Samsung Spinpoint HD753LJ again, any advice ?

[kekbom]

This has probably been covered here but I have to ask some info. I need it desperatively and to be honest a I haven't used forums much. Regardless of my computer studies.

Any advice would be invaluable.

The only HDD that contained ANYTHING of sentimental (given) value died just died the other day. May have occurred when I was sleeping but no errors or warnings of any kind. Just noticed in Win7 that I have one SATA drive too few and started to find out why...

Actually it's a story of my life : I bought too tiny Corsair SSD and moved all my pics, recently (years of family material, and not talking x-rated ones) to HDD to save space to this SATA2 Samsung Spinpoint HD753LJ.

A move I didn't think twice. In hindsight I may be lucky that my HDDs have lasted...

My PC's PSU is Antec's Truepower 650W and in same the SATA power supply chain that my Corsair SDD and 1 TB Spinpoint. Both working normally all this time, no problems to this date and hour. So a power surge ?

Unlikely but possible...

It does nothing. Nothing is given to BIOS, no mechanical indication of any kind. Just calm silence. I haven't made current measurements but nothing indicates a short. Ohmically +5 and +12 rails seems nominal to me.

I have access to a electronics rework lab (full-scale measuring), X-Ray machine, several BGA rework machines, deep but narrow knowledge about mobile devices etc... but not HDD

Been analyzing the PWB :

1. Both trancient Zeners OK, on both rails
2. Oscillator dead (supplied by MCU itself ?)
3. Oxidation relatively low, both sides of PWB
4. No DC-DC activity seen, step-down or SMPS (combined MCU & EM ASIC / PMIC ?)
5. Nothing warms up
6. Froze the drive to -15 degrees celsius, nothing
7. Rose it to +50 degrees celsius, nothing
8. Heated MCU stack, motor driver (SH6125B), diodes & ST MOSFET with Heat Gun, using common sense. No help
9. Did a triple visual check of solders & X-Ray analysis
10. Zero ohm resistor OK


Nothing !

This is frustrating without scematics & lay-out.

MCU / PMU dead ?

I think the adaptive parameters inside MCU stack ? can be omitted to get my pics from the drive. I already purchased a "new" working one with the same revision : A

The only difference is that Samsung's date ? on my drive is 2008.04 and the doner has markings 2008.01.

Is there some code inside the HDA itself that HAS TO MATCH with "correct PWB (or PCB)". Size same, model same, revision A same (revision of what PWB ? or whole S... ??)

On PWB : Rev5. Don't know about the other. I don't have it yet.

Is the date "just a manufacturing date or I'm a possibly looking at a different firmWare versions which mismatch with HDA code combining some kind of certificate.

From my point of view the difference got to be marginal. Of course this is not marginal if the clash with HDA checksum fails.

There is no external EEPROM so it's nicely integrated in MCU stack with other flash & RAM.


If anyone could advice It would be great. I'm sure this the place where someone knows. Scematics ?

I really hope that the HDA mechanics or ele hasn't given up the ghost. I think I was more screwd, regardless that the data doesn't vanish without a magnet

I have equipment and knowledge even to change the old MCU to another board but don't want to open the sealed pack.


Thanks in advance !

Attachments

[fzabkar]

If this is your PCB ...

http://farm8.staticflickr.com/7053/6921 ... 9e0e_z.jpg

... then the two 4R7 coils would be the inductors in the Vcore and Vio buck regulators that supply the MCU. The 5 plated through-holes would be the respective test points. The MOSFET choppers and Schottky flywheel diodes would be internal to the motor controller.

The anode of the diode adjacent to the 1R2 coil would be the test point for the -5V supply for the preamp. The STJ009 IC would be the PWM controller for this inverting buck regulator.

I realise that you have stated that there was no PWM activity in these areas, so that would suggest that either the motor controller is dead or is not receiving its correct supply voltages. You may wish to probe around the heavier traces for +12V and +5V.

BTW, you have a nice lab! (Do you work for Nokia?)

[kekbom]

Thanks for you're reply fzabkar !

That is the board, yes.

I know now more about the board.

No activity seen of any kind. I first probed the coils. VIO & VCORE definedly WON't rise, haven't measured if the VCM controller supply is OK.

But that didn't help with my question about the SWAP of a whole PCB from another drive which is surely OK :/

If the SWAP works I will get my data and probably thrash if after this because the wrong adaptive parameters make the usage unsure for best.

It probably won't last. Or swap the PCB back to the HDA it was tuned with.

Aside the fact that the HDA could also be dead, is there other reasons byside tuning that the SWAP of PCB to get my photos hell out of it won't work ?

FirmWare or other. AND it this FW completely inside EERPROM inside MCU ? code inside the HDA or just mechanics and electronics that works like the PCB commands ?

Problem with mismatch of code ? Because PCB should be exactly the same. Manufacturing date differs.

And yes, Nokia brings bread to my table At least for now...


Again, any advice about the PCB swap success possibility and why the failure after PCB swap might occur is invaluable !

I can change the motor controller from one board to another. According to previous guru this part may be the culprit, if it's fed.

VIO and VCORE are more than familiar to my but the SMPS's are inside the VCM cotroller with the external coils ? Seems logical because in this case the MCU is not fed with VCORE & VIO. AND still logical : No indication of life sent to BIOS through the data lines.

Nice day by the way, sun ! At least here with reindeers, LOL

[fzabkar]

I'm not a data recovery professional, so I can't advise whether a straight board swap will work. In fact the professionals argue over this subject amongst themselves. FWIW, I have seen threads where a straight PCB swap has worked, and others where a "ROM" transfer was necessary. I suspect that your chance of success may depend on the model. AFAICT, it won't hurt to try.

The PCB contains some boot firmware, but the bulk of the firmware is stored in a reserved area on the platters. I can't answer your questions with any authority, but perhaps the following articles and service manuals (for older models) will provide some insight into your drive's architecture.

newbie-info-from-and-for-newbies-about-firmware-etc-t6562-20.html
http://files.hddguru.com/download/Datas ... 20Manuals/

You are correct about the SMPSs being incorporated within the motor combo controller. In earlier designs the chopper and flywheel diode were outside the chip, with the PWM control coming from within.

The following datasheet may shed some light on how a typical motor controller works:

L7250, SMOOTH, spindle motor + VCM controller, ST Microelectronics:
http://wandrew.regruppa.ru/PCInfo/TechDoc/L7250(Smooth).pdf
http://www.datasheetcatalog.org/datashe ... Xyuswx.pdf

Here is another article that may be of general help, but it discusses a much earlier model:
http://www.users.on.net/~fzabkar/HDD/Tu ... 0411N.html

[kekbom]

Thanks for the L7520 datasheet.

Gives some directions but been unable to find TI SH6125B PIN diagram. The layout of pins do not match, regarless the similar design approach.

But...

HDA's PreAmp is unpowered at all times :
"The anode of the diode adjacent to the 1R2 coil would be the test point for the -5V supply for the preamp" (flat) !

So,

I should find

- preAmp feed from 1R2 coil = zero
- VCORE for Marvell's nightmare from 4R7 coil = zero
- VIO for digital IO from another 4R7 coil = zero

No activity near any of those 3 bigger coils. These leads to conclusion that SH6125B is not breathing. It's start-up logic is also a mystery what does it need the wake ? +5 at least (to generate VIO & VCORE, clock?, +12V), some other wake up signal.... ?

I got to trust that this part is changeable and not a security component with AGAIN memory.

Actually the only +5V I measured / found was from the capacitor next to DIP MOSFET.

Trying to understand but I need the scematics & lay-out of this board to even figure out is the SH6125B itself powered.

What are the pins ? Left my dear microscope at work so it's kinda hard now when I'm home. Well, another day tomorrow and my donor drive should be in mail by now.

Good side is that this seems to indicate a PCB failure (holding my breath). Changing the SH6125B from the other board is not a problem. I understand that this component can withstand high temperatures during it's own operations. So maybe it can take the heat from FineTech's PICO without damage when I change the sucker The middle ground is possibly hard to detach but it will come out.

Should I first try a direct PCB swap OR maybe change SH6125B or MOSFET... I think I will try luck with swap. But the amount of opions about the people who are positive that this can damage the HDA. But others say, hey, no problem man just do it !

Cannot even try to figure out how to copy a part of ROM from a dead MCU stack's flash. But first I need POWER and it obviously isn't generated ! Neither is the oscillator buzzing

I gotta admit that I don't exactly understand what is the DIP MOSFETs role here (AMP ?) but found +5V near it so I might to study it also.

BTW +5V rail resistance to GND seems very low. Just measured it : ~2.25 kohms.

+12V to GND is ~270 kohms

+5 to +12V about 275 kohms.


Samsung's scematics seems next to impossible to find. Do they have a vault where they stash them and nothing ever leakes ??

[fzabkar]

I think you have Buckley's chances of finding schematics or datasheets. If they've leaked into the data recovery domain, then you'll never see them. If you watch this forum for long enough, you'll come to understand that even the most trivial pieces of information are guarded as if they were potentially industry busting revelations.

In any case I don't think schematics will help very much. A datasheet would be very helpful in recovering the MCU's internal flash memory contents, though. AIUI, Samsung PCBs have a serial console interface that gives you access to the MCU. Of course you would need to fudge the Vcore and Vio supplies, and you would also need to control the POR (power on reset) signal to the MCU. POR is generated by the motor controller's power monitor circuitry.

If by "DIP MOSFET" you are referring to the STJ009 device adjacent to the 1R2 coil, then this is the chopper in an inverting buck regulator. There should be an example in the application circuit in the L7250 datasheet.

As for the +12V and +5V supplies for the motor controller, I would expect that the +5V rail would supply the chip's digital logic, and +12V would power its analogue section. I believe that each rail should have decoupling capacitors, so those are the components I would be using as test points. I expect that one of the larger capacitors will be part of a charge pump and will have a voltage of the order of 18V.

[BlackST]

In any case I don't think schematics will help very much. A datasheet would be very helpful in recovering the MCU's internal flash memory contents, though. AIUI, Samsung PCBs have a serial console interface that gives you access to the MCU. Of course you would need to fudge the Vcore and Vio supplies, and you would also need to control the POR (power on reset) signal to the MCU. POR is generated by the motor controller's power monitor circuitry.

"Opinions" ...

[kekbom]

Success, all data recovered and drive works "normally" again

Thanks for the help !!

AND learned a LOT !

Straight PCB SWAP failed but heard that the motor woke up from the dead and fault left to PCB.
Wrong PCB(!) revision. Dates on drive(!) : 2008.01 (new), 2008.4 (original)

Revisions 004 and 005. If I remember correctly.

Out again and heat gun on !

Swapped VCM controller and no-problems on repaired one anymore, IC was simply dead.

PreAmp got -5V now and VIO & VCORE jumped up. Found eventually life from those coils. No hassle after the chip change except that it was quite a thrill even for me who has changed thous... a lot of uBGA & LGA components. Not really a novice in that area despite the ugly looking change

It really puts a very cold circle under your ass when you're own data is in question

My original was TI SH6125B got replaced by SH6125B_P (only difference I found near P ending version was that new ASIC's two lines were jumpered together with SMT jumper).
Two bottom solder terminals (adjacent), one 0 ohm resistor which was forming a triangle of three jumpers with other two. A solder bridge would have resulted the same

Measured ~2.5 V during operation when attached, big difference without (don't remember). No idea what lines those were (even the SH6125B_H version, works without it though) !

BTW never use oscilloscope probe to measure the external crystal during operation, it will crash the drive. And my Agilent's probe sure isn't bad
Or at least it did it for me and took a long while before it recovered. Scared BTW

"A datasheet would be very helpful" ? YES, indeed !!!

For example what the heck was that zero ohm resistor / jumper and what's with this "_P" version ??

I will try to hunt one of those TI SH6125B (ummm... maybe ten because at this rate this could be seen as a problem in the future, got three Samsung SpinPoints) from somewhere to repair the PCB which got ripped from it's power ASIC. I hate to so see the sad and dead HDA on my workstation

Come on Samsung ! give my something harder ! Or NOT....

Attachments

Got a finnish version of CrystalDiskInfo... "Hyvä" is "Good" :) sorry about that. Sure has served some hours also :o

It ain't the prettiest job I've made but it will hold ! The GND cooling plate under it will make it last more than me...

[Vulcan]

Success, all data recovered and drive works "normally" again

Congratulations

In case any of the remaining components received EOS during the failure of that chip, (e.g. did Vio or Vcore exceed tolerances during the failure sequence?) I would not personally use that disk except for non-critical work.

BTW never use oscilloscope probe to measure the external crystal during operation, it will crash the drive.

This is a common issue with crystal oscillators - attaching an oscilloscope probe can (depending on a couple of factors) potentially stop the oscillator.

And my Agilent's probe sure isn't bad

For this specific task, the make of probe is not as important as the type of probe, in my experience.

Again, congratulations.

[fzabkar]

Hyvä work.

[kekbom]

Thanks !

Especially fzabkar


This site is great, hopefully this thread can help others also especially if the VCM controller fail occurs like it blew on me.

For non-critical work ? Indeed !! Now the sentimentally valuable data is stored optically and to another, new drive.
Actually still curious will the drive actually last and work on long run but as far as I can tell the chip change succeeded.

But of course, the chip I used is old ALSO. But the heat needed for soldering was immerse (bottom plate soldering).

So definedly nothing critical goes to that drive.

Does the pictures of old girl friends count ??

What it comes to measuring the oscillators I have also witnessed same behavior before with mobile phone oscillator's also. Many times the measuring attempt of systemClk or sleepClk has crashed the phone HW also.
Some times seen wave, sometimes a crash

Given that my probe has seen a lot of use and served me long and mark can be irrelevant. Serves it's purpose.

Happy ending here and studying the HDD was also interesting in some weird way. Can't deny this.

We also have a clean room here so even the opening could have been possible but luckily it was not needed.