Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Hard Drive Failure *Cries* - Engineers to the rescue?

Status
Not open for further replies.
The interesting thing is I explained what was happening to Kevin at OnePCBSolution.com, one of the companies who do the board + ROM swaps and he kindly replied with "To be honest, I don't think your drive has a circuit board problem.

How exactly did you explain the problem to him? Because, from what I know, having or not having the drive show up in BIOS is entirely the effect of board firmware. It doesn't have anything to do with the disk and it's integrity. It's the first step in the process, and is very low level. It would be hard to imagine that a disk problem could make the board SoC unable to tell your system who and what it is. Unless of course it was so bad that the motor/arm load exceeded specs to the point that it messed with the rest of the electronics.

I would try a board swap. Don't even bother swapping the chip, just the whole board. If it's not at least recognized in BIOS, then maybe his point is valid. Then again, you DID buy a used sacrificial HDD, which potentially had board problems of it's own.
 
Hi Oblivion,

This is the email I sent describing the problem:

"The problem with the laptop drive is that it's not being detected in Windows. When I take it out of the laptop and put it into a USB caddy, it still is not detected by windows but it comes up in Device Manager. All recovery programs show it to be 2TB large, when it is actually only 320GB. After it failed, for a brief period of time I was able to access some data but now nothing. The drive spins up fine, but it's just a constant faint spinning, and it doesnt sound like any data is being accessed. No clicking or bad sounds, drive sounds fine. Do you think the PCB is faulty from this description and is this something you can fix?"

Haven't had any other responses from the other companies yet regarding their thoughts, but it's only been a day or two. In the meantime, I tried soldering that little ROM off one of the PCB boards I have left over from my earlier experiments on order to get some practise in. What a big fail. Even my finest soldering tip doesnt come close to being able to get in close enough to heat the area and lift the IC off.

On my journeys across the youtube land, I did find an interesting alternative method for removing such IC's with a halogen desk lamp!
https://www.youtube.com/watch?v=HkO71G4wvo4&list=UL

I actually have one of these lamps, but my bulb is rated at 11W. Would be interesting to try this technique though! I've heard that if you're worried about damage to other components, you use something called Kapton tape which is sort of a yellowy-gold colour and is able to withstand high temperatures. Just cover up any other components with it. Doubt I'll try this for now but a good interesting idea nevertheless. I'm just going out now to see if I can find a friendly TV repair man/shop to get the chip off for me and put it on the new board :)

Regards,
Megamox
 
Kapton is heat resistant. I have never checked its heat insulating properties. A ready alternative is aluminium foil. The air space provides insulation. If you crinkle it a little it may work better. It can be easily conformed to what you need.

()oblivion has already given you the best answer for desoldering, unless you have a special tool or hot air. Although a halogen lamp gets quite hot, you can't direct its heat easily, and you will end up heating a large area. If you have a good hot air gun, that might be preferred for removal.

John
 
From that email, I too would probably believe that the board was fine. But... having read this thread, I know more about the whole situation than that person did, and that makes me think differently.

It is likely being detected in device manager when in the USB caddy because the USB caddy has it's own board and drivers that go with it. *THAT* is what is really being detected. This does not mean that you are getting any closer to the data in the drive, in fact, you inserted an extra layer in between the drive and computer, probably making it harder. The result you are seeing can easily be explained by the USB caddy try to enumerate the drive and failing because the board is bad, but of course the USB caddy isn't bad, so the it is able to do it's thing to some extent. This would be enough to be seen in windows, as it is still getting detected at the USB layer.

So... I am STILL convinced that the board might be dead. Probably because of the excessive load seen when it was trying it's damned hardest to smash the head onto the platters while the head assembly was bent to hell. I could see this taking out the arm coil driver chip, and then some of the rest of the electronics when the chip was being seen as a short to GND/VCC.

Anyone that disagrees with this assessment, feel free to chime in.

The EEPROM
If you can't do the soldering work, you can always try and find someone else that can. I would probably do it for nothing, and even rip the firmware for you for my own kicks, but you live in the UK and I live in the US. Now it's somewhat rude to volunteer someone else for something, but I believe Alec_t lives somewhere around that continent, plus he is highly skilled. Maybe if he feels like it and you're super nice to him, you could get(pay) him to do it?

No guarantees of course.
 
Hi Guys,

Thanks for your responses. I've got some more opinions to share. I visited the local TV Repair man and he couldn't remove the chip, said he didnt have the correct tools. As I was walking out, he said 'But I do know someone who can repair your hard drive for you, would you like to speak to him?'. YES PLEASE I SAID :)

So he called him and put me on the phone with him, here's how the limited conversation went:

Me) Hard drive isn't detected by windows/device manager, just spins quietly. Have tried putting a new PCB into it with the same firmware but still didn't work. Considering swapping the firmware chip to another donor, what do you think?
Him) Firmware corruption is probably the cause. It's on the chip you're considering transferring. You'll be transferring the only thing wrong with your PCB onto a brand new PCB.
Me) Oh.
Him) Firmware repair cannot be done at home unless...
Me) Unless what? I'll try anything at this point.
Him) Firmware corruption can be fixed in some cases by using a special cable and some special commands through Hyper terminal.
Me) So I connect the cable, talk to the firmware and reset it? Where can I find out more?
Him) Search for the BSY firmware repair guide, but the results are not guaranteed.

So as soon as I got home, I looked it up and sure enough, a whole forum of people fixing firmware errors on their seagate drive.

https://www.overclock.net/t/457286/seagate-bricked-firmware-drive-fix-with-pics
Equipment: **broken link removed**

Apparently BSY means the drive has somehow got it self stuck in the BUSY mode, so nothing can access it. Seeing as the firmware is split between the firmware chip and a service area on the platter, it could be 50/50 as to which one is corrupted to cause this.

It seems like officially Seagate have not said much about this, but the problem seems to be identified and surrounding barracuda 7200.11 drives. Mine is not one of them, so I'm not sure if messing with something in this way will cause more damage.

Oblivion, you mentioned you had experience with firmware? Does any of this sound reasonable/plausible?

Any Ideas?

One thing I'm more confused about now is the idea of swapping EEPROM chips. I figured if I swapped this chip from one board with firmware (DE05) and put it into another board with (DE05), then all I'm really doing is making sure the new PCB board has what it needs to read the data off the platter correctly. But what if the chip contains both the DE05 firmware + the unique data the hard drive will need to read the data off the platter, all in one chip. I have always thought the firmware and this data were seperate and so by swapping the ROM chip into an identical PCB of the same firmware, I could bypass any firmware issues. But if they're both on the same chip, and that chip has issues, then moving that chip around is just moving those issues around?

Does swapping an EEPROM chip transfer a corrupt firmware to a new PCB? If this is the case then it sounds like I've only got two hopes, try and repair the firmware or somehow reset it (The other working board has the same firmware DE05, maybe I can transfer it somehow). If this miraculously works and the rest of the PCB is okay, then it might work. If the rest of the PCB is not okay, chips blown as previously suggested, then I'll have to transfer the ROM chip over to the new board, then fix the firmware. Arghh!

Finally, there seems to be a seagate firmware demo tool available here (**broken link removed**), it seems to be able to read/write firmware and repair the actual issue mentioned in the above thread. Maybe this, with the ebay cable might do something? But right now it's just a complete guess and compared to swapping out dead hard drive heads, this seems a nightmare:)

Time to give up? Try the ROM swap anyway? Try the Firmware fix?

Any opinions or ideas or suggestions welcome :)

Regards,
Megamox
 
Oblivion, you mentioned you had experience with firmware?

I have experience with 8-pin SPI/I2C EEPROMs. If that is what the IC on the bottom right of your circuit board picture is, then yes... I know how to deal with it. This is not all that special as just about half the rest of the people here on the ETO forums can do the same. EEPROMs are quite common parts. I have a drawer full of about 20~30 of them, and those are just the junk ones I have taken out of crap. Some of that junk actually being hard drives.

Does any of this sound reasonable/plausible?

What you mentioned sounds like a fairly specific bug that plagued some Seagate drives enough to generate internet buzz and embarrass Seagate a bit. However, we know very specifically that your drive physically failed internally, as the mechanics were quite obviously mangled. This does not entirely rule out the possibility of that bug, but it is much less likely to be the problem. When I see paw prints, I think dogs, not hyenas. This doesn't mean hyenas don't exist, just that in my area, dogs are far more likely.

That being said, That tool looks suspiciously like a good ole' generic FTDI FT232RL USB to serial IC break out. If that's true, you may not have to buy one if you have an older PC lying around that happen to have a serial port. You would need to make a minor buffer circuit, as a serial port outputs ±10V and the FTDI chip is only 5V-0V. The breakout would be far better. In any case, knowing that there is a possibility to get into the drive board through raw serial coms is a definite step in the right direction. If the board still works enough to be able to talk with a com port, then you might be able to grab your calibration data with out removing the chip. Same goes for replacing it into the good board.

General HDD rundown.
*VERY* generally speaking, the main SoC (big square chip, center stage) almost always has it's own internal ROM. This can be up to a MB or two, and it holds the main program for the whole drive. This program is usually a carbon copy across all boards of the same model, sometimes even stretching across a few very similar models. Now, depending on the design, the SoC will have either internal or external EEPROM for storing data that changes over time. This would be that special calibration data that is unique to each drive, and would be mostly useless for any other drive. If there is information stored on the disk that has anything to do with calibration and such, it is almost certainly slightly higher level stuff, like which sectors are bad, various offsets in disk rotation, partitioning, and crap like that. Although some things will not work without this information, the majority of the drive should still work fine. Though you might need to use a special tool/commands to get the drive to do a "raw read"of the platter to get all the bits off of it.

Calibration data
If there is calibration data that you must copy in order to get the drive to read data off the actual disks the normal way, (and there probably is) it will almost certainly be stored in the EEPROM, not the SoC program ROM. If the SoC does not have external EEPROM (If that is not what the chip on the bottom right really is) Then the EEPROM is internal to the SoC itself and may not be recoverable at all without some skill. Probably that break out and magical mystical manufacturer command witchcraft that I can't even explain or spell. Otherwise, you would have to physically get into the chip and try and read the EEPROM directly. This would involve strong chemicals, a very expensive microscope, several micropositioners, some free time, and a whole lot of knowledge about IC architecture though. Far far far to extreme(impossible) to consider for this project.

Board swapping.
If you swapped the board from the sacrificial drive onto the repaired drive and it did not work, then chances are it is because of the EEPROM data. I really can't see any information that is stored on the platters from preventing POST detection, no matter how special it may be. I'm fairly sure I have a few dozen drives around here that would happily be recognized in POST while just being bare boards. The only way this would not be true is if the manufacturer went out of their way to check for disk stored information, and intentionally abort the low level communications upon finding a flaw in it. This would be quite counterproductive as it would make it very hard to do any kind of diagnostic and repair on the drive. Though, This is not necessarily something that the manufacturers would have a problem with if they had a better way in, and they usually do.

In any case, that information is on the disks you are trying to recover, so we don't need to think about it. The objective is to put as much of the variables from the old drive together and in working order such that you can copy your critical information. That means EEPROM and the original disk stored information are coming for the repair ride.
 
Last edited:
Very interesting! Given that the original problem with the drive was mechanical, I suspect you're right. It's best to not look for phantom Hyenas just yet :) Besides at this point, I'm rather curious myself as to whether an EEPROM swap would magically fix it. In the scheme of the repair, having already replaced the head, I think the general consensus amongst the internet DIY community would be be to try the chip swap next. This at least would provide a good practical example for others on whether this process is worth pursuing. If that doesnt work, I dread to think of how much I'll have to learn about Serial Hard drive protocols in order to talk to this drive to see if I can get it to unstick itself from whatever firmware loop it might be stuck in. The hyper terminal commands look so esoteric. Maybe I can find a book on it in my local library. I can think of a few 4 letter words right off the bat, I'd like to serially communicate to it already though! :)

Chip swap next..!

Regards,
Megamox
 
Last edited:
Maybe I can find a book on it in my local library.
Hope your local library is better than mine. Shelf after shelf after shelf of politically-correct rubbish, and the whole of science squashed into about half a shelf :(.
 
Hope your local library is better than mine. Shelf after shelf after shelf of politically-correct rubbish, and the whole of science squashed into about half a shelf .

That's pretty sad, but my local library is really not much different. The good news at least is that if they don't have a book on shelf, I can get it within a few days if I simply fill out a request for it. Pretty much anything goes so long as it doesn't belong in a museum or in a obscure corner of the bathroom closet (porn). No fee.
 
Hi guys :) I'm afraid mine isn't very good, the engineering section is literally about 4-5 books. Strangely, there's an awful lot of books on lock-picking, perhaps somebody keeps requesting them. Sometimes I go to Amazon which lets you preview some books for free and so I can find things out that way. Sometimes I just email companies and university lecturers and some really kind ones, even email back :)

The solder gun is arriving soon according to Amazon, so if I dont kill the EEPROM chip, I should be able to post a result fairly quickly. Having never done this before I'm not sure what temperature to use. I have a rough estimate of 300-370 C - any ideas anyone? Dont want to burn this poor IC. Apparently I should start by swirling the gun above the area of the chip for about 10-20 seconds from a distance to heat the area a little. Then work on waving it back and forth over the legs and lift it off. Then clean up the area a little with wick. Flux it and apply solder paste to the pads, put the new IC on and repeat the process. If there's anything crucial I'm missing, do feel free to let me know! :)

Regards,
Megamox
 
I use a bump key or bolt cutters when I need to get past pesky locks..... allegedly. Though I am not a stranger to conventional lock pickery.


The solder gun is arriving soon according to Amazon

(0_o) What solder gun? Are you talking about the surface mount hot air soldering station? If so... having one of those doesn't make up for not knowing the technique you need to use to properly achieve victory. You can screw up just as bad using a hot air station as you can with a pencil style iron. Though admittedly, your chances improve dramatically when you have the right tool to be sure. Maybe it will fit you better, doing right it with a standard iron is kinda elite.

I personally am not familiar with the hot air irons, so take what I advise with a grain of salt. But as I understand you need to be sure to heat a large area of the board, heat it evenly, and making sure to bring the heat in and out some what slowly. The whole trick with them seems to be don't give the work piece any form of heat shock.
 
Thanks, sorry yes I mean the hot air soldering station. I bought the ATTEN AT 858D+ from amazon for £45 as I cannot find anyone locally who can remove it. I just looked up this 'Bump Key' - I'm shocked! I didn't realise my front door lock could be bypassed like this. I need to add some extra security!

In regards, to the technique, it will be my first time desoldering such a tiny chip so I will try and practise on one of the working boards I've got left over before I do it for real. Then again, everything I've done so far in this thread has been a first, so I suppose I should be used to it by now :) Just watching youtube videos like the one below, seems to really show how professional the results can be if you know what you're doing. If only my attempt goes so well!

https://www.youtube.com/watch?v=E3OmchO_mDc&list=UL

Best regards,
Megamox
 
Hello,

Well not necessarily. For example, JPG files use an elaborate compression stream, which if even one bit in the whole 250k file is flipped, will render the data from that point on almost worthless. Files like this often show up as 1/2 a picture or 1/2 a picture with the remaining half badly distorted.

My massive experience with hundreds of pictures rendered useless
 
Hi Oblivion!


If CRC errors are recoverable, that would certainly be useful, if anyone knows how to do this, or what software to recommend (Trial ones please!) I'd happily test it out and post results. The interesting thing is data which is marked as un-copyable by windows, is still accessible on the original test drive. It views just fine, but you just cant copy it to another place. Perhaps drive imaging techniques such as Ghost or Acronis may be a better way to preserve the overall data but it's been a really long time since I've used them. Any advice would be appreciated!

I also do appreciate all the points and corrections, they serve to keep the information and data much more balanced which I think is fantastic :)

Kind regards,
Megamox

Spinrite from Gibson Research has been around a very long time, and i have used it for data recovery. i have version 5 which requires the data to be on a windows partition. version 6 will work with linux and mac partitions as well. spinrite works by directly controlling the head amp gain, grabbing the data it finds, and comparing against the CRC, until it has what looks like valid data, then rewriting the data to a known good portion of the drive (if it determines the section it's reading from is unreliable). i've used it from about 1992 until the present. it's not perfect, as some data may be totally gone, but it does work quite well. https://www.grc.com/spinrite.htm (btw, i guess it's time for me to ger version 6, since i no longer use windows, but switched to linux several years ago...)

one company i worked for back in 1992 used spinrite to test every hard drive they replaced. many of the errors that were found on the drives were correctable by spinrite. back then, spinrite also could low level format drives and make them usable again, but more modern drives don't provide access to the low level format process anymore. some drive makers have software that approximates a low level format, that they make available, but it doesn't rewrite the timing and sector information as could be done with RLL, MFM and some early IDE drives
 
Ironically my copy of spinrite is actually on the dead drive :) I've never actually used it to recover data myself or fix a drive, but I have in my travels to fix the issue I've got now, heard from more than one independent source that it can be quite aggressive on the drive. This statement sounds like it needs qualifying though, as it certainly seems to work for many.

My hot air gun should be arriving at some point over the next day or two so hopefully I'll be able to swap out the EEPROM. I've also discussed the issue with some of the experts on the HDD Gurus forum and they've been kind enough to lend me their thoughts. They mention that in addition to platter damage and contamination from opening the drive outside of a clean room environment, it could also be that I've swapped in an incompatible head. Hopefully that isn't the case, but as I'm finding out bit by bit with hard drive repairs, don't rule out anything :)

Best regards,
Megamox
 
Hi Guys,

Hot air gun arrived today and I thought, finally - let's do this! I couldn't find anything to practise on unfortunately, so just grabbed the boards, cranked up the temp to 420 C and started my working on the chips. I have to say, this hot air station is really cool - the chips came off within 30 seconds. Glad I got it now. The SMD paste was a little fiddly, had to apply it to the pads with a sewing needle but once heated up, it disappeared nicely and doesn't require cleaning. Rom swap took about 3 minutes in total (Pics below).

So... once the EEPROM was transferred, what was the result?
SUCCESS :):):):):):):):):):):):):)

The drive popped up in Windows/Bios and I didn't waste time copying the data off! I've grabbed all the important stuff off so far and it's still copying! Feels so good to get my data back! I've recovered 200GB so far, with only a handful of CRC errors. At this point if it carries on or fails, it doesn't matter to me, I have everything I need now.

So right.. let's tally up how much this recovery has cost me:

2 X hard drives to practise techniques on from ebay = £30

£27 - USB/SATA external cable - this was really useful!
https://www.amazon.co.uk/Bipra-SATA...sr_1_sc_2?ie=UTF8&qid=1360185046&sr=8-2-spell

Hot air Station - £45 - to exchange the ROM chips
https://www.amazon.co.uk/858D-Rewor...HM28/ref=sr_1_1?ie=UTF8&qid=1360185146&sr=8-1

Epoxy anti static Tweezers - £11 - to handle components
https://www.amazon.co.uk/Rolson-591...ef=sr_1_1?s=diy&ie=UTF8&qid=1360185182&sr=1-1

SMD Paste - £3.60 - to Resolder the IC
**broken link removed**

So about £115 for a home DIY recovery. However as I've got the air station to keep now and the USB cable which is very useful, I'm gonna call it £50 even. Not bad and much less than any professional recovery quote I had received. I think Seagate wanted to charge me £1,000 + VAT.

It's taken lots of research, lots of support from everyone here at the Electro Tech forums, practising on a test drive, a head swap, an EEPROM swap but finally... I've got my data back... 2 weeks to the day from when I started the thread!

Relieved :)

Recommendations on what should be done with the drive once data has been copied off? Could attempt a platter removal or something else. If I lived in the USA, I'd be tempted to buy a gun and shoot it :)

Best regards,
Megamox
 

Attachments

  • SMD SETUP.jpg
    SMD SETUP.jpg
    95.7 KB · Views: 240
  • REMOVED.jpg
    REMOVED.jpg
    105.3 KB · Views: 237
Congratulations. I share your glee.

As for the old drive, you could list it on eBay as "vintage, was working , but can't test fully." ;)

John
 
Hah! Evil John.. evil :) The hard drive platters look like they'd actually make pretty good frisbees, but I think they're made out glass so that might be a bit dangerous. I'm not sure what the penalty is for hitting an innocent bystander in the head with one is, but it can't be good. Perhaps I could use them to make a home made reflecting telescope, they are polished really well and already have a hole through the middle.

Megamox
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top