Author Topic: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN  (Read 1179 times)

0 Members and 1 Guest are viewing this topic.

Offline EEVblog

  • Administrator
  • *****
  • Posts: 32916
  • Country: au
    • EEVblog
EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« on: June 08, 2021, 01:41:49 pm »
Dave's Western Digital RED NAS hard drive failed. Bugger the warranty, teardown time!



Failure video:
 
The following users thanked this post: WN1X, PKTKS

Offline gjsmo

  • Contributor
  • Posts: 11
  • Country: us
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #1 on: June 08, 2021, 03:23:14 pm »
@EEVBlog The head assy is undoubtedly machined out of a single piece of metal. We regularly make parts more complex than this at my workplace. There's no good reason I can see to cast it - the part is fairly small and that level of precision requires post-machining operations regardless. That doesn't necessarily mean that it's done using conventional machining, I can certainly see a possible use for wire EDM here.

The housing on the other hand is undoubtedly cast, with certain areas post-machined (motor spindle mount, head mount etc). As it's larger and doesn't require much precision except in a few areas, that would seem like the most likely candidate. The soft corners are a giveaway too.
 

Offline Rdx

  • Contributor
  • Posts: 9
  • Country: de
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #2 on: June 08, 2021, 03:29:42 pm »
Wire EDM seems way too expensive for such a component (and this quantity).

Where do the 4 parallel connections of the main motor flex bugger off to?

Also, the head coil, how does that work? I can not exactly wrap my head around the magnetics going on between that weirdly shaped coil and the two magnets tbh.
 

Offline eti

  • Frequent Contributor
  • **
  • Posts: 463
  • Country: gb
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #3 on: June 08, 2021, 04:43:26 pm »
Dave, the “dessicant bag” is not that. It is, in fact, a special pad to collect stray oxide particles which are shed and flung off the discs, hence its height and position.
 

Offline gjsmo

  • Contributor
  • Posts: 11
  • Country: us
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #4 on: June 08, 2021, 05:31:09 pm »
@Rdx It depends entirely on the precision required. I would assume the moving assembly needs to be fairly well balanced, but as I don't work in the hard drive industry I couldn't tell you how well.
 

Offline Rdx

  • Contributor
  • Posts: 9
  • Country: de
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #5 on: June 08, 2021, 05:57:30 pm »
The quantity manufactured and the price for those hard drives already tell you that wire EDM is not being used.
Also if you look at the shape of the manufactured part you see some internal convex rounding along the center axis of the arm. This would be extra complicated to make using EDM, so it makes it really even more unlikely. The tolerances you can achieve with milling nowadays is quite impressive.

Also mentioning balancing, I would guess that the small holes in the central platter holding flange which Dave interpreted to possible be for assembly of the screws are more likely for balacing of the rotor assembly i would guess. Not quite sure though.
 

Offline egonotto

  • Frequent Contributor
  • **
  • Posts: 341
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #6 on: June 08, 2021, 06:44:43 pm »
Hello,

in german headcrash is "Spanabhebende Datenverarbeitung" (metal-cutting data handling)

Best regards
egonotto
« Last Edit: June 08, 2021, 06:50:03 pm by egonotto »
 
The following users thanked this post: thm_w

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 9006
  • Country: de
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #7 on: June 08, 2021, 07:14:37 pm »
How the Coil for the head drive is working is a really good question. With just a simple magnet, there would be no force, once the coil is fully inside the magnte.  So I would expect the magnet to be actually 2 sides of opposide polarization. So the 2 halfs of the coil would see opposing external fields, so that the forces would add up. So it makes sense to have the coild so large to cover halve the movement range.

EDM is nice for special parts made of hard material, but it is pain-stakingly slow. A bit like a snail in slow motion. The arms look easy to maching and are from a light relatively easy to machine material.

It may have been a different defect and not a head crash. A somehow broken head or amplifier would be a possible reason.
 

Offline thm_w

  • Super Contributor
  • ***
  • Posts: 2622
  • Country: ca
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #8 on: June 08, 2021, 09:34:45 pm »
Here is one I found in a dead server PC.
The lines/grey areas are where the head crashed and tore the hell out of the disks. You'd think it would detect friction and shut down, nope.

 

Offline CChin254

  • Regular Contributor
  • *
  • Posts: 51
  • Country: us
    • My Blog
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #9 on: June 08, 2021, 09:37:22 pm »
IC Identification/Block Diagram:
[attachimg=1]
[attach=2]
 
The following users thanked this post: EEVblog, thm_w

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 6747
  • Country: gb
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #10 on: June 08, 2021, 10:00:36 pm »
From microscope examination of similar head suspensions, I think those additional "test pads" at 20:30 are actually piezo actuators, used for fine tracking.
Chris

"Victor Meldrew, the Crimson Avenger!"
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 14942
  • Country: us
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #11 on: June 08, 2021, 10:11:17 pm »
Here is one I found in a dead server PC.
The lines/grey areas are where the head crashed and tore the hell out of the disks. You'd think it would detect friction and shut down, nope.

Not much reason to bother really, by the time there is excessive friction, the drive is trash. Most of the time a server drive is going to be part of a RAID array and hopefully has backups so it should be rare for there to be any reason to try to limit the damage once a disk is destroyed by a crash.
 
The following users thanked this post: thm_w

Offline EEVblog

  • Administrator
  • *****
  • Posts: 32916
  • Country: au
    • EEVblog
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #12 on: June 08, 2021, 11:07:02 pm »
How the Coil for the head drive is working is a really good question. With just a simple magnet, there would be no force, once the coil is fully inside the magnte.  So I would expect the magnet to be actually 2 sides of opposide polarization. So the 2 halfs of the coil would see opposing external fields, so that the forces would add up. So it makes sense to have the coild so large to cover halve the movement range.

Nope, just a single coil.
It already has a sensor, the read head itself, so just use encoded tracks on the platter.
 

Offline 3roomlab

  • Frequent Contributor
  • **
  • Posts: 816
  • Country: 00
  • Did you buy into a new audio scam call MQA?
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #13 on: June 09, 2021, 01:28:04 am »
for those who are unaware, there are these guys
https://www.backblaze.com/blog/
winxp same filenames overide press "yes to all" one button. win7 same filenames overide press 2 times apply to all and copy. why do they make modern software stupid? and bloated?
 
The following users thanked this post: EEVblog, SeanB

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 6747
  • Country: gb
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #14 on: June 09, 2021, 08:41:48 am »
From microscope examination of similar head suspensions, I think those additional "test pads" at 20:30 are actually piezo actuators, used for fine tracking.

Edit: If you look carefully in the video, you can see that one surface of each of the piezo elements is grounded to the head arm, whilst the opposite faces are connected to the outer traces of the flex-print which branch off and curve back.

Some more anorak information:

The "high friction" of the spindle are down to a couple of factors. Firstly, the spindle motor assembly magnets and stator are still inside the hub, and secondly, the spindle uses Fluid Dynamic Bearings (the days when the surface vibration and runout from ball races were acceptable are long gone). You're flicking them below operating speed, and half the time, spinning them backwards (which causes metal to metal contact). The construction and surface milling of FDBs are an interesting subject in their own right... https://www.nidec.com/en/technology/capability/fdb/

The head stack ball race bearings are about the closest tolerance bearings that you can sensibly find. You can salvage them by carefully cutting down the thinnest section of the head stack assembly with a hacksaw (along the bearing axis and using tape to avoid any microscopic swarf getting past the bearing shields) and then 'spreading' the hole with a vise (or screwdriver in the slot). This allows the whole pre-loaded bearing assembly to come out as a single component. Don't try pressing or drifting it out.

Even with the ultra-smooth head stack bearings, with such high track densities it is impossible to achieve individual track resolution in a sensible time - hence the piezo head steering. It's also much faster to fine track the low mass head tip than the moving mass of the whole head stack and voice coil assembly.

The heads use Giant Magneto Resistivity (GMR) for read. The heads are effectively multi-layer dies (which obviously span many tracks) and have the aerodynamic properties to 'float the heads by an infinitesimal amount. I believe the write coils are implemented as multi layer traces on the die. https://www.pctechguide.com/hard-disks/hard-disk-gmr-technology

It would be great to get @Noopy to make one of his excellent micro-photographs of the head surface.
« Last Edit: June 09, 2021, 10:24:54 am by Gyro »
Chris

"Victor Meldrew, the Crimson Avenger!"
 
The following users thanked this post: ve7xen, thm_w

Offline Neilm

  • Super Contributor
  • ***
  • Posts: 1493
  • Country: gb
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #15 on: June 09, 2021, 06:44:07 pm »
I am glad that Dave has checked his drive is mirrored...

I had a disk "crash" some years ago - there was a little bit of noise one day then an almight BANG. I think the platter dirve motor had seized.
Two things are infinite: the universe and human stupidity; and I'm not sure about the the universe. - Albert Einstein
Tesla referral code https://ts.la/neil53539
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 7647
  • Country: us
    • SiliconValleyGarage
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #16 on: June 15, 2021, 02:38:49 pm »
some information ( harddisks was my playground for 12 years prior to me saying farewell to the semiconductor ratrace)
That baggie is not dessicant but a dust filter. if you look carefully at the construction of the cavity and where that sits : there is a channel coming off the side that blows air molecules through the filter and back on the platter.
the spinning discs create quite a bit of airflow and this is carefully guided towards the baggie.

the disc cover is actually a mulitlayer sandwich of materials mainly designed to suppress acoustics. ( the noise of the drive ). you can peel that apart. the placement of the dimples is on purpose.

The platters are so shiny because 1) they are diamond coated , 2) they are lubricated with a special polymer. This is of course at near atomic level. the recording surface sits buried quite 'deep' in the platter. the core of a platter is either a borosilicate glass plate or aluminum. Then there are various metallic layers followed by the magnet layer. this is not a contiuous film but rather looks like salt grains. it is very rough (microscopically speaking) . you need about 50 to 100 grains to store 1 bit. The magnetic field is trapped in a grain with little boundary action. ( the grains don't attract or repel each other as the field is trapped in the sphere. you simply need multiple ones to have a strong enough response when reading. Contrary to what you would think, the head does not sense the magnetic field directly. The magnetic field does not move electrons in the pickup head, it only changes the spin of the electrons. The rest is quantum mechanics. if i remember right it works this way : A neutral material has both spins in a near 50/50 ratio. The head has multiple layers , one of which sets the spin of the electrons traveling to the pole tip. the pole tip is influenced by the spin in the magnecules ( clusters of magnetic grains and the polarisation directed towards the pole tip. ). this can flip the spin direction ( or leave it) now the electron travels from the pole tip to another layer with the same spin as before. the net effect is that, if the spin is unchanged the electron moves forward , if the spin has changed it scatters. basically the current changes , which in term induces a voltage.
the read head is a 4 wire element. the horizontal wires carry a current that creates the spinning electrons. the vertical wires detect the passing and deflected electrons ( differential sensing ). the signals are in the GHz range.

These amplifiers are not a joke they have bandwidths from a few kilohertz to 6 Ghz and beyond with near flat gain in the range of 40 to 60 dB (programmable). The filter is also programmable.
The write driver is even more nuts as it sends hundreds of milliampere through the write coil with very fast slew rates. bandwidth here is again in the 6 GHz range. The write current is 'shaped', it is not just on/off or a square wave. the controller sends a logic level and the write driver creates the correct pulse shape.

Flying height is controlled by a heater on the pole tip. air pressure between the head and platter controls the height (bernouilli effect) , but since pressure is temperature influenced we have to take care what happens if the head warms up ( during writing for example ). so a heater can mitigate that. the heater is again programmable.

the chip on the headstack ( visible in the video ) contains the read and write amplifiers, the heater controller, the bias control for the GMR element and the drivers for the piezeo actuators. in case of HAMR it also contains the laser driver , an now in case of microwave assisted writing it also contains the magnetron driver. as well as the voltage boosters. These chips have two communication channels.
one channel is the head data : 1 diff pair for read , 1 diff pair for write. The second channel is the control data for all the drivers (piezeo,laser,microwave, bias generators , filter settings , gain settings of the amplifier etc )
Your average device contains hundreds of registers , mostly 16 bits or 32 bits long that control all of this. now, these registers need to change fro time to time,  and this can be needed so fast that there is no time to transport the new register set. ( the bus is a 3 wire spi like datastream clocked at hundreds of MHz ) so the chip has multiple register banks that can hold presets. with a short command the entire banks can swap. there are also commands to increment/decrement certain registers. these commands are very short , only a few bits. so as opposed to transmitting a full 32 bit register they can just transmit 4 bits for example . The first bit in the packet tells you if it is a register content or a command. registers are always the same length. commands vary. the most needed commands are the shortest. the commands are a tree structure. This protocol is proprietary and the register allocation and layout is chip dependent. you can't swap one for another. They are all different.

the piezo elements on the arm allow for torsion. this keeps the head perpendicular to the track. it's the old needle-in the groove problem first encountered in record players. as the arm pivots around a fixed point the angle of the head to the track is not a constant. the middle track is the narrowest as there the head is perfectly tangential with the rocrding track. as the head moves inward or outward its angle changes making the track wider. the piezo elements can create torsion in the arm to make sure the head remains in line irrespective of the arm position.

the head mechanics themselves is another thing. it contains carefully crafted channels and righes to control the aerodynamics of this thing flying.

The voice coil motor works as follows : we create a current int he coil . which causes the coil to move in the magnetic field. the amplitude controls the speed of movement.
if the heads are on track there is no current in the coil. if we need to move up we send current one direction , or the other direction if we need to move down. A servo loop compensates for drift. the read processor looks at the amplitude of the received bitstream and can commandeer the motor controller to alter current polarity and change amplitude to position the heads.

There's much much more going on in the drive. i have posted many times ont he forum details on how all this stuff works. just search for it.

the heads are biassed
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 
The following users thanked this post: lowimpedance, amyk, SeanB, thm_w, BradC, BrokenYugo

Offline andy3055

  • Frequent Contributor
  • **
  • Posts: 693
  • Country: us
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #17 on: June 15, 2021, 03:47:47 pm »
I hate WD. Totally unreliable. Always had WD drives fail, as against other brands. If you want reliability, replace them with at least Seagate. IBM drives would run for 15 years non stop with no fails! Good things no cheap. Cheap things no good.
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 14942
  • Country: us
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #18 on: June 15, 2021, 04:16:27 pm »
I hate WD. Totally unreliable. Always had WD drives fail, as against other brands. If you want reliability, replace them with at least Seagate. IBM drives would run for 15 years non stop with no fails! Good things no cheap. Cheap things no good.

I've had excellent reliability from WD drives, less so from Seagate. You always hear things like this from people swearing by one brand and insisting another brand is junk but the reality is that every brand has produced a few turds. The worst I ever dealt with was Maxtor but they're long gone. Back in the day I saw a lot of bad 1.2-1.6GB Maxtor drives. Had a lot of Seagate 2GB drives fail too but they had excellent warranty service, several times someone gave me a dead drive and I sent it in under warranty and got a refurb in return.
 
The following users thanked this post: SeanB

Online amyk

  • Super Contributor
  • ***
  • Posts: 7393
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #19 on: June 16, 2021, 12:23:04 am »
Seagate has had some really bad models, however...

https://en.wikipedia.org/wiki/ST3000DM001
https://www.backblaze.com/blog/3tb-hard-drive-failure/
https://en.wikipedia.org/wiki/Seagate_Barracuda (look at sections(!) titled "Firmware bugs")

...and in all the reliability data I've seen, they consistently rank near the bottom. Even the highest AFR in the chart posted above is from a Seagate. Also, look at the data recovery forums - Seagate outnumbers WD and the rest in terms of number of posts mentioning them.
 

Offline Johnny B Good

  • Frequent Contributor
  • **
  • Posts: 506
  • Country: gb
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #20 on: June 19, 2021, 03:24:08 am »
 Just came across this HDD teardown video. I noticed your disappointment at failing to see evidence of a head crash so thought I'd offer this possible explanation.

 The magnetic formulations moved away from the "rust" based coatings over a decade ago and are, in fact, extremely hard and durable. However, they're not immune to the effect of high temperature causing loss of magnetic retention (Curie temperature) so if something causes a head crash (an occupational hazard as far as laptop HDDs are concerned), the resulting friction can magnetically damage the data and embedded servo information leading to the symptoms you observed without necessarily leaving any obvious witness marks.

 As others had said regarding the supposed "silica gel packet", that's simply a 'catch filter', a feature that's been present in all small form factor (5 1/4 inch) drives from the 20MB Seagates and up, over the past 3 1/2 decades or so (some features just don't change).

 Several years ago, I came across a laptop with an HDD suffering from a rash of bad unrecoverable sectors about 20GB into the 60GB's worth of disk space. I was able to re-partition it into three partitions (a 19GB primary drive C, a 1GB logical drive labelled BAD SECTORS and a logical drive D using the last 40GB of good disk space. Testing had revealed that only a well defined region of a few megabyte's worth had turned bad, hence the above work around to bypass them.

 My tests had suggested that there'd been a head crash which had induced "friction burns" rather than gross physical damage and a release of abrasive particles to wreak further havoc (built in catch filter notwithstanding) so felt pretty confident that bar any angry lid slamming or accidental high G events (like a free fall onto a hard surface), there'd be no further increase in the bad sector count. I did advise my customer to consider using the opportunity to purchase a replacement capacity upgraded HDD ASAP and come back to me whenever it was convenient to do so (afaicr, he'd been in a hurry to get the laptop working again). I don't recall him making a return visit so I guess my 'repair' must have proved sufficient for his needs.

 Mind you, whilst a laptop drive would face such high G head crash inducing events as an occupational hazard, the risk of a similar event  with a 3 1/2 inch drive safely housed within a NAS box seems vanishingly small barring rank carelessness in attempting to physically move it whilst up and running or it suffering a severe jolt by someone bumping into it accidentally.
« Last Edit: June 19, 2021, 04:06:26 am by Johnny B Good »
John
 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 4646
  • Country: de
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #21 on: June 19, 2021, 09:02:20 am »
I once heard the saying:

There only 2 type of drives;
Those that fail and those that will eventually fail.


So true from my experience.
I had failing drives from all manufacturers.
Lately WD drives have been very reliable for me.

 

There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline Johnny B Good

  • Frequent Contributor
  • **
  • Posts: 506
  • Country: gb
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #22 on: June 19, 2021, 03:30:40 pm »
I once heard the saying:

There only 2 type of drives;
Those that fail and those that will eventually fail.


So true from my experience.
I had failing drives from all manufacturers.
Lately WD drives have been very reliable for me.

 Surprisingly true for me too but I've only had experience of WD drives during the past eight years from a sample size of three.

I had to check out the 2013 email archive (a trip down memory lane) to find the only two SMART error warning emails that weren't test emails ::) In this case, it involved a 4TB Hitachi Deskstar that hadn't quite clocked up 6000 hours (so within the vendor's one year warranty which I collected on after purchasing an identically sized WD RED to replace it - still going strong to this day).

 The older 3TB 5700rpm Hitachi which I'd managed to purchase from, of all places PC World, for a mere 150 quid (this was during the Taiwanese flood induced shortages) had always shown LBA errors almost from the start during the first 10815 hours of its life stopping at a count of four for the next few years before it was finally retired with no untoward consequences.

 The 4TB Deskstar 'wonder' didn't show any errors until that first email which I didn't see on account of my disdain for email as a "high priority, you must deal with this straight away!" messaging system, only seeing those first and second emails when checking to see whether there had been any such SMART warnings after seeing actual problems reading and writing data to the FreeNAS server (as it was still known back then).

 I was very lucky in that, after a 48 hour run with ddrescue, I managed to recover every single sector to its replacement WD40EFRX some three days later. Said replacement is still going strong to this day, almost 8 years later, some (reported)  32252 power on hours - with WD drives, you need to remember that the PoH counting algorithm suffers from "Dorean Grey Syndrome" >:D.

 At that time, I still had a couple of 2TB Samsung Spinpoints in the box (never ever used RAID - couldn't afford it, plus knowing me, I'd most likely cock it all up if I ever had to resilver a replacement into the array anyway). One had clocked a mere 168 thousand head unloading events, the other a staggering million plus, neatly explaining the ensuing Multi_Zone_Error_Rate of 26081 versus a zero value for the other. They'd both accurately reported some 25,000 PoH by then (three years' worth - the longest period of active service of all the ever increasingly larger sized drives I'd fed to my home server to keep just ahead of my ever increasing storage requirements).

 Actually, the 1 million plus head unload events drive had clocked 2,000 hours less PoH. I can only surmise that in my experimenting with the various power saving options, I'd somehow managed to get it to emulate the same head unload behaviour you get by default with the WD drives (8 second head unload time-out FFS!).

 If you're setting up a home server, whether a ready made or, best of all, something like FreeNAS or its FreeBSD based derivatives, it's well worth monitoring the SMART logs to make sure there aren't any hidden 'timebombs' like WD's infamous 8 second head unload time-out to send your investment to an early grave. Also, don't let the power savings on the annual electricity bill tempt you into placing your drives under more stress than they already have to contend with - let them spin 24/7, they're far more likely to survive to the next disk capacity upgrade that way. >:D The three WD drives (10+6+4 TB's worth) in mine have all been spinning non stop for the past 746 days of up-time as shown on the status page of my XigmaNAS box.

« Last Edit: June 19, 2021, 03:33:38 pm by Johnny B Good »
John
 

Offline Johnny B Good

  • Frequent Contributor
  • **
  • Posts: 506
  • Country: gb
Re: EEVblog 1398 - Western Digital RED 6TB Hard Drive TEARDOWN
« Reply #23 on: June 19, 2021, 07:36:20 pm »
 Following up on my last post:-

 It has occurred to me that Dave's remaining three drives may also be about to expire... from head unload counts that exceed WD's stated lifetime limits of 300,000 for commodity class and 600,000 for enterprise class drives. He really should take a closer look at the SMART logs to verify the situation.

 This has been an issue that came to light over a decade ago with all WD drives. To my knowledge, an issue first raised by rather irate fielserver sysadmins discovering such counts exceeding those limits within just a year of being commissioned (and by quite a large margin).

https://www.storagereview.com/news/how-to-stop-excessive-load-cycles-on-the-western-digital-2tb-caviar-green-wd20ears-with-wdidle3

 Incredibly, even as late as October 2018, I was rather dismayed to discover that the shiny new Helium filled 10TB WD100EFAX-68LHPN0 I'd just purchased was also in need of the WDIDLE3 (or equivalent) treatment to cure it of this malady :palm: :wtf: :wtf: :wtf: :wtf: :wtf: :wtf:

 After seeing head unload counts on some WD 2.5 inch laptop drives in the range of 3.5 to 5 million with no apparent deleterious effects, I've come to the conclusion that this was a power saving strategy that not only doesn't scale to desktop sized drives but is also an unnecessary power saving feature that needlessly afflicts access performance.

 Such head unloading on a very short time fuse in laptop drives is actually beneficial in more ways than just a simple power saving mechanism where any such reduction in access performance is swamped by other power savings. For one thing, the chances of the heads being parked when a laptop is nudged into a free fall induced appointment with a high G event (or a venting of frustration with a satisfyingly loud lid slamming shut operation) are quite high, thus affording significant immunity to the consequences of such shocking events.

 I guess WD had suffered almost as stupendously as  large a 'brain fart' as Seagate's use of a totally stupid misuse of a permanent 10 or 15 minute idle timeout spin down power saving work around to the problem of preventing their very hot running drives inside of unvented external USB enclosures from overheating, forgetting that this simply doesn't work on a half hour or longer backup or restore operation. :palm: :palm: :palm: :palm:

 The only saving grace in WD's case being that they did at least provide the WDIDLE3 utility to allow the savvy end user to disable (or at least extend the 8 seconds timeout period to a less damaging 5 minutes). In the case of those Seagate "Specials" (literally identified with a letter'S' suffix!), there was no cure for the spindown restartup delay other than to repurpose them as the boot/data/pagefile drive in a Vista system which of course, completely eliminates (seemingly be design) any chance of such idle time induced spin down events ever occurring.  >:D

 Presumably by now (but I was making this assumption nearly three years ago  :( ) WD have finally seen sense and defaulted this head unloading feature to DISABLED. However, I'm not yet ready to test this with another WD RED NAS drive purchase in order to answer this question so I'd be rather interested to see if others have checked their more recent purchases of WD product in order to answer this important question. It would also be nice to see some feedback from Dave in regard of the SMART log head unload counts on the remaining survivors in his pool of drives. It might just indicate that I was the last purchaser to have gotten hold of a 10TB RED still afflicted with the infamous 8 seconds head unload time out feature (er BUG!).
John
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf