Info on HDD standard operation, repairs, anything ?

[MathWizard]

I have a 3TB HDD that out of the blue started making whirring sounds like it was stuck in some loop. I heard it and shut it down within a minute. I was playing a movie in the background at the time, and closing the video-player didn't help, had to reboot.

But in fact I took out the drive right away, and never checked it in the PC again, I just checked a few things with a DMM and it all seems fine enough.

But were can I find some useful info, on HDD's, in general ? I don't expect schematics, but just something better than some computer tech writing a marketing article on HDD's (that's about all I find on google).

Once I know how all the parts work, like stepper motor's , I'd have more idea what to look for, but I'm not there yet.

The main motor controller and DC-DC converter controller is some costum chip, so I have no hopes of finding anything on this stuff. But some of this stuff must be out there.

In the meantime since there's no deadshorts or anything toasted, I should just try it again in the PC.

[SilverSolder]

I just spent some entertaining hours taking old hard disks apart - they are amazingly well made mechanical devices and fun to play with.

The motor that spins the discs is a synchronous motor - it needs a 3 phase supply, generated by one of the custom chips on the PCB. 

The head motor is a complete custom design, not a stepper motor.  It is basically a flat coil that moves between two closely spaced magnets, not unlike how a loudspeaker voice coil moves.  The electronics on the PCB uses feedback from the recording on the discs to position the heads accurately.

The whole thing is built and sealed up in a dust proof factory environment.  I think that it would be difficult to open one of these things up in a home environment and expect it to remain clean and reliable after that.

[MathWizard]

I have another truly broken 3TB HDD somewhere that I kept. I really want to find that 1 now and hook it up on the bench PSU and scope.

That one has a real problem, but nothing looks burnt, or reads funny enough for me to have noticed 2-3 yrs ago when looked at it, but I learned a lot since then.

Seems there's 4 big inductors in DC-DC converter's on the main PCB. I haven't tried to map it yet, but again, now I should just put the PCB back on and try it again. It's not like windows would never have fallen partly alseep or something, with a movie playing while I was playing BF1.

[SilverSolder]

You always learn a lot from taking broken/obsolete stuff apart!

[MathWizard]

I forgot how corrorded all the underside PCB pads/and GND pads were, so I just cleaned them with the fibre-glass brush, but no luck.

I pluged it back in and booted, same noise, were- witicha- tow, x3-4 then wuuurreerrrreehhhh, and repeat. For the musically inclined the pitich of the were-witicha-tow is like a ramp up then down, I can't guess the pitch of the top of my head even compared to music I know, maybe 10Khz peak (so is that the platter motor , no but maybe 7200rpm motor on startup I think) but I should actually make a circuit and record it........but not today.......and the werrrreeeerhhh being slower/lower/longer, around 2-3s long, low low 100's of Hz .

And I shut down rather than log in to win10. So maybe that's the head seek motor and/or sequence failing to find whatever, and looping, or the control for it looping due to some failure ? It sound's vaguly familair, but that's a long time ago.

So now the fun begins, I'll have solder on a bunch of test leads, and put it on the PSU/DMM/scope

[TheMG]

Kind of hard to know the sound you're describing without hearing it with my own ears, a recording is worth a thousand words.

Anyways, if the platters appear to be spinning, the noises are likely from the head assembly, trying to seek to a particular area of the platters but unable to find it or read the data that's there. Head positioning information on modern hard drives is on the platters themselves.

Most of the time I don't think the failures are due to anything on the PCB. Usually it's one or more of the heads or some problem on the platter surface itself.

If the data on the drive is important, send it to a professional data recovery service. If not, and the drive is out of warranty, toss it (or take it apart and tinker around with it as you like). But what I will say is it is extremely unlikely you would be able to repair it. As soon as you open the drive up in a "dirty" environment you're already doing more harm than good, and swapping parts like heads requires specialized tools and skills to do properly. Data recovery if there is damaged tracks on the platter (note that this may not even be visible to the naked eye or even under a microscope) requires specialized software and knowing how to use it.

They are very very precise and extremely delicate mechanical devices.

[Renate]

They are very very precise and extremely delicate mechanical devices.

That's not really true.
The Diablo 5 MB disks had a track density of 100 tracks per inch.
Give me a pair of pliers and I'll do a seek.

[Mechatrommer]

you can search for threads about HDD damage recently there are few... esp by "classicsamus87 the unbelievably bot like" member created where free_electron replied, there you can read and enjoy understanding the details what happened inside the spinning disc. i believe youtube also has plenty of it. to summarize, any kind of singing sound you hear, esp when it comes with delayed/halted reading on your OS, is very not a good sign. the PCB you cleaned underneath, even if its still clean in the first place is very likely not has anything to do with the problem. i just hope you are not keep asking and digging about what the name technology behind the SMART before 2000

[helius]

They are very very precise and extremely delicate mechanical devices.

That's not really true.
The Diablo 5 MB disks had a track density of 100 tracks per inch.
Give me a pair of pliers and I'll do a seek.

Bending the head mounting arms irrecoverably in the process?
A Diablo or DEC RL01 weighs over 50 times as much as the drive the OP is talking about. It is still sensitive to small mechanical problems and cannot be repaired without a special alignment pack. So TheMG is correct about "specialized tools and skills" even for your example, despite it being an anachronism by ~50 years.

[fzabkar]

The motor controller IC is a multifunction device. It controls the spin motor, voice coil motor and the external power supplies. There is a public datasheet for the L7250 SMOOTH controller. That should give you a general understanding of how these chips work. Current chips appear to be custom designs for a particular manufacturer and model.

Your drive sounds like it has a head or media fault. When the drive POSTs, it unpacks the ROM on the PCB, performs a few tests, and then fetches the bulk of the firmware from a reserved System Area (SA) on the platters. If the drive cannot reach the SA, or if one of the firmware modules is damaged (eg SMART, defect lists, translator), then it doesn't come ready.

The hard drive -- a computer-within-a-computer;
http://www.hddoracle.com/viewtopic.php?f=56&t=2600

[MK14]

The link you just posted, have triggered warnings (from different sources), that they are BAD/MALWARE etc.

The site ahead contains harmful programs
Attackers on www.hddoracle.com might attempt to trick you into installing programs that harm your browsing experience (for example, by changing your homepage or showing extra ads on sites you visit). Learn more

Additionally:
https://www.virustotal.com/ confirms it as a dangerous link as well.

DETECTION
DETAILS
COMMUNITY
CyRadar

Malicious
Fortinet

Malware
G-Data

Malware
Google Safebrowsing

Malicious

[rob77]

i'm afraid you have a zero chance to repair the drive.. if you need the data then you'll need to send it to someone for data recovery, but expect a steep price.

heads are driven by a voice coil (exactly like in a loudspeaker) so the movement is analog - the hard drive has a servo loop using the head readout to follow tracks (it's not a stepper for decades).
and actually if the head stack is broken there is no readout and the drive is not able to find the tracks on the platters - that's most probably your case. in a data recovery shop they can transplant a head stack from a working donor drive and save your data (in a clean room/ clean hood using special tools to keep the heads apart and not damage them)

if the electronics is broken, then it can be transplanted from a working donor drive pretty easily but you need special tools/software to read the service area of the platters - without that step the new disk electronics will not recognize the data. so again it needs to be someone doing data recovery.

if you want just play with the broken drive, then the platters are driven by a BLDC motor, you can use any ESC for driving that BLDC (e.g. ESCs used in model aircraft), just hook the motor to such an ESC and a servo tester and you can play with the spinning platters (but it might be a bit dangerous - fast spinning discs are not safe to touch )

[fzabkar]

The HDD Oracle site hosts tools which deal with HDDs at the firmware level. These result in false positives.

The URL I provided links to my own text based tutorial. There are two attachments, both of them ROM dumps. The rest is only JPEG images and various trustworthy reference URLs.

If Kaspersky ever complains about the site, then I'll listen. Otherwise I'll treat these alerts with suspicion.

Edit:

One of the links is to a tool which I wrote and uploaded to my own web space:

http://www.users.on.net/~fzabkar/temp/wdROMv12.exe

It triggers two false positives:

https://www.virustotal.com/gui/url/1a523a115bfc39bf64679ea6cf4bece526222f79818d5e669ecf2918606b9b4e/detection

This is the FreeBasic source code (plain text) which also triggers the same two AV software:

http://www.users.on.net/~fzabkar/temp/wdROMv12.bas

https://www.virustotal.com/gui/url/eccefa8da737c603cd027f37df2a086cfbdf62644daa3b888ab21ae7c86ebc2d/detection

BTW, my tool does not access the hard drive, it only analyses a file containing a ROM dump.

[rob77]

The HDD Oracle site hosts tools which deal with HDDs at the firmware level. These result in false positives.

The URL I provided links to my own text based tutorial. There are two attachments, both of them ROM dumps. The rest is only JPEG images and various trustworthy reference URLs.

yep , nothing wrong with that link.. actually a very nice article 

[MK14]

The HDD Oracle site hosts tools which deal with HDDs at the firmware level. These result in false positives.

The URL I provided links to my own text based tutorial. There are two attachments, both of them ROM dumps. The rest is only JPEG images and various trustworthy reference URLs.

If Kaspersky ever complains about the site, then I'll listen. Otherwise I'll treat these alerts with suspicion.

Edit:

One of the links is to a tool which I wrote and uploaded to my own web space:

http://www.users.on.net/~fzabkar/temp/wdROMv12.exe

It triggers two false positives:

https://www.virustotal.com/gui/url/1a523a115bfc39bf64679ea6cf4bece526222f79818d5e669ecf2918606b9b4e/detection

This is the FreeBasic source code (plain text) which also triggers the same two AV software:

http://www.users.on.net/~fzabkar/temp/wdROMv12.bas

https://www.virustotal.com/gui/url/eccefa8da737c603cd027f37df2a086cfbdf62644daa3b888ab21ae7c86ebc2d/detection

BTW, my tool does not access the hard drive, it only analyses a file containing a ROM dump.

Thanks for taking the time to analyse what had happened.

I'm surprised, even basic source code language text in a file, can trigger an alert.

I was relying on the quality/accuracy of the reports that seemed to be triggered via the link, and was trying to warn people, in case there was a risk.

Probably better to be a bit too sensitive (virus checking programs), than risk missing new virus's and things.

It's crazy the level of complexity and work that goes into hard disks. They are a real marvel of technology (although becoming gradually obsolete, because of SSDs).
A long time ago, Dave/EEVblog, did a tear down of a huge ancient hard disk, which was of enormous dimensions (compared to modern drives). Here it is:

[HobGoblyn]

I’m sure a few others on here have used/seen these, when I was a computer operator in 1985 we had a few disk drives that were the size of a washing machine. They were something like 330 MB.  Once every few weeks/months a lady in a white coat appeared and remove the platters to clean them.

They were used on a Gould Sel computer.  It had 6 of those huge vacuum reel to reel tape drives and also had an inbuilt basic chess game.  If you played the game, all the tapes would stop while it decided what it’s move was going to be

[fzabkar]

I used to work on those 300MB Control Data monsters. One of them is now my tool trolley. I got AU$70 just for the scrap aluminium.

I did my first data recovery from one of those:

http://www.hddoracle.com/viewtopic.php?f=117&t=1203

[amyk]

The data recovery industry is pretty cartel-like. I wouldn't be surprised if they got AVs and such to silence such detailed information about HDD operation, just like cracks and keygens get blacklisted by AVs.

[SilverSolder]

[...] I'm surprised, even basic source code language text in a file, can trigger an alert. [...]

This is because e.g. Basic code can be run directly by Windows machines, in the right circumstances.

The problem is similar to a batch file, which is also just text, but which can cause a lot of damage.

[MK14]

This is because e.g. Basic code can be run directly by Windows machines, in the right circumstances.

The problem is similar to a batch file, which is also just text, but which can cause a lot of damage.

That makes sense. I hadn't thought of it like that. Now you mention it, I think some Microsoft stuff, can execute/interpret basic scripts, as if they were batch files (approximate recollection of mine, so exact details may be different).

Some of the latest internet security advice is mind-boggling, on the face of it. Lack of password security, can actually lead to you being killed as a result.
Basically, people loose or don't change the password on home video cameras (or however the hackers manage to get 'into' the cameras), then hackers watch what is going on in the home and make fake police calls, so that swat teams are seen making a raid on the innocent peoples homes (hackers victims), which risks the householders being incorrectly shot.

Link Here:
https://www.bbc.co.uk/news/technology-55499164

[MathWizard]

There's nothing that really matters on drive, I'm just really really hoping some discrete part has failed. Years ago I saw a some 2 terminal SMD fry on my HDD with a wisp of smoke. I wonder if that one would be easy for me to fix now.

Just how bad is household dust for a drive ? Is it just that you end up with a layer of dust and debris on the platter ? I've opened drives up before, but never a working 1 (as a HDD).


I have a Bus Pirate, I'm just creeping into programming, I never really played with that yet, surely there's something cool I can do with that on this.

[MK14]

Just how bad is household dust for a drive ? Is it just that you end up with a layer of dust and debris on the platter ?

Unlike floppy disc drives. Where the read/write head(s), are in physical contact with the disc surface.

Hard disks, rely on being able to have the head(s), extremely close to the disc surface, but not quite touch it. I.e. an almost microscopically small distance away (bernoulli effect).
Hence the very long lifes of tens of thousands of hours, you can get, without wearing out the disc surface.

But even a tiny dust particle, is big enough to cause a head crash and/or friction/damage, because it can fill the tiny gap, between the head(s) and the delicate disc surface.

tl;dr
As long as you don't intend to get any important data from it, or intend to repair it (and use it again). Then getting dust into it is fine.
But if you do (get dust in it), it probably won't last very long.

[fzabkar]

There's nothing that really matters on drive, I'm just really really hoping some discrete part has failed. Years ago I saw a some 2 terminal SMD fry on my HDD with a wisp of smoke. I wonder if that one would be easy for me to fix now.

Just how bad is household dust for a drive ? Is it just that you end up with a layer of dust and debris on the platter ? I've opened drives up before, but never a working 1 (as a HDD).


I have a Bus Pirate, I'm just creeping into programming, I never really played with that yet, surely there's something cool I can do with that on this.

HDDs have TVS diodes at the +5V and +12V inputs. An overvoltage will usually short circuit them. There may or may not be a fuse or zero-ohm resistor in series.

TVS Diode FAQ:
http://www.hddoracle.com/viewtopic.php?f=100&t=86

Dust is serious. The fly height of a disc head is about 3nm. That's the average distance between two molecules of air.

If you want to pursue this task, then at least tell us the model number of the drive. Different models have different solutions and different software tools. The problem with your 3TB drive is most likely internal, not PCB related.

I have written several tools to parse the "ROM", if you manage to dump it. It won't help you to repair your drive, though.

[amyk]

free_electron has written some very interesting (and surprising) commentary about HDDs, including head crashes, here: https://www.eevblog.com/forum/beginners/what-to-do-with-a-hdd-motor/msg129452/#msg129452

[MK14]

free_electron has written some very interesting (and surprising) commentary about HDDs, including head crashes, here: https://www.eevblog.com/forum/beginners/what-to-do-with-a-hdd-motor/msg129452/#msg129452

Thanks, that was very interesting. When I wrote the post above, I was thinking to myself, you don't seem to hear about hdd disk crashes these days.
But, dust/contamination, may also (if I understand the situation correctly), somewhat rapidly, wear out the tiny/delicate head(s), as well. So, even if the dust doesn't cause damaging head crashes because of modern, design improvements, I think it would still, seriously shorten the head(s) life (but I'm NOT sure).

I think my original post, came about, because in the old days, of mainframe computers, actual destructive head crashes, was a thing. So it seems (unknown to me, and probably others), modern hard disks, have largely/fully solved that problem.
Except where the drive accidentally gets a hard knock, while on/spinning, and any automatic protection against that (mems accelerometers, I suspect/know), hasn't had time to work, or is not present.