During the winter break of 2014/15, my Dell XPS 15z laptop suddenly died during use. The power button would not turn the computer on and the battery indicator on the side would not light up.

Checking the power supply, I noticed that the light on the brick was out, even when after unplugging it from the computer. Unplugging the supply from the wall and plugging it back in revived the light, but it would go out instantly when plugged into the computer.  Dell power supplies will shut off when their output is shorted and only reset if they’re unplugged from mains power for a few seconds. It became apparent that I had my laptop’s motherboard had shorted itself. I confirmed a dead short across the DC input jack with my multimeter. Great.

Apparently this was a common problem with this laptop, with many similar complaints on support forums. Being 1 year out of warranty, Dell was not going to repair it despite being a documented issue. They offered to sell me a new motherboard for $250. No thanks.

At the time I wasn’t able to focus much on getting it back running due to work and school, so I bought a new laptop. Having a bit more free time, I pulled it out of the closet last year and found that the fix wasn’t terribly complicated, as documented below.

Since the battery indicator on the side of the laptop wouldn’t light up anymore I figured the short was on the system’s main positive power rail and not isolated to the DC input. That would mean the battery was being shorted as well. My initial guess was that one or both FETs of a synchronous buck converter latched up and was providing a direct path between the system’s power rail and ground (shoot-through.) Time to pull out the motherboard!

The first thing I did was measure continuity across the battery connector, which also showed a dead short. I looked for visibly damaged components on both sides of the board near the battery and DC input, as that is where the majority of power electronics reside on this board. In the image above, that would be the lower-left section of the board. I did not find any damaged components, nor did I smell any residual magic smoke. At the time I was not able to pin down the schematic (which can now be found by searching the web for “Dell XPS 15Z Quanta SS8”) so I  tried a trick I’d seen Louis Rossmann use in some of his videos: Pass a constant current through the shorted rail and see what heats up or smokes.

In my initial attempt I tried passing current to the board through the DC jack. I connected a snipped Dell DC cable to a CC/CV buck converter based on the XL4005 buck converter, commonly found on Ebay and Aliexpress for a few bucks. (Constant-current is not a feature of the XL4005 but is implemented with an op-amp and current shunt on the board.)

I powered the buck converter with a 11.1V LiPoly battery and set the current to the minimum. The voltage setting doesn’t matter much here since the shorted output is going to force it to operate in constant-current mode. Here’s the setup:

Now we turn up the current and look for heat! In Louis’ videos he often dabs isopropyl alcohol on a shorted board and looks for evaporation as an indication of heating components. I had a Flir Tau 2 handy so I decided to see if a thermal camera could spot the problem:

As I increased the current it appeared the DC input connector on the board began to heat. The image above was taken a few minutes into the process and the heat had soaked into the board. I took a look at the other side of the board, as there are some FETs there that may be the culprit. They didn’t seem to stand out:

Here are both sides of the board with the thermal image superimposed on them:

I shifted the colors in the image above to make the thermal image a bit clearer. 

The DC in jack seemed to have a higher resistance than whatever was causing the short. I decided to take a more direct approach and power the board through the battery jack and see where that got me. I soldered two wires to the battery jack on the motherboard as shown below:  I connected my power supply to the wires and repeated my test with the thermal camera:

This looks a lot more promising: The long QFN and inductor, located between the battery and DC input jacks, immediately began to heat. After a minute the board began soaking the heat but its still quite clear in the image below what is sourcing the heat:

Superimposing the thermal image onto the board gives a clearer indication of which components are heating:

The QFN is a Texas Instruments BQ24765 battery charge controller with integrated MOSFETs. The functional block diagram on page 15 shows two MOSFETs across the IC’s DC input and ground, used to drive the synchronous buck output of the charge controller. One or both of the FETs must have failed in the on state and were causing a direct short across the board’s input positive rail and ground. The large inductor just below the IC is heating because it has a series connection between the battery jack and the PHASE output of the charge controller, as shown in the schematic diagram on the first page of the datasheet.

I don’t have access to a hot air rework station, so took the board to the office the next day and used a hot air gun to remove the charge controller:

And melt a couple of connectors in the process…

Nonetheless it came off quite easily. A continuity check across the DC input with a multimeter indicates that the short has been broken.

That evening I reassembled the laptop without the battery (as the inability to charge makes the battery useless) and plugged the DC adapter into the laptop. It worked!

Dated and permanently tethered, I currently use this computer as a general desktop computer and for stitching large photo sets that would tie up my current laptop.