[NeXT] got himself an IBM ThinkPad TransNote and yeah, we’re pretty jealous. For the uninitiated, the TransNote was IBM’s foray into intelligent note transcription from roughly fifteen years ago. The ThinkPad doesn’t even have to be on to capture your notes because the proprietary pen has 2MB of flash memory. It won an award and everything. Not the pen, the TransNote.
Unfortunately, the battery life is poor in [NeXT]’s machine. The TransNote was (perhaps) ahead of its time. Since it didn’t last on the market very long, there isn’t a Chinese market for replacement batteries. [NeXT] decided to rebuild the replacement battery pack himself after sending it off with no luck.
The TransNote’s battery pack uses some weird, flat Samsung 103450 cells that are both expensive and rare. [NeXT] eventually found some camera batteries that have a single cell and a charge controller. He had to rearrange the wiring because the tabs were on the same side, but ultimately, they did work. He got the cells together in the right configuration, took steps to prevent shorts, and added the TransNote’s charge controller back into the circuit.
Nothing blew up, and the ThinkPad went through POST just fine. He plugged it in to charge and waited a total of 90 minutes. The charging rate was pretty lousy, though. At 94% charge, the estimated life showed 28 minutes, which is worse than before. What are your thoughts on the outcome and if it were you, what would be the next move?
Fail of the Week is a Hackaday column which runs every Wednesday. Help keep the fun rolling by writing about your past failures and sending us a link to the story — or sending in links to fail write ups you find in your Internet travels.
35 thoughts on “Fail Of The Week: Battery Packin’”
First thing that pops into mind is resetting internal eeprom calibration..
Those crazy, rare cells are likely either a different chemistry or charged to a different voltage than the camera batteries can handle; maybe it would have been better to check the specs on the battery and try to rebuild that with round cells instead of flat, since camera batteries are meant for, well, cameras. Further, he does say that the cells from the batteries, which are nominally the same, are brandless cells not rated for computer use.
I think I have seen some battery related IC’s that monitor the total amount of current that has gone through each direction. It would still be up to the controller to read those values and make something of them. If the capacity is increased by adding more/larger cells, it may require a few discharge/charge cycles before the controller knows what to expect, unless the values are hard coded in.
The estimated lifetime is just that. Estimated. Probably uncalibrated to the new cells. Has the ACTUAL lifetime been measured?
tl;dr. If he tested it and it did last for 28 mins, something must be wrong. But if it lasted more than what was estimated, the battery must have not been calibrated yet, or whatever.
Thanks for your compelling and insightful contribution.
A lot of older cell phone replacement batteries also have these cells in them. They can usually be had for cheap, too. Unfortunately, I don’t remember the model number of the battery of phones that I have found these in. At least his management circuit let the new cells work. I have an old CF-41 whose battery management has decided the battery is bad, and has turned it ‘off’, so no matter what I hook to the management board, no power comes through. I am still waiting for somebody to come up with a good way to reset them without having to spend over $150 on some unheard of software from a questionable website.
While they bring some neat benefits, ‘Smart’ batteries can really be a problem for a lot of us who collect and work with older portable computers.
Correction: CF-M34: the tiny one with a touchscreen. Hackaday, you really need to add some kind of post editing mechanism!
Try jumping the positive pack voltage to the computer-side positive voltage input. This should ‘Jump-start’ the battery management circuit and get power flowing. That’s what I had to do when I re-celled my CF-M34 pack.
Wire up a dc jack and use a wall wart of the right specs.
The “computer” inside the battery pack remembers that the cells are dead, so replacing them will yield no usefull result. Try to contact some russian or chinese people, that does in replacing cells in packs. I many parts of the world it’s common practice to do this.
The old pack still functioned, though not well. I think the battery manager will kill the pack if it detects a fault, but if the pack is still operating, I think it can relearn the capacity.
those camera batteries he got are around 300mAh :)
original cells are probably at least 1500mAh
there is a picture of the camera batteries, it says 3,7V 2200mAh.
all 4 in series is 14.8V and 2200mAh, which in theory should be more than the 1.6Ah of the old battery.
It’s probably just the classic case of chinese battery manufacturers being a bit “Over-optimistic” of their cell capacities. :)
chinese 2200 = 300-400mAh
Some of the really cheap batteries out there are rejects that don’t meet specifications.
Those battery packs have charge management chip that has EEPROM memory, internal or external. It stores data about packs, if old packs reached for example 500mAh those values are stored in chip and charge management treats them as 500mAh. Even when you replace cells with 2200mAh. To completly rebuild pack you need to know which values in EEPROM have to be replaced with those of the new pack. Mobile phone batteries are simpler, they have only protection circuit, no memory so you can replace cells with new ones and it will work OK.
there is an additional issue with 3.6 vs 3.7 volts Liion batteries – they are never going to charge to 100%, because the charging circuit will stop early. you should expect 80 %. charge never the less. try to charge / decharge cycle a few times, that helped me in the past with a similar upgrade.
The nominal rating has nothing to do with it, the 3.6 vs 3.7 has to do with which edition of the iec standard the are certifying them to, the charge voltage could make a difference, but the nominal voltage rating has no real correlation to this
103450 is the most common prismatic cell, i dont see how they are rare at all. The battery life estimation is based on the fuel gauge and is likely tied to the cell that was in their before. I would have to see the pack personally but it should be pretty trivial to knock it off, this was in the days before sha-1 handshaking was standard in batteries.
To add a couple more of the traps :
Whoever posted above about the capacity mismatch is probably correct. I doubt the replacement cells come close to the original cells.
I would not look at samsung/sdi for replacement cells, until a few years ago they were impossible to source in the usa thru proper channels and are still pretty impossible to get on the secondary market, look at the lg equivilant to teplace them.
Check the original data sheet for charge voltage, some of the cell in laptops/computer equipment works off a higher (4.35) nominal charge that the standard 4.2.
And finally with prismatic lion cells typically the can is positive and the button negative, this is reversed from typical cylindrical cells, ive seen many a shorted pack because this delta was overlooked.
And in fact he is correct. After I submitted this article I was determined to find out if this was just the controller or crud batteries. It’s a mix of both. The discharge curve in these cheap chinese cells are several volts lower than the original Sanyo cells which were indeed probably rated for heavy loads. When you completely disable the TransNote’s power management (I swear I turned hibernation off but apparently I was wrong) you can watch it give a low battery alarm when the cell is actually at 70% capacity and when the controller “thinks” it’s at 7% (about 60% in reality) it would hibernate however with that switched off it will drop to 6% and the voltage drop will stabilize for 35-45 minutes, then it will proceed to dip down until completely discharged three to four minutes later.
In all I stand corrected. I’m actually seeing almost an hour and a half out of the rebuilt pack, mind you the pack originally had a capacity for a more useable two and a half hours but the controller is too confused to give me that hour and a half.
Unfortunately the traditional cylindrical cells can’t even be hacked into this pack as unlike all other laptop batteries which are on the outside the battery on the transnote is on the inside of the book-style enclosure so you couldn’t close it.
There are a lot of mysteries regarding this charge controller however it appears to not be as smart as we all assume. It’s actually changed it’s life expectancy average with repeated cycles and if you reinstall the four old cells the charge rate bumps back up to 1.5Ah but either because it has bot been “retrained” enough times yet or because the previously mentioned cell discharge curve is hard coded it will continue to assume that the battery is dead long before it actually is.
Things were a little bit easier in the Ni-Cd/Ni-MH days. Lithium is a strange creature, at least to me. :)
Any chance of pulling a part number off either the charge controller, safety IC or fuel gauge? Safety is likely seiko ic of some type, and the fuel gauge is probably either maxim, TI, or benchmark (bought by TI). Lithium isn’t bad at all, you have to handle it with kid gloves more than nimh but the advantages are huge.
Also take a look at the dataline, based on a quick look of the circuit board and the inclusion of the uchihashi fuse (a sure sign of a substandard design, it likely failed UN 38.3 the first time around and the corrective action was to add a fuse to pop when the fets go into runaway on overcharge) Its probably communicating via HDQ, which can be a pita to parse unless you have the right tools, but iirc someone out there did write a arduino library that does it moderator well.
Well, for a start the guy should have wrapped the cells to reduce the chance of a nasty short, but, oh well.
As others have said, that pack, has some smarts, or at least an EEPROM with an ID, configuration, callibration, and historical information.
First thing to do is to discharge to 3.0v/cell, outside the device then do a full charge cycle in the device, then do a rundown in the device. A couple of those might get it to recalibrate itself. If that doesn’t work there is the possibility of rewriting pack id info.
How depens on whether it is just a dumb board with an eeprom, or a full smart battery system with an MCU that can be talked to over SMBus (i2c). Even then, IBM probably locked out acess to the relevant config parameters.
Also, he can probably get better cells if he looks for ultrabook or tablet packs. I just pulled some Very similar looking prismatics a pack from a 10y old toshiba tablet. I also have some 6 year old Dell 9T119 packs that have these. I know there are packs made in the ladt 4 years with similar cells, but I don’t know which ones.
Was the actual battery life tested by running the laptop until it ran out of power or did the person only go by the estimated battery life displayed on the screen?
Until recently I ast times used an old Dell Latitude. The KDE desktop I was running kept complaining that the battery was at 17% of its original capacity and may be broken while the estimated battery life was just minutes. However, after the battery gauge informed me that the laptop battery was almost dead the laptop kept running for another 3 to 4 hours.
If the laptop was not allowed to run out of power while testing then I would call this fail article a potential fail. (sorry, couldn’t resist)
Looks like the original cells
I would see about getting these LG ones as a replacement:
Also, try to find them with the tabs on, unless you have access to a resistance welder…. Trying to solder to the battery can is a no no.
1. The donor battery NP-120 is for a camera market in 2005 – 2007. Its not unlikely that the chines battery has some years on the shelf and done a deep discharge.
If you have a RC-charger try discharge the battery to 3V at a low speed 0,3C and then charge extremely slow 0,05C for a day or two. Repeat 2 – 3 times. This sometimes give 80% – 85% of original capacity. Takes a week to finish … but hey! If it works?
2. You do not need to weld tabs. Solders works fine, you need a special “fast solder technique”.
3. As long as the number of cells / chemistry / configuration match you can have a external battery pack + cable, Not pretty but you can make it 10 times bigger and use “normal cells”: 18650 or a large number of cellphones battery, some in parallel.
4. If its ok to have a external battery pack and a external fast charger, you can use one (or several) DC – DC step up / down to mimic the correct voltage, from a car battery or something. Battery is no longer a problem. The battery meter is not going to work so… the computer is going to die when the voltage drop fast.
1. not an issue with lithium batteries
2. Also wont work with lithium, there is no memory effect with lion.
3. WRONG and bad. trying to solder to a lithium ion cell can is never recommended. You can go ahead and do it, you can also play with contact explosives or do alot of other dumb stuff. Best case you cause localized heating and possible premature failure of the separator, worst case you pop the cell and vent organic acid all over yourself… never mind you are now handling something on fire.
Has anyone checked to see if the 15 year old cells are even li-po or 3.7v? Could be the old type that were 3.0v or another chemistry. Something else those camera batteries could just be of horrible quality and not have the rated capacity claimed.
see above where i posted the data sheet.
Found some batteries for the guy http://www.directron.com/ibmthin2675b1.html
My ThinkPad (Twist) runs for 40 minutes with 0% battery.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)