Hi,

I have a Samsung HD204UI drive that had its STJ009 mosfet burnt up from the board. I was hot plugging a chain of drives, and this drive - the first in the chain - possibly got a surge of some kind. Upon booting, the drive made a bunch of clicks, at first I thought it was one of my CD drives on the fritz, since one of them had a disc inside and sometimes they make odd noises as they struggle to read in a disc on bootup. There was also a burnt smell, which I disregarded at first because previously I was working on an unrelated project that involved heat gunning a cable harness; it had the same smell, which I mistook for residue from that project.

Unfortunately upon bootup I found that this Samsung drive was missing.

On a second bootup attempt I noticed the fast clicking sound again, and after removing and inspecting the drive I saw the physical damage on the PCB: the STJ009 mosfet was completely destroyed, it's top cracked open, and left smoke traces on the PCB. After doing some research, I concluded that my choices are
1. replacing the blown STJ009 mosfet
2. replacing the entire PCB from a donor drive, then exchanging the firmware between the two drives
3. professional data recovery

As I have a desoldering station and could source another Samsung drive that used the STJ009, I opted for the first choice. The other two ones cannot be put into motion during the weekend anyway.
A problem was that the traces around the mosfet had been burnt up so bad, that they could not be desoldered from the corpse of the mosfet and had to be cut off. However, I managed to map up the traces, partially due to others on this forum posting pictures of the PCB with the chip removed, and partially because the drive I used as the ic donor (a Samsung HD103UJ with over 48k hours clocked on it) used the exact same layout.
The mosfet pins 5-8 are connected to a b240 diode and a choke. Pins 1-3 connect to the left side of the 474 resistor, as well as the bottom of the component immediately to the right of the mosfet (a ceramic cap?). Pin 4 connects to the right side of the 474 resistor.

I wired up everything in this way using 28AWG kynar wire, double checked if there are no shorts. I also checked all resistors on the board (all show resistance), as well as the TVS diodes in the power input (which also show resistance), so I was hoping that only the mosfet was bad.

Unfortunately upon boot I got the same clicking noise, and the mosfet burned up again... this time I removed it once it started clicking, but the top of the chip bubbling up and it had a burnt plastic smell as before. So it is most likely dead now.

What else can I diagnose on the board - voltages and such, to narrow down the problem as to why the mosfet burns up? I have a basic multimeter, as far as I know it can check for voltage, resistance, amperage (never used it for that though), continuity, and some form of battery test. I may be able to source a third mosfet on the weekend, and I'd like to do anything within my skills to restore the data from the drive - as long as it does not cause potential damage to the platters themselves.

Does the repeated mosfet burning, and the clicking of the drive at power-on indicate permanent damage in the head motor, that would require professional data recovery? I mean, do the symptoms indicate a damaged motor? The mosfet was blown, but none of the diodes near the power connector were, and the resistors on the PCB are all in working order. Upon powerup, the drive spins up properly, but the read head clicks 15 times and then the entire drive spins down. There is no buzzing sound that indicates a stuck motor, and no chirping that would indicate the head damaging the platter (but perhaps I'm just being overly hopeful when saying this).

As an experiment, I might connect up the hd103uj donor drive that is now missing its mosfet, just to see if it produces the same symptoms, provided that this does not cause permanent motor damage...

You need to try good pcb. If you need one, I can send it to you.

I've seen several instances of the same problem. In all cases the problem was not fixed by replacing the MOSFET. This leads me to suspect that either the coil develops shorted turns, or the motor controller fails in such a way that the gate drive to the MOSFET is permanently on rather than PWM-ed.

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

Typical inverting buck-boost converter:
http://www.hddoracle.com/download/file. ... &mode=view

_________________
A backup a day keeps DR away.

That sounds very troubling, it would mean that even if I switch the PCB, the motor will just burn out that one as well, wouldn't it?

Is there any measurement I can do on the board, to determine if this is the case, to avoid destroying another PCB?

I've ordered some irf7205 mosfets, which are pin compatible with the STJ009, and that someone reported to be able to use as a replacement.
Wiring them up to the HD204UI's burnt board again results in the chip burning up.

However, on the working HD103UJ that I extracted the previous STJ009 from, I could put in a IRF7205 in place of it, and it has been working fine for the past half hours (over 100gb data verified on it so far). So if anything, I can confirm that that the IRF7205 indeed works as a replacement for the STJ009. I tested it on an expendable drive, but luckily it survived.

I have a used working HD204UI arriving in the next few days, I'll be performing a PCB swap between that and the broken one once it arrives. I hope for the best, but if the replacement board burns up as well, then the problem is indeed most likely to be in the head preamp/motor. Then I'll have no choice but to use the services of a professional data recovery company to swap the heads out.

If I may ask, have you seen a PCB swap performed in any of those cases? If so, did it help, or did it burn up the swapped board as well?

Yes, ISTR that a member of this forum successfully replaced the PCB after similar problems with the MOSFET.

However, I would first confirm that the preamp is not shorted. To do this, measure the resistance between the anode of the diode (the -5V test point) and ground. Do this with the PCB on and off the drive.

_________________
A backup a day keeps DR away.

Okay, I measured the following:




Resistance:
Between #1 and #2, I got 1025 ohm when powered off, 180 ohm when powered on. This was a bit interesting, if I turned up the multimeter from 2000 to 20k, then the multimeter said -1.43 (with the board powered on). Not sure if that is relevant.

Voltage:
Between #1 and #2, I got a zero reading... so the test point is not getting -5V, which would explain why the mosfets burn up.
Then I tried the three pins which are all connected to that area leading to the test point, and got 0V on the bottom two. The pin marked as #3 read as 5V however.

This is all with the board removed from the drive.

Oh, and I should note that the resistance between #1 and #2 was the same (~1025) with the PCB on the drive and the drive powered off.

You should not measure the resistance of a component when it is powered on.

The resistance of 1025 ohms confirms that the preamp is not shorted.

Those 3 pins with the heavy traces are the -5V, Ground and +5V power pins for the preamp.

_________________
A backup a day keeps DR away.

OK, the donor HDD finally arrived today. Funny story, they sent a brand new WD SE 2tb enterprise datacenter drive instead by mistake... any other time but now and I could've appreciated that immensely. But I digress. The used HD204UI worked fine; had no errors that I could tell, seems it had an EXT4 partition on it once according to Getdataback. It was also a HD204UI/Z4 type drive like mine, and the PCB matched exactly other than the being a different shade of green (probably printed by a different manufacturer, but all the serials, and the component layout and order matched - I've seen some boards with the same PCB that had a few extra components). A good omen.

I swapped the PCBs, carefully put the drive upside down so I could observe if anything burns up again. Upon boot, the disc correctly spins up, no burning, and no noise to indicate errors. Unfortunately, the BIOS does not see the drive.

So I removed the firmware from each PCB (the small 8 pin Winbond 25x40 below the RAM) using a hot air station, and resoldered them swapped, being careful to keep the orientation of the chip right. There was an arrow on the PCB indicating the correct orientation, that should match the small notch on the corner of the firmware IC. Just to be extra sure I also made photos of the PCB and soldered it back according to that.

After the firmware swap, the BIOS sees the drive... fingers crossed, I boot up the OS, and the drive is there, exactly where it used to be, as if nothing has happened! I'm currently archiving all the content from it to a new Toshiba 2tb drive. This old Samsung now only needs to work for 5-6 more hours until the archiving is done.

Looks like the story will have a happy ending, and I can recover all my data by myself with the only cost being an extra drive. Thank you for the help fzabkar!
I might get a spare PCB later on so I can put the donor drive into service as well, as an external backup... better to have more of those than less.