EEVblog Electronics Community Forum

Electronics => Beginners => Topic started by: lordvader88 on August 07, 2018, 07:16:57 pm

Title: How often do discrete smd parts fail on hard drives ?
Post by: lordvader88 on August 07, 2018, 07:16:57 pm
I have non-working 3TB HDD, and I never probed anything digital like that before. I think it has a big jtag buss/pads. So far I found which chip is the motor controller, and ram, on the net, but I know this is all mission impossible. I should buy a replacement PCB

It has some corrorded pads, that I've read people re-tinned and got results, so I'll do that.

How often does a discrete part like resistor, cap, or inductor fail ? VS a chip goes bad ?

There's only 5 main chips, 2-3 voltage/current controller types, and I don't see any smd transistors, just R,C,L and a few diodes.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: james_s on August 07, 2018, 09:42:32 pm
I think it would be hard to say, in at least 99% of hard drive failures the drive is simply disposed of with no attempt made to perform a component level repair. I once repaired a hard drive that had failed because someone spilled a drink on the computer while it was on and one of the SO8 mosfets driving the spindle motor cratered. That's the only time I've done anything like that though.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: Gyro on August 07, 2018, 09:50:42 pm
Have you tried removing the PCB and checking the motor and head/actuator connections? On many drives these are pressure just contacts onto pads that can tarnish.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: Buriedcode on August 07, 2018, 09:52:42 pm
Not often.

Generally hard drives die slowly as the number of bad sectors can no longer be reallocated.  The pad corrosion thing would most likely cause a drive to be unusable by the host system, as it would hinder the connection between controller and heads.  I did the "clean/polish the pads" thing when my last hard drive bought the farm and it made no difference - doesn't mean to say it won't with yours, just that it must have been working with this corrosion before it went.

The most common board failure I've seen in my very limited experience is the TVS diode on the power inputs.  But I would say any "power" related devices, regulators or motor drive switches, the stuff that heats up would be the most likely failure.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: tooki on August 07, 2018, 10:20:21 pm
FYI, you cannot swap the PCB anyway. It stores a bunch of drive-specific stuff (like the map of remapped sectors) in EEPROM.

Buy a new drive and use that one as a doorstop.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: james_s on August 07, 2018, 10:53:19 pm
Occasionally you can get away with swapping the board, it's not an uncommon technique for data recovery. Back when 1GB drives were still respectably large I swapped a board and made one good drive out of two dead ones I was given, I used that drive for at least a couple years.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: Nusa on August 07, 2018, 11:48:48 pm
FYI, you cannot swap the PCB anyway. It stores a bunch of drive-specific stuff (like the map of remapped sectors) in EEPROM.

Buy a new drive and use that one as a doorstop.

If that's actually a showstopper, and I'm not convinced it is in most cases, then one can swap the EEPROM between boards. Granted, that takes a little more ability to do that than the mechanical board swap, but it's doable.

A few basic diagnostics on the original board are probably called for first, just to verify it isn't an easy fix like a shorted cap.

Whether it's worth attempting is another matter. It usually comes down to how important the data that nobody backed up is. New drives don't cost that much.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: tooki on August 08, 2018, 01:13:16 am
Occasionally you can get away with swapping the board, it's not an uncommon technique for data recovery.
Yes, for data recovery firms with big budgets and labs and expertise. Not for a hobbyist who has no clue what's wrong with the drive.

Back when 1GB drives were still respectably large I swapped a board and made one good drive out of two dead ones I was given, I used that drive for at least a couple years.
I'm not an expert in this, but I wanna say that I heard that in modern drives it works even less often, I forget why. It might have been something like calibration values or something.

If that's actually a showstopper, and I'm not convinced it is in most cases,
I think it's a showstopper for someone like the OP who has no clue where to start.

then one can swap the EEPROM between boards. Granted, that takes a little more ability to do that than the mechanical board swap, but it's doable.
Just a little! ;)

Whether it's worth attempting is another matter. It usually comes down to how important the data that nobody backed up is. New drives don't cost that much.
This was the crux of my argument. If your time is worth anything whatsoever, it's simply not worthwhile for a drive that's $75 new nowadays. Component level repairs on a hard disk make sense only in the context of data recovery.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: ttelectronic on August 08, 2018, 01:20:23 am
I would say a majority of drive failures, at least the non-ssd drives I've ever come across, it's usually mechanical.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: tooki on August 08, 2018, 01:28:31 am
Oh, for sure!!! I'd guess that in an HDD, electronics failure is probably orders of magnitude rarer than mechanical failure.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: BradC on August 08, 2018, 02:01:58 am
FYI, you cannot swap the PCB anyway. It stores a bunch of drive-specific stuff (like the map of remapped sectors) in EEPROM.

That's not entirely accurate.

There *are* drive specific adaptives that are stored in eeprom/flash on most of the new drives. Sometimes they are in an external chip and sometimes they are in the main uC. Some of the major aftermarket drive service tools (we're talking >$10k tools) are capable of re-building those to some degree. Some can copy the adaptives between boards. On boards with an discrete external memory, that can be swapped over.

The adaptives hold enough information on the mechanical charactersitics and tolerances to allow the drive to load its second stage firmware from the service area (SA) on the platters. It is also in the SA that things like SMART attributes,  defect tables and additional adaptives are stored.

Quite often the repeated "click, click, click" signaling drive death is either the drive unable to read the SA due to electronic or mechanical failure, or the SA having been corrupted. In the latter case the right tools can often re-build the SA and get access to the data. In the former then mechanical repair or board swaps are generally required.

In some cases and with a bit of dumb luck, a straight board swap might work. If the boards have a discrete eeprom, and are similar enough in hardware/firmware then a chip swap will do the trick.

As always, if the data is valuable, talk to a professional.
Title: Re: How often do discrete smd parts fail on hard drives ?
Post by: tooki on August 08, 2018, 02:59:06 pm
Thanks for the info!!

Sounds like it's worth it for data recovery, but beyond silly as a repair.