tap water is a strong electrolyte
Many thanks to all good people here who have posted schematics, board layouts, images and explanations of the power circuit. I wouldn't be able to fix my friend's mac without your help.
This it to report how I did it, in case somebody runs into a similar problem.
In fact, I had two fix two separate problems, both caused by the same fault; more on that later.
Cause: water spill; running wet for two days
Symptoms seen at various stages:
- Machine turning itself on and off spontaneously
- Battery not changing
- Battery not recognised
- Can only start from battery
- With battery depleted, can only start with battery connector unplugged
- Original MagSafe plug blown
- Replacement MagSafe not turning its light on (light is visible, but dim)
- Charger not recognised
- Signs of corrosion on the logic board
- No damage to keyboard or power button
Culprit: corroded via to pin 5 of U6901 (
=PP3V42_G3H_ONEWIRE)
Ancillary problem: battery inoperative due to deep discharge
Hardware: MacBook Pro A1278 13" 2.26GHz K24 BETTER (the best ID I have)
Schematics:
A1278-820-2879.pdf (not an exact match, but close)
Board layout: "
A1278 820-2530 K24.brd" (
posted here; the board file inside this archive needs to be renamed to drop the
.zip suffix)
Back story: My friend toppled a glass of tap water near her mac, with some water splashing on the screen. She wiped the screen and forgot about the incident. Said no water had been seen on the keyboard. Later on the same day, her mac turned itself off while she was working on it. After a while, it did so again. Having now linked this new behaviour to the water spill, she decided to leave the machine turned off until the following Monday, with the view to ask me to look into it. An hour later, the machine turned itself on. Now seriously concerned, my friend decided that saving her data from the errant mac was of greater urgency. She spent the following two days fighting it.
Investigation:
- When I first saw the damaged machine, it was turned on and still appeared to be functional, except the battery was fully depleted and "Not charging" was showing in the battery indicator.
- I took it apart and examined the damage. There was still liquid water on the bottom cover under the power circuit, and some parts were soaked in water. Part of the board was covered with white residue, showing copper-blue around some large components such as electrolyte capacitors. Apparently, the water got in through the air intake on the right side of the body.
- The keyboard appeared dry, but in view of the report of erratic power-on, I thought it wise to examine the power button. That dragged me into an unwarranted quagmire, but at least I can say I know how the power button is made. (An unfortunate casualty of that endeavour is keyboard lighting. I failed to align the light guide and now only the middle two columns are illuminated. Seeing how fragile the keyboard was, we chose to continue without keyboard lighting)
- Cleaning and drying the board presumably curtailed the damage but lead to no improvement.
- The first attempt to run it from external power led to the destruction of the MagSafe adapter. Its light shined unusually bright, then became intermittent and died off. That adapter could no longer be used with this or any other mac and had to be retired.
- With the new adapter plugged in, the machine would not start.
- Without external power, the machine could only start while there was still some energy in the battery. Because it was depleted beyond the safe limit, it would only run for a few seconds before shutting down.
- Eventually, the battery went so dead it could not start the machine.
- After much fiddling, I found that the machine was basically alive and could run from external power, but only with the battery disconnected. I did not know about the SMC bypass and did not know you guys existed (and google failed me), so I settled on this start-up procedure:
- Connect external power
- Plug the battery connector in (not 100% certain it was necessary but seemed to help)
- Unplug the battery
- Push power button and release after 3-5 seconds. Much shorter or much longer intervals did not seem to work
This procedure worked most of the time, with a higher failure rate in the beginning, probably due to residual moisture in the board still interfering. Recently, it worked without failure. I judge it to be a better way of running a damaged mac because the SMC remains online and can be tested. Had I known about the SMC bypass, I could have missed some important information.
- Connecting the battery while the machine was running resulted in an immediate power-off.
Repair
Having considered the above symptoms, I came to a tentative conclusion that there was irreversible damage in the logic board and decided to wait for a good deal on a replacement. Now,
a whole year later, no such deal came about, and I was almost ready to send the board to a fellow in Taiwan (
cf. "ebay logic board repair") who promised to fix it for $90 provided it was fixable. But I did not feel good about it, because it was going to cost a lot of money in any event (shipping and all), and I wasn't even sure what the problem was. So I gave it another look just to be able to describe it properly. First, I made quite a bit of progress by fixing the presumed-dead battery, and then I discovered this wonderful forum (which really needs to be digested to some usable form, by the way).
- Battery
Since there was apparently no way to charge the battery in situ, and buying a new battery just for testing seemed like an almost certain loss of money, I ripped the dead battery's housing open, cut the red and black wires inside and used them to connect it to a "dumb" 9.6V Ni-Cd power tool charger. The "dumb" charger does not monitor the battery, so you can use it any way you want. The initial current was 0.5A, and I left it running overnight. By mid-day next day, the current was about 0.2A, with the voltage drop across the battery close to 13V.
Left to sit for several hours after charging, the battery showed no appreciable loss of voltage, and I thought it was up for a test. I put the cells back in what remained of their housing and taped it all together. (NB.: If I were to do it again, I would warm the battery with a heat gun or a hair dryer, or put it in an oven for a few minutes, to soften the adhesive that is used to attach the cells to the housing. It was relatively easy to crack the housing open -- which is best done when it is cold -- but taking it apart without first warming the adhesive can damage the housing and the cells. Also note that there is no extra space to accommodate any enlargement of the battery, so the final taping must be minimal. Use just a single layer of thin scotch tape over the parting line and one layer to cover the open part of the housing)
Incredibly, the so treated battery worked fine. The machine was able to start with or without external power. The battery indicator reported "Not charging" and "Service battery", but while the former was a problem yet to be addressed, the latter was likely due to old age. The profiler informed:
Health Information:
Cycle Count: 819
Condition: Check Battery
Battery Installed: Yes
Amperage (mA): 1.6
Voltage (mV): 12206
At any rate, at this point, the battery was talking to the SMC (which fact I could ascertain with a scope), and it even turned out that the "Not charging" message was a lie. After my rough-and-ready external charging, the charge level as seen with CoconutBattery was 89%; it went up to 100% a few hours later.
I felt I could even leave it at that. The machine was fully functional, the battery was charging; it could sleep and wake up and was again a real laptop (as opposed to a crippled desktop it was year ago). But the SMC not talking to the charger and the MagSafe light not being operational represented a usability problem (however slight), so I pressed ahead with the charger and that was when I discovered this forum. I was amused to see somebody suggest jerry-rigging a charger for the battery in one of the first several posts; that was the first idea that came to my mind, but I'm not sure it would have worked with the smart interface this battery has, which shuts itself off when the charge gets depleted. BTW, I found the charge to be around 7V, evenly distributed among the cells, even after a year in storage, with only a few mV showing on the outside.
- Charger circuit
I soon found that the charger was working and therefore I did not need much of the priceless knowledge about the FET gates shared in this forum. But finding the remaining few bits about the charger sense signal took some digging.
This was the condition I saw when I restored the battery and plugged it in:
AC Charger Information:
Connected: Yes
ID: 0x0000
Wattage (W): 0
Revision: 0x0000
Family: 0x0000
Serial Number: 0x00000000
Charging: No
While no data was coming in, it was clearly there and working, so "Not charging" was blunt misinformation.
The battery charger U7000 was happy, with its =CHGR_ACOK on pin 14 showing high.
On first testing, the charger sense line appeared to be severed. It only seemed to be clamped by a diode to -0.2V, or maybe a Zener, but I did not reach the Zener breakdown, the way I tested it, so don't know which one it was. There appears to be a Zener diode on the DCIN board. I tested the line with the ambient field by keeping my finger on the tip of my scope's probe. It went up all the way, but hit the floor at -0.2V. Past the diode, it felt like it wasn't connected to anything. The DC resistance in each direction was several megohms, which could be just dirt. No DC at all was to be seen where wise people tell me I should expect 3.4V.
It took me a while to determine that my "1-Wire OverVoltage Protection" circuit was entirely packaged in a single IC (the relevant schematics cited above can be found in a couple places in this thread, but they differ from the most-discussed version and are not easy to find).
While poking around this IC (U6900, MAX9940) and its gate (U6901), my probe fell through when I touched a test pad for =PP3V42_G3H_ONEWIRE (shown in the first picture). Looking at it through a microscope, I found that it had no metal in it; it was rust all the way in.
It turned out to be a via, with its opposite side still connected to C6908, where I did find the promised 3.42V. This line is apparently routed on the inner layer of the board and this via is the only point in this vicinity where it comes to the surface.
To repair it, I first cleaned it out with a piece of 34-gauge steel wire. Copper did not work. All needles I had were too thick and promised to ruin whatever metal was still left on the good side of the via. Doing this under the microscope, even steel felt like a flaccid sausage. It positively did not want to stay in that hole, let alone go deeper in it. The copper wire I tried first felt like a squirt of toothpaste. It is amazing how the microscopic perception of these materials differs from our everyday experience with them. Later, trimming that steel wire with a utility knife felt like cutting through butter.
Eventually, I was able to thread the steel wire through the via and clean all the detritus out of it. I soldered the same wire to the remaining part of the trace leading to U6901, and to the pad on the opposite side. The result is shown in the second picture (poorly -- my microscope is not instrumented, so I had to use a handheld camera).
That was all that needed to be fixed. The circuit that seemed to be severed (sing an ode to the FETs) came back to life and all is well now.
I have also attached a couple pictures of the relevant part of the board. These may be helpful in further discussions of this particular layout.
