I just got this drive. Customer told me that it "stopped working", which it indeed did. Does not respond to anything. I took a look in the microscope and found a damaged MLCC capacitor. Other than that I cannot find any visual damages.

The issue here is that the drive is encrypted via Bitlocker. Hardware encryption is disabled, though, so I think the only way to recover it is to repair the PCB in case that's the problem. However, I cannot rule out that the controller (Infineon) is dead or not. I cannot get a response from anything this far, so this - for me - one of the more trickier cases I've had so far. What I can rule out is that there is no short circuit on the PCB

Any tips of what I can test to get a hint of what the root error is?

I would help you, but your photos are cr*p.

Measure the voltages at the inductors (I can see 5).

Tip: How to take a good PCB photo:
http://www.hddoracle.com/viewtopic.php?f=112&t=120

Tutorial - Linear and Switchmode Regulators used in HDDs (and SSDs):
http://www.hddoracle.com/viewtopic.php?f=59&t=231

Quicker-and-dirtier switchmode regulator:
http://www.hddoracle.com/download/file.php?id=6446&mode=view

_________________
A backup a day keeps DR away.

you have to consider Open Circuits as well as shorts. if a fuse or resistor has blown open circuit, you wont find a short..

@fzabkar: My friend's cellphone was obviously better than my EOS 400D, so here are some way better pictures to look at.

@HaQue: You are right, ofcourse, but other than the blown MLCC (right next to the text "U26") I find no other hints of open/shorted components.

If you can't discern any part markings in your photos, then neither can I.

Sorry, this is the best I can do.

_________________
A backup a day keeps DR away.

I ran out of time before I had to pick up the kids at school so I never managed to do some exhaustive measurings. Will continue tomorrow.

What I managed to check was that it's a PIC16 as you guessed, and DMM shows a stable 2.8V on a Vcc-node. I will measure the inductors and scope some nodes, because something tells me that your hints are probably where the root cause is. If Vcc is only 2.8 then my hunch says that I will either find noise or some other discrepancies when I hook up the oscilloscope.

This case is currently a bit out of my league, but it just got very interesting.

Measure the voltages at the inductors. If any voltage is out of spec, check for shorts to ground. Also check that the enable pin of the corresponding buck regulator (switching component) is active.

If any of the enable pins are inactive, then check the power controller IC (PIC16Fxxxx). This chip sequences the various regulators in a preprogrammed order.

The fact that you are measuring 2.8V somewhere indicates that the e-fuse is OK.

Datasheets for SSD and Flash Drive ICs:
http://www.hddoracle.com/viewtopic.php?f=100&t=1577

_________________
A backup a day keeps DR away.

The two large inductors on the top of the image (L2, L3) shows a voltage drop of 0V. L1, the third large, drops 2mV. I am unable to measure on the two small ones, L4 and L5, due to them being soldered beneath. I had to reflow them both and move them a tiny bit to the side to be able to get down with the probes. They both now show about 1mV drop.

Also, after I moved them, L3 now shows 50mV drop instead of 0V. L2 is still 0V, though.

So, four of the five inductors shows a drop (I am unsure if they are within the specification or not), while the path near the fifth seems to be shorted somewhere.

The two top regulators are both MP28258DD, and both enable pins are high.

The large 2*regulator between L1 and L4 is marked "B7DG\n314" but I cannot find a datasheet for it. The last regulator, straight under L5, is marked "6QD\n249" but I cannot find a datasheet for this one as well.

So two regulator-chips out of four are confirmed being enabled, while the two others are unconfirmed.

Measure the voltage at the junction of the inductor and capacitor, ie between the inductor and ground. That's Vout.



Datasheets:

MP2109DQ, Monolithic Power, Dual 1.2MHz, 800mA, Synchronous Step-Down Converter, 2.5V - 6V in, marking B7xx:
https://www.monolithicpower.com/pub/media/document/MP2109_r1.4.pdf

MP28258DD, Monolithic Power, 3A, 4.2V - 20V Input, Synchronous Step-down Converter, marking AAA, QFN12:
https://www.monolithicpower.com/pub/media/document/MP28258_r1.14.pdf

MP2136, Monolithic Power, stepdown converter, 2.5V - 6V Input, 2MHz, 1A, marking 6Q, QFN6:
http://www.mouser.com/ds/2/277/MP2136_r0.9-478180.pdf

https://www.monolithicpower.cn/cn/documentview/productdocument/index/version/2/document_type/Datasheet/lang/EN/sku/MP2136/document_id/430/

_________________
A backup a day keeps DR away.

Now I've confirmed that all enable-pins on all regulators are high.

Also, I've found a short:



The top right capacitor is shorted, so I removed both capacitors, both inductors as well as the capacitor near JP19 (all marked with a blue dot), but the short still remains. The components are fine (verified with DMM when desoldered) so the short is somewhere else.

I will have to dig around more to see if I can find it.

There was no need to remove components from the second regulator circuit -- they are two separate entities.

I suspect that the two regulators supply the Vcc for the NAND arrays on each side of the PCB (or Vcc + Vccq). Hopefully there is a shorted bypass capacitor at one of the NANDs, otherwise a NAND itself may be faulty. ISTR that @pcimage was able to locate a shorted cap by sensing its temperature with his finger.

_________________
A backup a day keeps DR away.

Thankfully I own a FLIR-camera, so finding a shorted component that lets through some current is never difficult. However, for this case I see no thermal signatures at all otherwise on and beneath the controller. There are about 10 capacitors right under the controller that belongs to the shorted path, but desoldering them all yields nothing: there is still a short from Vout to GND.

The one thing I haven't checked is to desolder the controller (BGA) to see if that's the one containing the short. I really hope that one of the NANDs are not shorted internally.


EDIT: the FLIR says that the controller lies steady around 36 degrees C. It never goes beyond 40, and if there was a short the temperature would show a significant larger value I think.

What are the values of the working supply voltages? Maybe you could eliminate potential suspects by determining the voltage requirements of each component. It does look like the controller may be faulty, though. :-(

If none of the bypass caps at the NANDs are shorted, then the NANDs should be OK.

_________________
A backup a day keeps DR away.

On Vout near L2: 2.86V
On Vout near L3: 1.03V

Vcc near the NAND-chips all show 2.86V.

The resistance at the Vcore supply will normally be a low value (~100 ohms ?). Hopefully you are not confusing this with a dead short.

_________________
A backup a day keeps DR away.

Now I realize that when I checked for shorts, I used the continuity tester. I assumed beep = short, but just now I made an actual ohm-measurement and I see 27.5 ohms on the path I assumed was shorted. The continuity beeps up to 80 ohms or something like that.

I have in other words spent ~4 hours chasing my own tail. How embarrasing. On the plus side is that I have now learned something from it, which is good. On the minus side is that I'm back on square one.