Jump to content
CCleaner Community Forums
Guest Keatah

Why to never buy 3.5" external HDD's

Recommended Posts

Guest Keatah

Why to never buy 3.5" external HDD's? A good number of reasons. But 2.5" are ok. If I haven't warned you guys & gals before then read up!

 

1- The drives that go into these external enclosures, like the WD MyBook series for example, are sub-standard.

 

2- More surface errors. Externals have more errors on the surface right out of the factory than their equivalent desktop models do. Every single HDD ever made in the entire world has surface defects. But externals have more than their desktop counterparts.

 

Every HDD comes with a bad sector map. The defective spots are recorded in a chart and the drive skips over them, substituting a bad sector with one from a reserve area. Thus the disk "presents" a 100% error free drive to the OS. And in the old days, you had to manually enter the defective sector list by hand according to certification chart taped to the drive at time of mfg. Today it is stored in th firmware.

 

Accessing these spare areas takes a few mSec more than a normal sequential sector - because the head has to swing to the reserve area. This latency (for external disks) is buried out of sight deep inside the USB interface. But on a hi-performace Enterprise class drive, speed loss due to defect remappings are noticable and thus they get the best platters.

 

3- Balance. You'll note that on average external drives generate more vibration. This is from sub-standard bearings in the motor/spindle and out of round platters. While each HDD (if you take it apart) seems ultra-precise, there's a real difference when looking at their tolerances on a microscopic scale.

 

4- Cheaper engineering. You'll find less cache in externals along with single processor cores as opposed to 2 and 4 core controllers. That, and number of heads. As the number of heads increase, the reliability decreases. The desktop equivalent will have less surfaces, thus less parts to fail, and fail they do!

 

5- Overall machining tolerances and materials are worse in the externals. The materials are more sensitive to temperature changes, parts expand and contract more, producing different geometries of the heads and surfaces, and therefore requiring more compensation from the electronics to keep things on track. Incidentally, this creates more heat. Typically 5-10`C. Might as well throw in +5`C more due to rougher and less than "perfect" machining of the bearings. And unbalanced platters require a bit more current to keep them spinning steadily. More stress more heat. Unbalanced platters add about 3`C to overall tempertature gain. All this consipires to compound things. The metallic coating (where your data is stored) has its coercivity affected by hotter temps, and presents a less defined magnetic pattern.

 

6- Larger parts, platters, actuator arms, motors, bearings, fly-height adjusters, micro-positioners, voice coils, things like that, it all takes more power to operate.

 

7- Lack of any form of cooling! That's correct. There is no cooling fan anywhere in most external disks! You're taking a desktop drive and basically putting it into a box. No air circulation. You don't do this with your desktop drives do you? You've got some kind of airflow, or if not, you've got the disk attached to a metal frame that absorbs some heat. In an external USB-style drive you don't any of those two cooling mechanisms. You might as well put the disk in a baggie and then a shoe box. This one factor alone is responsible for the high failure rate of external disks! Externals can typically run 20-30`C higher than their desktop cousins.

 

A hotter running disk from the get-go, trapped in a box with no ventilation, is just asking for trouble. And manufacturers know this! Evidenced by the shorter warranty 1-year vs. 3 or 5 years! They know the disks will fail, they know about the lack of cooling as the #1 reason. They are too cheap to put in a $5 fan, citing a number of excuses - one of which, ironically, is extra moving parts.

 

---------------

 

I've seen Western Digital external drives run at a blistering 68`C !! How can you expect a cheap consumer grade product to operate reliably under those conditions. You can't! And I have to say that over the years I've had many drives both professionally and personally, and without a doubt, every single Western Digital 3.5" external disk that has no cooling fan has died prematurely. ALL OF THEM! 100% fail.

 

Manufacturers expect this. They expect a short lifespan. They know this. Remember that 1-year warranty as opposed to 3-years or 5-years!

 

It should be noted that 2.5" disks don't exhibit many of those cascading failure scenarios. They are engineered to work in an enclosed compartment with no ventilation. A laptop! They handle vibration better. Parts are smaller, better spec'd. Less stress, less power consumption. Better surface verification procedures. All a necessity with the higher density of 2.5" platters and portable environment.

 

So if you're looking for external USB disks buy only 2.5", afterall these smaller drives were engineered for the portable environment from the ground up. They were not afterthoughts and hey let's do this types of hacks.

 

And eventually, once SSD's become ready for the consumer market, then I will recommend those. For now, stick with 1TB 2.5" externals for your cost-effective storage needs.

 

Get two 1TB 2.5" disks as opposed to a single 2TB 3.5" external.

Share this post


Link to post
Share on other sites

Perhaps the title needs changing.

You give arguments against 3.5" drives,

but say nothing about 5.25" drives

(I remember how much better a 5.25" Floppy Disc was than the earlier 8" Floppies.)

 

4- ... That, and number of heads. As the number of heads increase, the reliability decreases. The desktop equivalent will have less surfaces, thus less parts to fail, and fail they do!

Please explain further. Your statements surprise me because your reality differs from my preconceptions (I admit I have never taken a HDD apart).

 

I assume that you are referring to physical heads,

and not the fiction shown by the "properties" of my 600 GB Western HDD WDC6401AALS-00L3B2 that it has

129 Heads accessing 4 off 512 byte Sectors per track on 2422991 Cylinders.

 

I understood that an HDD might have one or more platters, with two heads per platter (one on each side).

I assume that there are technological constraints upon how close together the tracks could be, and how close together the flux transitions may be,

and consequently how many bytes may be held on a given surface area,

and hence a Desktop 3.5" platter could hold about twice as much data as a Laptop 2.5" platter,

so for a given capacity there might be twice as many surfaces and heads to fail on the Laptop 2.5" drive compared with the Desktop 3.5" drive.

 

Those are my preconceptions but I am open to education

 

Alan

Share this post


Link to post
Share on other sites
Guest Keatah

Yes, the title should read 3.5" disks. I do not know how to change the title of the thread. (Mod: edited title)

 

Regarding point #4: let me clarify.

 

If you have a 500GB WD desktop drive & a 500GB WD external usb drive, both 3.5" size, take them apart. You will find higher-quality and less platter count in the desktop variant. The lower quality and older versions go to the cheap consumer external disks. This way the mfg can blame early failure on rough consumer handling, after all, it is an external drive. Hard to argue with that!

 

The external usb-enclosure disk is likely to be last year's model with more heads and lower density platters. The heads will fail before the platters loose magnetic cohesion.

 

The track density or how close they can be is a matter of how precise we make the heads, and how much resolution the "spray on" magnetic coating has. Both are still at the mfg limit and have not run up against a laws-of-physics limit yet. Soon though.

 

When I say heads, I tend to refer to how many physical heads are in the drive, anywhere from 1 through 10 in modern disks, currently.

 

2005 = 500GB w/8 heads on 4 3.5" platters.

2010 = 1TB w/4 heads on 2 2.5" platters.

2012 = 2TB w/6 heads on 3 2.5" platters (thick size 2.5 drive)

2014 = 2TB w/4 heads on 2 2.5" platters (standard size 2.5 drive)

 

Understand that in laptop drives, they're limited to 2 platters. And any increase in capacity comes from areal density. Any mfg does not like to do more than two platters in a lappy drive. Takes a lot of energy spin up, makes them thick, generates more heat, weighs more.

 

And in a laptop, they will sacrifice capacity in order to maintain small & reliable mechanics. That's why 2.5 drives lag 3.5 drives in capacity, but always forage ahead, staying 1 step behind their 3.5" brethren.

Share this post


Link to post
Share on other sites

Thank you for the explanation.

 

Until today I thought the only disadvantage of a 3.5" external drive was that it took too much power for a USB2 port,

and so required a separate powered hub that should NOT be switched on at the same time as the P.C.

( I encountered a "race hazard" if the external drive was powering up at about the same time as Windows XP,

which resulted in System Volume Information violating my requirements.)

 

Supplementary questions about 2 platters with a total of 4 heads :-

 

How many independent actuators ?

i.e. Assuming access is required to 4 different files and they HAPPEN to be accessible to 4 different heads,

will 4 independent actuators move them instantly or sequentially ?

( I vaguely remember reading terms such as "Elevator Control" on high throughput enterprise drives one or two decades ago)

 

When a Partition is created on an empty drive, does it populate only one surface at a time until complete,

or does it spread across all surfaces simultaneously ?

 

Are the LBN file clusters in sequence on one surface, or stepped across from one surface to the next to maximise the rate at which the file as transferred ?

Will partitions populate

 

Regards

Alan

Share this post


Link to post
Share on other sites
Guest Keatah

Let's answer these supplemental questions, I get asked them quite often actually.

 

1- How many independant actuators? The majority of consumer and industrial drives have one big main voicecoil actuator. This swings all the heads in parallel (as one solid unit) at the same time. This is done for simplicity and reliability. This one linear motor positions all heads at once.

 

The latest and greatest enterprise class drives will also use 1 single actuator for positioning the heads. But for extra precision each head will have a piezo crystal that can jog the individual head into a more precise position. Make no mistake, all heads are reading track 420,763 on each platter. The piezo positioner just makes sure the heads are dead center as possible; compensating for out-of-round tracks, vibration, temperature, things like that.

 

Furthermore, to help compensate for temperature and air pressure variations the heads also have a fly height adjuster, this is like a heater wire that expands or contracts, thus it pulls on the wires and changes the angle of attack of the head surface - much like an old-school airplane with cables connecting the pilot's stick to the control flaps. This allows the head to float closer than ever to the surface below. The closer the head is to the surface, the greater the data density can be.

 

And the newest disks under development as we speak, for next year, will also rotate the heads like a compass, better aligning the head-gap (where the data gets shot out and into the platter) with the surface for even more density.

 

In the past, companies (Connor did the most experimentation here) have tried building drives with heads that swung in multiple directions at different times, each one moving independantly. And they even tried multiple heads per surface! This never worked well in practice because the technology wasn't advanced enough to implement these concepts. Too much heat, mechanicals too large, and so on. And not only that, they didn't have a controller/processor sophisticated enough to coordinate all this activity.

 

If that was tried today it would be a different story. But the industry isn't interested in that anymore, too much extra parts required. And we've since developed caches and buffers and algorithms to really optimize the spreading of data among different tracks. And in trying to be energy-efficient, its more effective to use caches than it is to be swinging heads all over the surfaces.

 

If you want multiple actuators chomping on your super-duper-ultra-sized PDF file you just get a RAID array going. Now you're cooking with gas! You got independant actuators jamming 2 sets of heads. And that's why RAID is so fast. And that's why it's more expensive.

 

2- Data is spread out among different platters simultaneously. A write operation will write stuff on platter 0 track 0, when that track gets full, it will go to platter 1 track 0, and platter 2, track 0 and then finally platter 3, track 0. Stepping its way through the read/write task. This way, the control electronics can access 4 separate data tracks without having to move and reposition the heads. Just a simple software switch to select the head/platter/channel.

 

When all 4 of the tracks (a cylinder) are are full, the head actuator moves the heads a notch and then there are 4 more tracks at the ready. Well it's not exactly precisely that way, there's more spreading and interleaving and switching between heads, but you get the idea.

 

Regarding partitions, Your partition will occupy all 4 surfaces in a 2 platter drive. It will be stepped and interleaved like any file. Your mega-sized Word document will be divided up between between surfaces too. So will a tiny post-it note sized text file.

 

But the very first drives from the 1950's and 1960's did not work that way. Think RAMAC. Each file or record generally went onto one track on one surface. No splitting up was done. But of course a file may spill over to adjacent tracks, or 2 files may occupy one track. But the 1st drives never interleaved and stepped between platters.

 

Clusters are stepped among different surfaces too. Sometimes they are grouped, surface 0 might have cluster 0-255, then surface 1 might have 256-512, surface 2 513-769. It really all depends on what the drive mfg wants to do.

 

But to have clusters 0 - 640,000 on surface 0, and 640,001 - 1,280,000 on surface 1 and so on, no, that would be almost forcing the drive to work on one platter and only using the others when the previous one fills up. Nope.

 

I hope all that makes sense. And you might find reading this (you'll have to make allowances for translation) quite interesting.

 

http://en.wikipedia.org/wiki/IBM_305_RAMAC

http://en.wikipedia.org/wiki/Hard_disk_drive

http://hddscan.com/doc/HDD_from_inside.html

http://hddscan.com/doc/HDD_Tracks_and_Zones.html

http://www.hddscan.com/doc/data-recovery-for-dummies.html

 

Ohh and by the way, if you think this is a lot of logistics and work to access a file, just wait till you study how an SSD manages its space. 10x more complex!

Share this post


Link to post
Share on other sites

Many thanks for your time and effort.

 

You have given me immediate answers to things that puzzled me,

but which I refrained from Googling because I get confused and misled because

so many articles either give no date of creation, or if a date is given it might be a revision date,

but refer to technology as if it was current and yet it might actually have been phased out more than 10 years ago.

 

I am now ready for bed, but will use your links tomorrow.

 

Thanks

Alan

Share this post


Link to post
Share on other sites
Guest Keatah

In modern disks (and disks from the past 20+ years too) there is one single actuator that moves ALL heads in unison. Doesn't matter how many heads, surfaces, or platters there are. 1 actuator to rule them all.

 

The controller listens to the OS grabbing files as requested. There are times when the OS will ask for multiple files quickly one after the other. The disk will go back and forth to try to fulfill the requests. Soon enough the drive heads start flying all over the place between multiple files. This is disk thrashing. And it is made worse by file fragmentation too.

 

Many newer disks support disk queuing. That's the elevator control. The disk loads one file, then another, than another. But it pays attention to 1 file at a time. There is little or no thrashing back and forth trying to access 4 files at once. 1 file has the complete attention of the controller. Thus avoiding unneeded head movements.

http://en.wikipedia.org/wiki/Tagged_Command_Queuing

http://en.wikipedia.org/wiki/NCQ

 

A partition is spread equally among all surfaces, as is LBN clusters.

Share this post


Link to post
Share on other sites
Guest Keatah

Some holiday reading there for ya..

 

It's important to keep in mind that as one technology or technique of doing things comes into favor, runs a course, then falls out of favor; there's going to be overlap of what methods are used in a product. And there are different degrees and rates of change in how this occurs.

 

Case in point is the pot-roast-sized 5.25 HDD from the 1970's and mid 1980's. Let's compare it to a modern 1TB 2.5" pocket drive. You'll see the similarities and differences right away.

 

Spinning motor technology - they all spin the disk but there are different methods for energizing the magentic coils.

OLD = stepper-motor-like control

NEW = continuous pulse-width-modulation

 

Head positioning techniques - or how do we get the head into position over the right data.

OLD = a direct drive gear assembly

OLD = a stepper motor that pulls and pushes a metal band connected to the heads

NEW = a linear motor, like an audio speaker, a voicecoil, with the heads attached to an arm

NEW = the above, but with added fine positioning piezo actuators at the head tip itself

NEW = fly height adjusters, controlling head position in a second dimension, up and down

 

Magnetic surface chemistry - there's hundreds here, all kinds of coatings, etchings, depositions, it just goes on and on.

Share this post


Link to post
Share on other sites

Do you have any mainstream sources for any of this? I have been using WD external drives to years without any problems.

same

Share this post


Link to post
Share on other sites
Guest Keatah

I am the source. Our research has shown that many 3.5" drives in enclosures reach temps of +62`C, and 400-500 hundred hours operating like that is sufficient to get you some (current pending sector) and (reallocated sector counts) rolling.

 

If you doubt it go on and try it. See if your disk hits +60`C. If it does you'll be having issues soon enough. The first errors will be handled by the disk without bothering you. You'll never know. When there are no more spare sectors left the S.M.A.R.T. counts will start increasing, and you've got a bad disk. A disk with heat worn bearings and warbling platters the heads can't easily track.

 

The extra effort just causes the disk to heat up more and snowball itself into rapid failure. If your disk doesn't reach the 60 degree range then you're probably alright.

 

3.5" external disks aren't ideal for 45 minute backup runs or 3 hour defraggler sessions. They are meant to run 10-15 minutes with intermittent access. Why do you think they agressively power themselves down? Prevent heat buildup.

 

Regarding long-term reliability. I've got WD disks from the early 1990's that work just fine. And then I've got WD disks from 2 years ago that blew up. So take it for what it's worth. But check those temperatures and see how hot (or not) they get.

Share this post


Link to post
Share on other sites

It seems your main argument against using 3.5 inch external drives is heat. I'll give you a few more.

 

1) 2.5 inch external drives use only 1 cable for usb & power. 3.5 use 2 cables.

2) 2.5 is smaller, so it looks better

3) Since it is smaller, it is more portable

4) Smaller drives also use less power

5) Easier to hide/setup/use/carry/etc.

Share this post


Link to post
Share on other sites

agree with your argument on shop brought, external drives.

I have always found it better to buy the drive and the enclosure seperately.

makes it easier to change when one of those parts goes pear-shaped.

the enclosures you buy yourself have better heat dissipation, come apart easily and usually work out cheaper than the off-the-shelf, moulded, sealed units.

Share this post


Link to post
Share on other sites
Guest Keatah

The main theme is cost cutting. Sub-standard tolerances place too much wear and tear on the mechanicals. The disk controller compensates, with increasing difficulty over time. This generates heat. The process snowballs and there you have it.

 

2.5" drives can't have as much cost cutting, or slack tolerances. If they do, then they fail immediately. So by default, they don't snowball themselves into a heat corner as quickly. The electronics don't need to bust ass compensating for mfg irregularities.

Share this post


Link to post
Share on other sites

in my experience, externals die most commonly from the good old accidental knock or drop whilst the unit is running, long before any other factor gets a chance to show its hand.

 

with 2.5" drives generally costing more $$$ for less gig, people are sadly going to buy based on the price tag and go for the 3.5" units.

 

maybe your thread Keatah will make people think before their next purchase.

Share this post


Link to post
Share on other sites
Guest Keatah

in my experience, externals die most commonly from the good old accidental knock or drop whilst the unit is running, long before any other factor gets a chance to show its hand.

 

Mfg's know that, exactly. No need to put in a long-lasting disk.

Share this post


Link to post
Share on other sites

Who do you work for? I'm not doubting you but your throwing around a lot of "absolute facts" that I've never heard mentioned before. I have WD my book drives that are about 5 years old and work fine. I've never read any articles in mainstream tech journals saying not to use 3.5 externals.

Share this post


Link to post
Share on other sites

The journal editor might be concerned that even if 90% of External 3.5" drives failed for any causes other than high temperatures or being substandard,

the readers of such an article might swamp the manufacturer with 50% of all failures, and not just the 10% that warrant replacement.

 

If the manufacture has been coasting along with only 1% of the failures his costs would be vastly increased,

and this might affect decisions upon which magazines should benefit from an advertising budget.

Share this post


Link to post
Share on other sites

All I'm asking for is a little more info to back up some pretty bold claims. I read lots of stuff about consumer and corporate technology and have never heard these claims. I'm not saying he's wrong I just don't think he's provided enough facts to support his claims.

Share this post


Link to post
Share on other sites
Guest Keatah

Sorry can't provide much more info than I've already stated, at this time. Soon enough a paper will be published for all to read. There's a Google study floating around, but IIRC they focus on the ability of S.M.A.R.T. data to predict failures. I'm most interested in how a disk behaves when you remove all the cooling, as in an external enclosure.

 

Even housings with vent slots aren't always enough. Sometimes the disk inside the housing has even a further "blanket" around it in the form of a mounting plate. The plate covers about 40% of the disk and provides the mechanical interface & mount-points to the external housing. A sub-frame if you will. It also holds the bridgeboard.

 

2.5" drives have none of this.

 

In the meantime, just minimize your risk by checking the temperature of your fanless & ventilation-less enclosures, after 1/2 hour of constant access. File copy operations, defragging, imaging - all good ways to warm up a disk.

 

If disk goes above the low-60's (Celsius) you're asking for trouble. I think we can all agree that heat is the #1 enemy of electronics and electro-mechanics.

Share this post


Link to post
Share on other sites

Who do you work for? I'm not doubting you but your throwing around a lot of "absolute facts" that I've never heard mentioned before. I have WD my book drives that are about 5 years old and work fine. I've never read any articles in mainstream tech journals saying not to use 3.5 externals.

 

Whether all the things he states are fact, I do know a few people that accidentally dropped/knocked a drive & caused immediate permanent damage.

 

Afterwards, all attempts to get the drive to work (short of data recovery lab) failed.

Share this post


Link to post
Share on other sites
Guest Keatah

That is the 3rd reason externals are not built to the same quality as desktop or mobile equivalents. In fact 2.5" disks are engineered to take more pounding than 3.5". It is very difficult to make 3.5" shock-proof. There's simply too much mass in the moving parts.

 

The mfg's know a drive is going to be abused a little more in the external housing. Knowing they can't cost-effectively make them withstand bumps and knocks, they reduce the warranty. Health of the company finances 1st. Health of the customer's data 2nd. This will never change.

 

The large heads in 3.5" will hit the surfaces on impact if drop the disk while it's spinning. The air cushion simply cannot keep them separated. Too massive. 2.5" disks have much much lighter heads, and the air cushion can support them a hella lot better. Less chance for head crash.

 

Don't you just love those WD MyBooks? Tip them over and now you need a pro-service! Great source of income.

Share this post


Link to post
Share on other sites

What company could you possibly not disclose that is testing hard drives? Seems a little sketchy honestly.

Anyway post your article when it's done, I'm interested but skeptical.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×