MOSFET or Capacitor ? which one creates the short ???

[hgg]

Hi guys,

I am trying to repair a used laptop I've just bought with a motherboard that refuses to power on.
I've installed a new mSATA SSD, installed Windows normally and after a few reboots and a power off
the laptop never turned on again. 

No burnt fuses on the motherboard, power comes in normally and everything I've checked turned out
to be working fine except the following.  Two of the 5V MOSFETS driven from the peak current mode
power controller look like they have a short.

Below is the circuit that has the short:



MOSFETs Q89 & Q90 have a direct short between drain and source.   I have removed all power and CMOS
battery, have discharged the gates and the short is still there.

Capacitors C601 & RFC30 are shorted as well.




My question is which component do you think its more likely to create the short?
One of the capacitors or one of the MOSFETs?   
I don't want to start removing components randomly, especially these tiny capacitors.
Is there any other test I can perform to isolate the faulty component?

Thanks!
p.s. The full schematic of the power controller:  https://postimg.org/image/t3nv3wycj/

[capt bullshot]

One way to find the shorted component could be:

Inject a few amperes from a lab supply into the shorted node and watch for rising temperature, voltage drop on the PCB traces, ...
The easier way would be to desolder the components and check them one by one.

[danadak]

I would unsolder L to isolate upstream side of converter. With your MOSFETs removed,
their gates tied to source, is Rds very high ? If so they are probably OK, asasuming
they are enhancement mode devices.


Regards, Dana.

[hgg]

So there is no other way to identify MOSFET vs Capacitor without desoldering these 'microbes'..?

captbullshot's idea sounds promising but it seems a bit dangerous if you do not know
exactly what you're doing.  Would 5V / 1Amp be enough across Q90 for example ?

danadak do you mean tre L26 inductor?

[xavier60]

If you have a multimeter that has microvolt resolution you can trace the current path to ground by measuring voltage drops across very low resistance current paths like through the inductor or even the ground plane. I use about 200ma and a cheap AN8008 multimeter.

[hgg]

Hi, yes I have one.
Can you elaborate further on the setup I should use to measure the voltage drop you say?

[xavier60]

You could feed some current in to the right side of L26. For a start, measure the voltage drop across L26 to see if the current is flowing through it or not.

[capt bullshot]

Connect "-" output of your supply to GND (one side of the suspected short)
Set the lab supply to a low voltage (near 0), and a limited current (maybe 200mA as xavier supposes, maybe you'll need more)
Connect "+" output to the faulty node, somewhere on the PCB trace leading to the suspect components (other side of the suspected short)

Carefully increase the power supply output voltage setting (stay below 1...2V to avoid damage to your board) and watch for the current rising, while voltage stays low. If the voltage rises and current is low, you've done something wrong, start over.

Now, using the uV DMM, try to figure out where the current goes to. Measure various voltage differences along the PCB traces to the suspected components, a "good" component will have near to none voltage drop along the trace, the shorted one will have a noticable voltage drop across the PCB traces.

Your PCB layout (from the photo) doesn't look well suited for this method, since the components have very short and wide traces, it'll be difficult to figure out the path.

Suitable points to contact your power supply might be: upper pad of the 2R2 resistor (for "+") and lower pad of the small capacitor below the 2R2 resistor ("-"), where your C601 arrow points to. Then you can measure the voltage across the traces from this point to e.g. the pads of the MOSFETs.

[hgg]

Ok, for starters I applied 30mA and 3V (maybe a bit high voltage?) to the right side of L26 and ground.
I got 0.12mV across L26.

Shall I continue as captbullshot suggested ?
Thanks.

[xavier60]

I took some measurements on 2 laptop inductors at 200ma. A 3.3uh 5v rail inductor dropped 2mv. A 0.36uh VCORE inductor dropped 200uv.
The voltage drop across a few mm of ground plane can easily be measured. I take most measurements on the ground plane to find where the current is returning to it. It is best to connect the power supply negative and DMM negative to different points on the ground plane.

[xavier60]

Ok, for starters I applied 30mA and 3V (maybe a bit high voltage?) to the right side of L26 and ground.
I got 0.12mV across L26.

Shall I continue as captbullshot suggested ?
Thanks.

30ma at 3v isn't really a short.

[hgg]

What do you mean its not really a short?

I used current limiting on my power supply...
Should I increase it to 2V 200mA ?

For a start, measure the voltage drop across L26 to see if the current is flowing through it or not"

What does the voltage drop across L26 has to do with which MOSFET or Capacitor is burnt?

[hgg]

Suitable points to contact your power supply might be: upper pad of the 2R2 resistor (for "+") and lower pad of the small capacitor below the 2R2 resistor ("-"), where your C601 arrow points to. Then you can measure the voltage across the traces from this point to e.g. the pads of the MOSFETs.

Its very difficult to apply voltage to these points.  They are extremely tiny and hard to reach.
For example the MOSFET is 3x3mm  But if I manage to solder some magnet wires across
2R2 and C601 which point pairs should I measure next?  Top of 2R2 and drain of Q90 for example?

Sorry for my many questions but I have not tried this method before.

[xavier60]

What do you mean its not really a short?

I used current limiting on my power supply...
Should I increase it to 2V 200mA ?

For a start, measure the voltage drop across L26 to see if the current is flowing through it or not"

What does the voltage drop across L26 has to do with which MOSFET or Capacitor is burnt?

To get an idea of what's happening, we need to know the simultaneous current and voltage readings. 
The voltage drop across the inductor indicates that the current is flowing through the inductor to something on the left side.

[hgg]

Yes, but the problem is on the right side.
How does this info differentiate a burnt capacitor vs a burnt MOSFET?

[xavier60]

Yes, but the problem is on the right side.
How does this info differentiate a burnt capacitor vs a burnt MOSFET?

To be certain of that, you can feed the current to the left side of L26, then measure the voltage across L26 again.

[capt bullshot]

Suitable points to contact your power supply might be: upper pad of the 2R2 resistor (for "+") and lower pad of the small capacitor below the 2R2 resistor ("-"), where your C601 arrow points to. Then you can measure the voltage across the traces from this point to e.g. the pads of the MOSFETs.

Its very difficult to apply voltage to these points.  They are extremely tiny and hard to reach.
For example the MOSFET is 3x3mm  But if I manage to solder some magnet wires across
2R2 and C601 which point pairs should I measure next?  Top of 2R2 and drain of Q90 for example?

Sorry for my many questions but I have not tried this method before.

OK, you got it somehow wrong. The intention is: solder the two wires to suitable points at your PCB. There might be better accessible points available than the ones I said, you'll have to find them. Leave them there, apply the current through them.
Then take your DMM, use small and pointy tips to measure the voltage along the PCB traces, starting from one of these points. E.g. point one tip to the top pad of the 2R2 resistor (where you soldered the wire), the other at the pad of the nearby MOSFET that is connected by the PCB trace to the starting point. Read and note the voltage (should be a few microvolts). Next: leave the first tip in place, move the second to the other MOSFET, read and compare the voltage to the first reading. Now make your decision where the current goes to (there where's the larger voltage drop). Follow more traces if necessary.
As xavier says, this procedure can also be done across the GND plane of the PCB, it depends ...
The GND plane (if exists) is often easily accessible through many points on the PCB, you'll have to find them.

[hgg]

@xavier60:   0.0030mV

[xavier60]

@xavier60:   0.0030mV

That indicates that the short is on the left side of L26.
 I still need to know what the power supply voltage and current readings are while your power supply is connected to L26.
Saying 3v and 30ma indicates 100 ohms, which isn't really a short.

[hgg]

Ah ok, I see what you mean.
The current is maxed at 30mA and the voltage is almost zero.

[hgg]

@captbullshot:  I am measuring across top of 2R2 & Q90 drain and top of 2R2 @ Q89 drain the
same voltage drop:  0.0030mV   I've applied 2V @ 30mA.

[xavier60]

Ah ok, I see what you mean.
The current is maxed at 30mA and the voltage is almost zero.

That is a short for sure. Increase the current to 200ma. you need to find the place on the ground plane that has the highest voltage across it. This will be at the place where the current is flowing through the shorted component to ground. Hopefully it is a faulty capacitor and not a large BGA IC which this method doesn't work very well with. Read the previous posts. 

[capt bullshot]

20mA is to low a current. You'll need more voltage drop along the traces. Try to increase up to 200mA, maybe 2A, but watch the voltage at the output of your supply. It is supposed to stay low due to the short.

Also follow other traces, especially to and through the inductor, as the short may also be at the other side of the inductor.

Measure the voltage across the 2R2 resistor, should be near zero, otherwise the capacitor in series with it failed.

[hgg]

It looks like this is beyond me...
I started measuring capacitors all across the board and they all display a short.
I am not sure how can you go debugging this.
Anyway, guys thank you for your help!!

[xavier60]

Sounds like it has more that one shorted rail, I wonder what happened to it?

[capt bullshot]

There might also be an issue with the meter, indicating a short across the capacitors in circuit due to the properties of the circuit. A somewhat higher test current will show the truth.

[xavier60]

Some of the low voltage rails will measure a low resistance like tens of ohms. This is normal.

[hgg]

Sounds like it has more that one shorted rail, I wonder what happened to it?

That's a mystery!
After I installed a mSATA SSD, I installed windows.
Everything went fine and I did a shutdown for a last time to install an antivirus.
It never switched on again!!!

[hgg]

Some of the low voltage rails will measure a low resistance like tens of ohms. This is normal.

Indeed I've measured some with 20-40 Ohms, but many have direct short across them.

[xavier60]

You should check to see what rail they belong to, maybe the same 5v rail fed by L26.

[capt bullshot]

Some of the low voltage rails will measure a low resistance like tens of ohms. This is normal.

Indeed I've measured some with 20-40 Ohms, but many have direct short across them.

My guess: these are connected to the same power rail as your MOSFETs. Ceramic capacitors can fail short without an obvious reason, though this is rare. For sure, they do not fail all at the same time.

Edit: I'll leave the forum for some real world stuff now ...

[xavier60]

That's right, all capacitors on that rail will appear to be short but they are not really, it will be only one of them.

[hgg]

Yes I know, but it would be very difficult at least for me to find the real cause.

This kind of troubleshooting would be a great tutorial.

Luckily I did not remove any components which in this case could be a bummer..
Do you know if there is any tutorial on the method you suggested by tracing the
voltage drop across the power line?

[xavier60]

Yes I know, but it would be very difficult at least for me to find the real cause.

This kind of troubleshooting would be a great tutorial.

Luckily I did not remove any components which in this case could be a bummer..
Do you know if there is any tutorial on the method you suggested by tracing the
voltage drop across the power line?

That is what I'm trying to explain how to do.
Or you can risk taking a short cut by applying more current and trying to find something getting hot.

[hgg]

Thank you for your time and effort to explain it.
I might follow the power line in the same way to see if I find anything.

[hgg]

Ok, only the 5V rail is shorted.

I've applied 3V / 3A at right side of L26 and this is what I got :
(don't blame me, blame Rossmann...)




The temperature was around 40C. 
You can see the power leads on upper left and between them L26.
Nothing else was getting hot and I started with low currents.
It looks like that there is a short in the main CPU !

Is this possible or common? 

[capt bullshot]

Is it the chip or a nearby component getting hot?

The main CPU of a laptop isn't connected to a 5V rail, afaik. They use lower voltages for core and I/O.

[hgg]

I know.  Its around 1.2V I think.

The hot spot though is on the lower left corner of the CPU where its mostly empty.
Back side as well.  The hottest spot is actually on the die (IC).

[capt bullshot]

Ouch. That's no good.

What's the white stuff on the vias around the CPU, between the capacitors on the chip carrier and at the edge of the chip carrier? Looks unusual.
If it's thermal grease, this isn't supposed to be on the PCB. Don't know if electrically conductive thermal grease exists or was used here, if so then this might be the reason.

[hgg]

The white stuff is just the silk screen.  I don't know why they marked the vias.
The stuff between the capacitors is thermal grease and the chip was full to the brim...
Its not conductive though.

I did the test again with 1.5A and you can clearly see the temperature ramping up
pretty fast at the lower left corner! 



What kind of short is that!?   

[capt bullshot]

Did you check the other side of the PCB?
From that picture, I suspect not the chip itself, but rather one of the small caps on the carrier, or something underneath / inside the chip carrier / PCB or a component at the bottom side of the PCB.

Get a ZnSe lens to improve the close-up capabilities of the thermal camera, Fraser has written some posts here regarding that. Quite interesting stuff.

BTW: I'd probably give up at this point ...

[hgg]

Yes I did.  Nothing.  Cold.
Its something on the top side of the board.  Probably inside the chip carrier or
some of the BGA balls underneath.  I am guessing that the motherboard is shot.   
What do you think?   If it is I might try to remove the CPU just to see what's going on.
I am not sure though if my Chinese hot air station Atten 858 is up to the task...

The ZnSe lens is very interesting.  If you could zoom in with the Flir it would be perfect!

[capt bullshot]

I have no idea what's the cause for this, IMO further digging results in destroying the board, maybe finding the cause but not able to put it together again. Depends on the equipment and experience you have.

Some people got really good looking results with these lenses and the E4 or E40 - https://www.eevblog.com/forum/thermal-imaging/flir-e40-upgrade-configuration/msg1299934/#msg1299934

Removing the CPU for curiosity's sake? Yes, why not. I've done such things with a common heat gun from your local home improvement store, just be careful not to overheat the PCB and don't expect to be able to solder it back again.

[hgg]

Very interesting!
I think I will give a try to the 18mm / 50.8mm FL lens.
https://www.ebay.com/itm/HQ-ZnSe-GaAs-Focus-Focal-Lens-K9-Mo-Si-Mirror-CO2-Laser-Engraver-Cutter-FL-1-4/261992318293?hash=item3cfff58555:m:mv8JVfumF6QQYwzUTQ4ZeJA

As for the motherboard I think that I will indeed remove the CPU just to see what's going on.
Very strange problem though.

[hgg]

After a lot of searching it turns out that the 5V rail was sorted by the VCC1R05B_VTT line from the CPU via
a switching regulator (Volterra VT357FCX) .

I removed the inductor L30 that was connecting the two lines and the short on the 5V rail cleared.



After that, I injected 0.7V/1.5A current on pin 2 of L30 and I got the same thermal signature on the
CPU as before.  It looks like that the CPU is causing the short but I am still not 100% sure.

When I check continuity from the input of the VT357FCX regulator (VDD) to the output (VX#D) I get
a dead short with the output floating.  (since I removed L30)

My question is, can you deduce from the above that the regulator is toast as well,
since its input is shorted with its output?

[capt bullshot]

My question is, can you deduce from the above that the regulator is toast as well,
since its input is shorted with its output?

Yes, consider it broken down. Interesting guesswork which one failed first - the regulator shorted input to output and toasted the CPU or vice versa ... I'd guess the regulator failed first, a short to GND on it's output usually doesn't kill them.

[hgg]

Interesting indeed.

When I installed the mSATA SSD, everything worked fine until that final shutdown.
After that, I opened the laptop, I removed the SSD and when I tried to reinsert it,
it was difficult and the SSD would not snap into place again.  I looked in the pins of
the mSATA connector and I saw that some of them were bent inwards..  Could that
be the cause of the short somehow?  Although it was inserted fine in the start and
Windows were installed without any problems.

[capt bullshot]

You'll have to investigate if the bent pins made a short between the 5V rail and some other power supply. Older SATA devices had 12V on the power connector, so a short from 12V to other signals could be destructive. I don't know if a short from 5V to the SATA data lines is destructive.
I wouldn't expect a short from 5V to GND to cause your failures, but one never knows ...

[hgg]

There is no 5V rail on the mSATA connector.  Only 3V & the VCC1R5B_WWAN line
which it might be connected with the 1.5V rail that has the short.

Left and right of that pin are the SMB_CLK_3WAN & GND lines and below are the
SATA2_RXN & GND.  Its more difficult to short the adjacent pins of the mSATA and
easier to short top and bottom.  But the thing is that the laptop worked fine, the
SSD was fully in position and only after the shutdown it stopped working.