Help identifying the faulty component on a laptop motherboard

[philipjfry]

Hi everyone,

I know that this is likely a futile endeavor, but I have a defective (yet optically pristine) laptop mainboard here which I would very much like to repair if possible. The CPU and GPU are still pretty powerful (i7 4720HQ, GTX 960M) and a replacement would be uneconomical, so it's at least worth a shot.

Due to the nature of the failure, there may be a glimmer of hope: It conked out under full load while running a game, drawing nearly peak power (around 100W). Now it is completely dead, not even the battery indicator light will turn on when I connect the AC adapter. I am therefore hoping that the issue is related to power delivery. The first thing I tried was unplugging the battery. Then I tried swapping the AC adapter and using this device's adapter with another device. The AC adapter works fine, but I noticed 2 things when plugging in the other (different model) AC adapter: It started chirping in ~1s intervals. Also, The original adapter (which is not chirping) will provide 19V when freshly plugged into the wall and its voltage permanently drops to a random low value (such as 3V or near zero) when I plug it into the defective board. It does not have a "smart" pin btw, so I am a bit curious why there are 6 wires coming from the jack.

I am not an electrical engineer, but experience leads me to suspect something shorted out? I therefore measured for continuity at the power jack, but my multimeter says it's an open circuit. However, since 2 wires become 6 at the board, there might just be some "smart" wizardry going on and the short circuit may only occur once 19V are delivered? The 3 cell battery also seems to be very "smart" with 10 wires coming from it and only like 0.01V coming from the 3x2 cell wires while it is unplugged.

The machine saw some heavy use over the last ~3 years, but the CPU was slightly undervolted, the GPU was mostly underclocked, high performance thermal grease was applied, the fans were regularly cleaned and operational and the laptop was even propped up a bit in order to reduce heat/strain, so I don't believe the CPU or GPU died.

I finally poked a few components such as the VRMs with my multimeter leads and didn't get any readings I would consider suspicious, but I don't really know enough about electronics to analyze the readings. I DO know that you can't really measure much while a component is still mounted though...

Does anyone have a suggestion what I could try next? It would be much appreciated. Besides the multimeter, I have a soldering iron with fine tips and enough experience to remove the bigger components. I do not have any tools for SMD stuff, but I know someone who could help me out with that if there's something worth trying. The images below are as high-res as the board rules will allow, if you need better quality let me know and I will upload them somewhere else.

Thanks in advance!

[Twoflower]

Is the orange jack the power supply connector? Check if the cable has developed some shorts due to high power consumption. The insulator looks a bit wired on the picture.

Edit: In general it would be nice if you can post the model of the laptop. That might help to sort things out a bit.

[philipjfry]

Hi!

Yes, that's the power jack. Good idea, I checked that rightaway, thanks! I Unplugged it from the board, inspected it closely and wiggled it around a bit while measuring, hete's what I learned: The 2 wires from the AC adapter are just split into 3 wires each and while there is of course continuity between these 3, there is none to the other 3 wires even when wiggled. Measuring the pins on the board, they are also 3 pins connected each and at first I measured a 650 ohm resistance between the pairs of 3. After plugging in the AC adapter, resistance actually went up and there is now no connection between the 3 pairs? Don't know what to make of that.

The board is from an Acer Aspire VN7-591G. I left that out because it's a bit embarrassing. But you are of course right, it may be helpful if for example someone knows about a common defect with that model that I am unaware of.

[Twoflower]

Hmm. that resistance went up might indicate that a poly-fuse triggers. That means that something is drawing a high current. After a bit further inspection I noticed something odd looking: On the Top side, on the left half there are 3 similar devices PU8202, PU8203 and PU8205. Can you check if they are looking OK? Especially the 202 and 203 seem to have something on the right side of the device that looks odd. It could be an artefact from the cam. That area seems to have a slight changed colored PCB too. That is an indication of heat. So a closer visual inspection of that area might useful to identify the problem.

But other than the power jack and the visual inspection I haven't anything I can add here. Sorry. But maybe someone else has an idea. I'm crossing my fingers for you.

[RomDump]

I am not an electrical engineer, but experience leads me to suspect something shorted out?

I finally poked a few components such as the VRMs with my multimeter leads and didn't get any readings I would consider suspicious, but I don't really know enough about electronics to analyze the readings. I DO know that you can't really measure much while a component is still mounted though...

Does anyone have a suggestion what I could try next? It would be much appreciated.

If you really think it is the power delivery Get the schematic and boardview and trace the voltages to see what is missing. The files are on the net, (you may have to pay a few bucks to get them). Let me know if you are still stuck.

[philipjfry]

this is great, thanks so much! even if i can't revive the board, it's still a wonderful learning experience.

i had a close look at the area you mentioned. yes, there is some discoloration due to ~2.5 years of being subjected to quite a bit of heat. it's the uncooled side of the hottest area on the board and acer ran the cpu and gpu pretty hot (95°C, equivalent lenovo machines throttle to only 85), even though the laptop had dual fans and heatsinks where most others only had one. this was nice regarding the noise level, but i was always a bit worried about the temps, hence all my countermeasures.

regarding the artifacts, sadly that was just some residue from a protective sticker. i was able to wipe it off with some naphtha and took some new pictures afterwards. i also poked the 3 devices with my multimeter and (considering that 05 is upside down) they read pretty much the same so sadly, it doesn't look like one of them burned out.

[xavier60]

Check for a short to ground on the VBAT rail. This is the main rail that supplies adapter or battery voltage to various Point Of load buck regulators. The current sense resistor just below PU401 is on the VBAT rail.
If there appears to be a short, it could be due to a shorted capacitor or a shorted high side MOSFET in one of the buck regulators.

[philipjfry]

well if a cpu/gpu/vram goes bad i usually get at least the indicator lights, but this board is completely dead, so i was hoping it is something rather trivial that is not a final death sentence. besides, of the hundreds or maybe thousands of machines i have built and serviced over the years, i have only ever seen one cpu go bad (that was a 4th gen iX intel cpu too though, hmm...). in my experience, the silicon itself is very hard to kill. fun fact: i even had a duron 600 with a partly cracked die once that still worked fine and overclocked nicely. must have been a deactivated area.

regarding the schematic, could you please point me at a resource where i could find that or what numbers from the board i should look for (the stuff in the hades box i suppose?). would i be able to measure anything meaningful as a youtube-watching layman and if as twoflower suspects, there is a protective poly-fuse involved? i'd be game to try but it seems a bit daunting with my basic knowledge (U=R*I).

[philipjfry]

xavier, thanks so much!

if i understood you right and your instructions were poke here and there (see image) then the answer is yes, there is a short there!

sadly, it's really late here in austria and i have to go to bed now, but i'll be back as soon as i can!

[xavier60]

It next depends on what equipment you have and being able to get a schematic.

[RomDump]

regarding the schematic, could you please point me at a resource where i could find that or what numbers from the board i should look for (the stuff in the hades box i suppose?). would i be able to measure anything meaningful as a youtube-watching layman and if as twoflower suspects, there is a protective poly-fuse involved? i'd be game to try but it seems a bit daunting with my basic knowledge (U=R*I).

I don't like posting links for copyright reasons, however google "Acer Aspire VX5-591G LA-E361P Wish_KLS C5PM2" with quotes and Vinafix link has the files. You may be able to get them for free but you have to google a bit.

[philipjfry]

well at home i only have a (cheap) multimeter and (good) soldering station, but thinking about it, at work i have access to all sorts of stuff like proper multimeters, scopes, bench psus, signal generators...along with someone to make sure i don't accidently fry any gear, hehe. not sure about smd soldering, will look into that tomorrow. what would be essential/ideal?

[RomDump]

what would be essential/ideal?

For just diagnosing:
 
-A good Multimeter
-Fine Point Tip Probe, (Like Fluke 8845A-EFPT) there is cheaper options
-A magnifier glass, (or Microscope)
-The Schematic and Boardview
-Your Brain.

[philipjfry]

All right, thanks to RomDump's input (thanks again!), I was able to find the schematics and boardview for free! I suppose I should not upload them but rather share them via pm?

Regarding the equipment front, I haven't been so successful (yet), but I can jury rig a decent magnifier with my phone / have pretty good eyes and the probes I have allow me to reach most of the spots on the board. Stuff like the feet on the "nuvoton" labeled chip or similar components would be tough, but if there is a problem with any of these, I'd have to hand it over to a pro anyway.

So I opened the boardview with BoardViewer and familiarized myself a bit with the software. Seems pretty straightforward, although by myself I still wouldn't be able to do much more than confirm that I have the right file.  I also found a page with test points in the schematic file and successfully found and checked the "DC in" ones. Granted, they're a bit small for the naked eye and my probes, but it works well enough.

From this point forward, I'm afraid you'll really have to baby me. I do have a brain, but I usually use it to write code for these chips, not diagnose them.

[xavier60]

Using the schematic, confirm that there is a short to ground on the VBAT rail. Is it called VBAT on your model? And try to take an accurate as possible resistance measurement of the short.
Do you have use of a bench power supply with a Constant Current mode?

[philipjfry]

Sorry for the slow response, but we just had a baby girl a few weeks ago and naturally she is demanding a lot of attention. 

If I am interpreting things right, this rail that provides power across much of the board is called DCBATOUT on my model (see attached screenshot) and there is definitely a short to GND. Using my two cheap multimeters, I am getting around 0.3 Ohms. Interestingly, the value drops a bit during the first few seconds of measurement, but I am assuming that's because my multimeters are so crappy (they are also doing this when I am just shorting my probes and I already subtracted the value that remains in that case). Will also measure with a proper ohmmeter at work tomorrow (it's the first chance I get).

Yes, I do have access to a CC bench power supply at work! 

[xavier60]

Sorry for the slow response, but we just had a baby girl a few weeks ago and naturally she is demanding a lot of attention. 

If I am interpreting things right, this rail that provides power across much of the board is called DCBATOUT on my model (see attached screenshot) and there is definitely a short to GND. Using my two cheap multimeters, I am getting around 0.3 Ohms. Interestingly, the value drops a bit during the first few seconds of measurement, but I am assuming that's because my multimeters are so crappy (they are also doing this when I am just shorting my probes and I already subtracted the value that remains in that case). Will also measure with a proper ohmmeter at work tomorrow (it's the first chance I get).

Yes, I do have access to a CC bench power supply at work! 

That appears to be a real short on the DCBATOUT, usually a faulty MultiLayer Ceramic Capacitor.
A shorted buck regulator high side MOSFET can cause what appears to be a short because the DCBATOUT rail becomes connected directly to the supply rail for one of the large ICs, the CPU for example, which have a low resistance on their supply pins. This can be as low as 5 ohms.
 There are various methods for finding suspect shorted MLCCs. I apply 0.5V limited to 0.5A to the rail and track down the shorted part by measuring  voltage drops across the ground plane. This might be too complicated for a one off.
 I haven't used other methods, so I should leave it to others to explain first.

[philipjfry]

thanks again for your valuable insight, xavier!

here's an idea based on your methodology: if it is okay to apply 0.5V / 0.5A for a while, maybe i can identify the shorted cap using a FLIR camera? i have easy access to a quite nice one, so i guess i'll just give it a shot!  oh and i do occasionally get laptops with defects like that (in fact i just received another one that is also completely dead), so if this works, i might not remain a one-off! they are usually much cheaper and therefore less repair-worthy though.

[xavier60]

thanks again for your valuable insight, xavier!

here's an idea based on your methodology: if it is okay to apply 0.5V / 0.5A for a while, maybe i can identify the shorted cap using a FLIR camera? i have easy access to a quite nice one, so i guess i'll just give it a shot!  oh and i do occasionally get laptops with defects like that (in fact i just received another one that is also completely dead), so if this works, i might not remain a one-off! they are usually much cheaper and therefore less repair-worthy though.

When already available, using a FLIR camera is a good method. If the voltage stays too low, there will be little heat dissipated by the faulty part. It should be safe to increase the current to 1A.
Others have successfully used many amps at a few volts to smoke out the faulty part although this adds the risk of causing extra damage.

[LateLesley]

FLIR cams aren't always a good fit for finding shorted components. easier ways are to use alcohol on the board, or freeze spray, and watch what evaporates quickest.

Here's a couple of video's of repair channels I watch, showing each technique. One is on a macbook, the other will be an iphone board,, but the technique applies to any board really.

Hopefully it helps you.

[philipjfry]

i know i'm repeating myself, but thanks xavier and lesley! with your help, i was really able to pinpoint the culprit!  it was exactly what and where you suspected and lit up brightly on the flir camera even with just 0,5A. upon closer inspection, there actually was a visible indicator that i previously did not notice: a small part of the top layer flaked off, see image below. it took a bit of work to remove the cap cleanly (ironically, at work we only have a crappy soldering iron and at home i have a really nice one), but i even managed to save the little brackets holding it in place.  now that it's gone, we are back up to 470 on the DCBATOUT rail.

not knowing much about these buck regulators, i'm now wondering if it is a good idea to do follow my urge and do a first function test as it is (without replacing the cap) and if there is a more elegant way of finding out what kind of replacement i need without removing a similar good cap...

[LateLesley]

if you've took the cap off the board, then it's probably safe to power it up and test, as it will just be a power rail filtering cap of some sort. The laptop might crash if it tries to boot, but it should at least power up. But for the longer term you should replace the cap, as it will be a decoupling cap of some sort, and being doubled up like that, it probably needs both to reduce current stress on the cap. A few minutes run won't hurt the one left, but hours of use may be a different story.

Anyway, nice find! I hope you find the value, and get a replacement.

PS what are the "little brackets" holding it in place you mentioned? Can you take a pic? These components usually just sit on solder pads on the board, and there should only be solder holding the cap onto the board - the metal ends are part of the component. Its the only part of what you said that is worrying me.

[xavier60]

Those capacitors are typically 10uF 25V. The schematic should say for certain. 470 doesn't sound right, I would expect it to be much higher. There usually isn't anything else wrong after finding the shorted capacitor. Are the meter probes being used the right way? Red to the rail, black to ground.

[philipjfry]

you're my heroes. PC8209 does show up in the shematics, but i would not have realized that 9SC10U25V5KX-GP (see image below) actually contains the information i need. 

regarding the new resistance value, i may very well not have measured it correctly. i'll measure again and let you know! i'll also take a picture of the "little brackets". i too expected that only the solder pads would remain, but the cap just slid out in one piece and the "bracket" remained. it looked useful for re-mounting (keep in mind i only have a soldering iron), so i decided to keep it on for now. 

[philipjfry]

hi again!

i have only good news: i reassembled the laptop as far as necessary for testing and when i plugged it in, the battery LED lit right up. when i proceeded to turn it on, it appeared to work nicely, immediately (and rightfully) complaining about a lack of bootable devices. victory! i ordered a strip of 100 of these 0805 caps rightaway, costing me a whopping EUR 2.09 on aliexpress. for comparison, replacing the whole board would have cost me around EUR 350 for a used one and considering the state of the remaining components, that would not have been worth it.

prior to all that, i measured the resistance from DCBATOUT to GND again and this time, i measured no connection at all in both directions.  that's back on my crappy multimeters though, the 470  came from a fluke. must literally have been a fluke. 

afterwards, i cleaned the crime scene (where the cap killed my laptop) a little with my good iron and used some naphtha to remove the flux, so that i can give you a better look at those brackets. looks quite messy under a macro lens, i guess i'll give it another pass before i insert the new cap. anyway, you can see from the pictures below that they actually all look like they are mounted on "stilts", possibly for better cooling? it really does get scorching hot there due to acer's unnecessarily flat fan-curve / late thermal throttling. kinda devious imo, it makes their devices look good in benchmarks but because it's bad for the hardware, it also makes sure that people need a new laptop regularly...

anyway, i would finally like to thank you all again for your time and effort! i learned a lot along the way and there's no way in hell i would have been able to do it on my own. you are the best!

[xavier60]

Good to hear about the success so far. I have never seen that mounting method for MLCCs. Ill do some searching.

[LateLesley]

  I'm glad it's working. And I've learned something out of these last pics too, I've never seen SMD caps raised on tabs like that before. That's a new one on me!

I hope your new caps arrive soon, and you have your machine put back together and working, along with the satisfaction that you fixed it yourself! Well done.

Edit : @Xavier60 I'm glad i'm not the only one who's not come across that method of mounting. The only reason I could see for it, would be for more airflow to cool the component, but then that tells me they're running it on the edge. And I'd think it would get better cooling mounted to the board, as the power planes wick away the heat.

[xavier60]

It is a countermeasure for cracking, https://au.mouser.com/new/tdk/tdk-flex-crack-countermeasure-MLCCs/
Not a totally successful one it seems.

[philipjfry]

interesting! nope, didn't help on my acer board 

i'm glad you guys and gals also got a little bit out of it other than the warm and fuzzy feeling of having helped a guy out.

yeah, fixing something and restoring value to it is most satisfying indeed! i often do it even when it's not economical, for various reasons (environment, being used to a particular device, no good alternative...). this acer laptop is a particularly troublesome device: even though i am treating it like a raw egg, i have had to fix it many times already (display backlight under warranty, hinges, 1 fan, battery) but i am refusing to let it die young because it was somewhat expensive, is relatively lightweight, has the best ips display i have ever seen in any device, great speakers, great performance and it almost passes for a business machine (not the usual brightly colored kid's toy). also, it was a trial balloon if acer still make highly unreliable devices and if i should consider them for any of my clients (yes they do and no i won't  ).