capacity. The reported capacity increased from 43Wh to 82Wh, but not to the designed capacity of 93Wh. But I know that the real capacity is about 100Wh. When the battery reports 0% the laptop still works for several hours. So, there might be some limit how far can the controller adjust the capacity during learing procedure.
I'm waiting for the adapter board to arrive and will try to fiddle with the eeprom.

And a word of caution to everyone who wants to go the same route. Figure out what is the max. voltage of the installed cells in the battery and order the same. I did this mistake. The original cells were 4.3V but I installed 4.2V. There might be possibility to change max. voltage configuration in eeprom but it is unknown for now.

Happy re-celling

[flyingfishfinger]

When the battery reports 0% the laptop still works for several hours.

Huh, interesting. I actually experienced the exact same thing. From having changed the capacity in the EEPROM, I know that the controller can decrease the "known" capacity by 10-20% per charge cycle, but that still doesn't explain why it lasts past 0%. I assumed in that case it should increase the capacity but according to Victor the controllers are pretty bad at increasing.

The reported capacity increased from 43Wh to 82Wh, but not to the designed capacity of 93Wh.

Same here. "Design capacity" is supposed to be 107Wh, but with new cells which I assumed to have 3500mAh, that's only around 102Wh. Given that the original cells supposedly had only 2800mAh, I have no idea why the design capacity is so high to begin with...

R

[RealBlackStuff]

My X201 has a 9-cell 42T4696, made by Panasonic.
Mfg date: 2010-06-22
First use: 2010-11
Cycles: 279
Design : 84.24Wh
Current: 84.33Wh, that is MORE than original design, after almost 7 years of use!

[sheinz]

Current: 84.33Wh, that is MORE than original design, after almost 7 years of use!

That is incredible result!
Did you use some techniques to prolong battery life? (like limit full charge using charge thresholds)

[RealBlackStuff]

No special settings, just default from whatever is in BIOS and in the W7-Pro (32-bit) Power Mgmt program (V.6-something).
I use it until it's about 10%, then plug it in overnight.
It's my travel companion and gets otherwise used daily, when I feel like it on the couch.
It's just a dammed good battery.
To quote Mercedes: The Best, or Nothing.

[E350]

RBS: Where can you buy Panasonic batteries for our Thinkpads?

My X201 has a 9-cell 42T4696, made by Panasonic.
Mfg date: 2010-06-22
First use: 2010-11
Cycles: 279
Design : 84.24Wh
Current: 84.33Wh, that is MORE than original design, after almost 7 years of use!

[Thinkpad4by3]

My X201 has a 9-cell 42T4696, made by Panasonic.
Mfg date: 2010-06-22
First use: 2010-11
Cycles: 279
Design : 84.24Wh
Current: 84.33Wh, that is MORE than original design, after almost 7 years of use!

7 years of use and only 279 cycles???? I have a feeling that thing rarely leaves the charger.

Also, panasonics happen to have an extremely low cycle capacity loss of 70% @ 500 cycles so there is probably 85% of the initial design capacity left, of which coincidentally happens to be more than the actual design capacity.

[vavet]

Interesting, thanks for sharing the experience.

I have recelled my x61 battery with Panasonic 3600mah batteries.
The original board got locked in the process as expected, but I then I tried a board from a dead cheap ebay battery, and apparently those do not get locked no matter what.
Got about 4-5 hours from a 4cell battery. Using this battery on a daily driver for about 2 years. Now it holds about 2-2.5 hours.

Not sure if all ebay battery boards have this but ones I tried back then did.

[fourthree]

How's the battery holding up after half a year or so? Any more statistics (wifi, brightness, cpu usage, etc)

[flyingfishfinger]

Hi,
Well, I don't treat it very well to begin with so it's lost a lot more capacity than if I had been "nice" to it. This laptop spends most of its time plugged in, but I never bothered to change the full thereshold so it's always at 100% (yes, go ahead and judge me. An oversight, if you will). Also, ThinkPad Power Manager doesn't even detect this battery and I don't know where to change the setting outside of that.

On the few occasions I do actually use the battery, I get several hours of use. The remaining capacity is quoted by Battery Bar as 93,946mWh (out of a possible 107,827mWh), which is 12.9% of wear.I'm pretty sure that it told me my actual capacity was lower than the design capacity from the moment I did the rebuild though, so i'll have to do the math with the actual battery capacity again and see which one is correct (did it before, forgot what the vaues were). That probably has something to do with how the theresholds and capacity numbers were reset when I reflashed the battery gauge though, since even the author of the hack doesn't fully know which bytes do what ("Zero out this set of bytes one by one until the remaining capacity resets" or something along those lines).

I've got my CPU power management pretty dialled in; it spends a good amount of time in C7 when I'm not doing anything. At lowest screen brightness with Monty's LED backlight and a WiFi connection, the machine idles around 11W and I've seen it as low as 8W and as high as 15W while the screen is on.

I downloaded & watched a 2hr movie last night with VLC and used roughly 30% of the battery, as far as I can tell.

So I'd say for what it is, it's holding up pretty well but I could probably take care of it better.

Also, does anyone know how to change the charge theresholds without using the Lenovo Tools in Windows 7?

[axur-delmeria]

Also, does anyone know how to change the charge theresholds without using the Lenovo Tools in Windows 7?

I've set up a small SSD (32GB) in a USB enclosure, with Debian installed. Once I boot from there, I can set the charge thresholds using TLP.

[ndandanov]

A very interesting topic! Has anyone managed to successfully re-cell the battery of a newer ThinkPad, e.g. from the xx40 series or newer?
Thanks!

[packardbell]

would love to see them pictures of the process of recelling. Just can't trust what you buy nowadays, even if it looks legit and all internally you cannot see it.
posting from Canada today.

[Pardalis]

Thanks for sharing your experience!

I also repacked my battery for x230. But when I did it I didn't know that it is possible to reprogram a controller. So, I did a cells "transplantation" without powering off the controller. The controller didn't lock up and the battery was working fine. Next I tried the battery gauge reset so the battery would learn a new capacity. The reported capacity increased from 43Wh to 82Wh, but not to the designed capacity of 93Wh. But I know that the real capacity is about 100Wh. When the battery reports 0% the laptop still works for several hours. So, there might be some limit how far can the controller adjust the capacity during learing procedure.
I'm waiting for the adapter board to arrive and will try to fiddle with the eeprom.

And a word of caution to everyone who wants to go the same route. Figure out what is the max. voltage of the installed cells in the battery and order the same. I did this mistake. The original cells were 4.3V but I installed 4.2V. There might be possibility to change max. voltage configuration in eeprom but it is unknown for now.

Happy re-celling

I see that both you and this thread are still active, so if you don't mind a few questions:
-How did you go about opening the battery? I've gone over the entire seam around it but can't find the bit that either clips or glues it all closed.
-Did you transplant by connecting the new set in parallel, letting the charge balance and then disconnecting the old set?
-Did you ever get the eeprom adapter? Do you consider it necessary for this if all you want is some extra juice in a very used battery?

I'm looking into recelling my X230's 44+ since BatteryInfoView reports 5000 mAh designed capacity and 1700 mAh full charge capacity.

[sheinz]

I see that both you and this thread are still active, so if you don't mind a few questions:
-How did you go about opening the battery? I've gone over the entire seam around it but can't find the bit that either clips or glues it all closed.
-Did you transplant by connecting the new set in parallel, letting the charge balance and then disconnecting the old set?
-Did you ever get the eeprom adapter? Do you consider it necessary for this if all you want is some extra juice in a very used battery?

I'm looking into recelling my X230's 44+ since BatteryInfoView reports 5000 mAh designed capacity and 1700 mAh full charge capacity.

I think opening the battery without ruining the case was the hardest thing
It takes time. The cells inside are glued to the case.

I prepared the new cells by charging them to the same voltage. It can be easily done with a lab power supply. Just measure the voltage of the old cells and set this voltage on the power supply to charge the new cells to this voltage.
If voltage of the new cells is much different than the old ones and connected in parallel huge current might flow to charge uncharged cells.

I did get an eeprom adapter. I read the eeprom, modified the capacity and wrote it back. At first the system shows my set capacity, but after a few cycles the battery overwrote this value. The battery measures charge/discharge curves of the cells and calculates the capacity by itself. As my cells are different they have different "curves" and controller wrongly calculates the capacity. Theoretically if swapped with identical cells everything might work just perfect.

[atagunov]

- Follow blog instructions on parsing the EEPROM dump and modifying necessary parameters. I zeroed out the cycle count and set the capacity to 7000mAh.

Sounds like fun On a whim I ordered a CY7C68013A.

Unfortunately https://github.com/karosium/smbusb is preciously thin on info. It's probably going be some effort to figure out how to connect CY7C68013A to the battery connector.. Guess I need to put that simulated load on the battery (a resistor + fan), etc. Would have been nice to save some time and find instructions on doing this..

UPDATE: things seem to be getting clearer.. this CY7C68013A which seems to be more officially known as Cypress FX2LP has got some pins marked out as GND, SCL, SDA. Apparently these are the ones which need to be connected. An online source claims the battery connector is

+, +, (no-pin), SCL, SDA, Thermistor, GND, GND

Guess just connecting those 3 pins would be a start..

UPDATE2: it seems interesting to see what powers the demo-version of this app happens to have: http://be2works.com/ It seems quite a bit of intelligence has been built-in. Apparently there are roughly 6 actions to do and roughly in this sequence: read general battery data (ok-notOk, charge remaining, etc), get EEPROM data dump, put the battery controller into "unsealed" mode, modify the EEPROM data dump, write EEPROM data back, possibly give some extra commands to make it run nicely. Demo version will lack the clever tricks for "unsealing", but it should still allow to enter commands manually. Perhaps for some of the batteries "unsealing" can be done that way even on Demo version.. It seems some "passwords" - 2 byte sequences - for doing this are available somewhere online for some batteries.. If the "unsealing" is successful, this be2works promises some nice things at least for some of the controllers - reset production date, cycle count and probably even meaningfully play with capacity. This is what would have been real nice.. Doesn't look like fun trying to figure it out on your own.. Unfortunately full-version price is still a bit steep for our use: $350 and I would generally expect this app to be in a decline now that Chinese manufacturers are full rage.. Like KingSener Still I have splashed out another £4 for CP2112 from AliExpress - that seems to be the controller required by this app.. BTW there people online advising to use some software for CP2112 to lock it's USB device IDs, since apparently they somehow sometimes get messed up with using be2works.com app.

So many of these controllers are TI produced and run either stock firmware or a firmware derived from stock offered by TI. In either case one of the parameters to set in EEPROM is cell "chemistry" code. Apparently TI have measured a number of cells, took note of some important characteristics (minimum voltage to allow?..) and made it part of the firmware. Modifying chemistry code seems like a rather obscure part of messing with EEPROM data.. Seems this be2works allows just to set it inside that EEPROM data dump. But set it to what?.. How to find the most suitable code for your cells?.. Also I'm not yet clear if/how to tune the controller to a change from 4.35 to 4.2V cells..

UPDATE3: with a sacrificial T61 (nVidia NVS140m 128Mb 08/08 yeay! it arrived! 15.4" only good as a source of spares for me) I received an original 9-cell 92P1133 10.8V 7.8Ah (so about 84Wh nominally) battery. The case of the battery was slightly damaged (somebody put it next to a cooking hob accidentally?..) so I couldn't resist the temptation to open it up. Honestly I was hoping to see something like bq8030 or some bq with 'z' in the name inside.. but no, it was an M37512. The cells are CGR18650E. 2.55Ah as expected with max voltage of 4.2V. 4.2V isn't bad - it matches currently available cells it seems. Opening it up with a pair of flat screwdrivers was a bit brutal on the case of course.. It was a bit damaged to start with and is more of a mess now. No idea if I will ever close it again with new cells.. I'm not a big fan of 9-cells, one of the reasons I didn't hesitate to opened this one up. I'm more keen on re-celling a 6-cell with 3.4-3.5Ah cells so that to reach around 74Wh with less weight. Though in theory this case could hold more than 100Wh if new cells were fitted and the controller was aware of it.. To my shame I haven't even properly tested this batt before opening it up, so keen I was to see which controller it is

My other problem is I don't have any kind of a lab power supply to charge the batts outside of a Thinkpad. Any ideas what I could use cheaply?