Teardown: Debugging a faulty MacBook Pro battery
If your wife’s MBP battery is destined for nowhere other than the trash can (or, recycling), I could take a look at it for you if you could get it to me.
Lee even picked up the one-way shipping tab! I was happy to take him up on his generous offer, the results of which (complete with photos) I'll detail in this post. Before documenting what Lee found from his detailed (and creative) analysis, however, I'll begin by telling you a bit about both Lee and his company, in his own words (lightly edited by yours truly for grammar, clarity, etc.).
Lee Kresge is a Research Chemist working for the Indium Corporation. When he is not in the lab formulating polymer products for the electronics industry, he wears the hat of "data acquisition guru" for the R&D department. In his off time, he enjoys building and flying R/C aircraft (and recently, boats), cycling and (despite his age) participating in watersports activities.
Indium Corporation is a premier materials supplier to the global electronics, semiconductor, thin film, thermal management, and solar markets. Products include solders, and fluxes; brazes; thermal interface materials; sputtering targets; indium, gallium, germanium, and tin metals and inorganic compounds; and NanoFoil®. Founded in 1934, Indium has global technical support and factories located in China, Malaysia, Singapore, South Korea, the United Kingdom, and the USA.
With introductions out of the way, let's get to the meat of the writeup. Lee kept me updated via a series of email exchanges as his analysis progressed, concluding with a 23-slide PowerPoint deck that combined photos and often-hilarious (at least I think so!) commentary and also showcased the contributions of his daughter. I've reproduced it all below in Lee's own words, again only lightly edited by yours truly. Take it away, Lee!
Due to cheap plastic welding, the battery pack opened fairly easily. Once started with a butter knife, most of the plastic spot welds were broken with upwards of 20 fingernails. (Don’t know what she does to them, but my daughter’s were razor sharp.)
The photo above shows the simple series wiring of the battery pack, referred to as a “3S” pack in the R/C world. (The red and black electrical probe is only there to hold the blue tape up for the picture.) The yellow and white wires are connected between the cells for individual cell balancing and monitoring.
Charge controller (and over discharge protection?)
Before I tore into the battery pack, I wanted to get a few pictures of the electronics and wiring. Along the side/end of the battery you can see a temperature sensor (later determined to be a NTC thermistor measuring 7kΩ at ~20°C and 9kΩ against an ice cube). A minimal amount of high temperature PI (polyimide) tape is used for a good part of the assembly.
One thing that I noticed right away was that the temperature sensor was not in physical contact with the cell it was taped to. In the high power R/C world, this sensor would be plastered down or glued to the center of a battery pack. Air (especially dead air space) is a great insulator, after all. However, I would not expect this quirk, a result of poor assembly technique, to contribute to the battery pack shutting down.
This, on the other hand ...
On the upper top side of the charge controller, I noticed a broken chip capacitor. It did not move easily when I tried to pull it back into place with my fingernail, leading me to believe it had been that way for some time. It is quite small, estimated to be a "0402" chip capacitor by my coworker, who is experienced in the handling of these common components.* The black wires in the picture come from the temperature sensor.
*0402 = 4 mils long by 2 mils wide; narrower than the thermistor wire residing next to it.
Could it be THAT simple of a failure to find?!? I had tools to use, software to run, data files to generate. I wanted to play! And here I’m shut down at the sight of a broken passive component. Sigh.
Depressed yet undeterred, I continued my testing after the holidays (I also had more free time.) I discharged the battery pack through the charge controller while directly monitoring the voltage of each cell, bypassing the charge controller. I chose a discharge rate of ~1.0A, as I thought this would be representative of what the battery would see under moderate load in a computer, and get the testing done in a reasonable amount of time. That the 1.0A matches the current that two boat trailer light bulbs draw while wired in parallel is pure coincidence.
The three cells in the pack were very closely balanced and nearly fully charged, at 4.114V, 4.112V, and 4.116V as received from Brian. They were discharged until the lowest cell just broke through 3.4V. This is a bit low, I know (should Amclaussen be reading), but it's not a voltage that should have damaged any of the cells.
The voltage dip ~3/4 of the way through the graph for the first cell is due to my daughter disturbing one of the wires (which I didn't notice at the time). I do not believe that this dip is representative of what the cell would actually experience in normal usage. The downward spike in the current graph ~90% of the way through the discharge is due to me trying to get revenge on her by shining the light in her eyes. So much for alligator clip connections …
I was a bit surprised that the charge controller did not shut the battery down, either immediately or before hitting 3.4-3.5V in the lowest cell (as a future extension of this project, I will try discharging the pack again to a lower voltage to see if the charge controller shuts down the discharge).
After discharging the battery, I next wanted to test the controller's ability to charge and balance it. As I needed this to be a "walk-away" test (my daughter's college application forms were vying for my attention), I opted to use a power source that was less than the 4.2V × 3 cells' total voltage, should the controller fail to operate as designed (although my fire insurance was paid up, I was not interested in exercising it). It turns out that if you play with a fully charged deep cycle boat battery long enough, its standing voltage will drop to 12.5V, conveniently just below 3 × 4.2V.
Charge testing through charge controller
The charge controller balanced the cells to within +/−2mV at the end of the recharge cycle. This is a pretty good result, considering that they differed by >100mV at the end of the discharge (i.e. beginning of the recharge). The post-recharge cell voltages summed to 12.5085V for the complete three-cell battery pack.
Output connector testing
Desperate to find an electrical/electronic explanation for the laptop shutdown (i.e. something other than the aforementioned capacitor quirk), I then turned my attention to the output connector. It is a very small (at least in comparison to my fingers) 9-pin flat connector. Three sockets on one side go to the negative battery terminal through a 0.020Ω shunt resistor (R010). The three sockets on the opposite side go to the positive battery terminal through the charge controller. And the three center connections were unknown in function and output. I decided to measure both voltage and resistance between them, as well as between them and ground while under load, to see if any signals might be generated that would tell the laptop to shut itself down.
Unknown connector terminal measurements
*Distracted by a famished teenager while preparing dinner during this particular test, I mistakenly took two measurements for the connector pins to ground, instead of pin-to-pin measurements.
Various voltage and resistance measurements, both between the unknown terminal pins and between each of them and ground, gave no indication of any generated resistive or voltage signal that would tell the computer to shut itself down.