EEVblog 1400 - Hard Drive Micro Actuators are AMAZING!
Вставка
- Опубліковано 26 вер 2024
- Dave takes a closer look the micro actuator used in the 6TB Western Digital RED hard drive.
Forum: www.eevblog.co...
Subscribe on Odysee: odysee.com/@ee...
EEVblog Web Site: www.eevblog.com
The 2nd EEVblog Channel: / eevblog2
EEVdiscover: / eevdiscover
Support the EEVblog through Patreon! / eevblog
AliExpress Affiliate: s.click.aliexpr...
Buy anything through that link and Dave gets a commission at no cost to you.
Donate With Bitcoin & Other Crypto Currencies!
www.eevblog.co...
T-Shirts: teespring.com/s...
#ElectronicsCreators
It still amazes me that HDDs are able to operate this precisely and reliably usually for a very long time, and the fact you can buy such thing new for just 10s of dollars.
Some of my old stock hard drives can't work anymore somehow, data can't be got.
I designed ICs for hard drives back in the mid 80's. At that time hard drive design was considered the most difficult and technically challenging of all mechanical engineering tasks. I'm sure it is still so.
@@tsites1 Thankyou for your service Tim :)
@@ray-charc3131 Maybe the electrons have fallen out.
@@EndlessDelusion perhaps the resistors are in rebellion as well?
For a given mounting orientation, you can fabricate piezo actuators to move in any axis (x, y, z, roll, pitch, yaw). For these disk arms, my guess would be they lengthen and shorten (rather then bend or curl). And they would likely need to be operated as an opposing pair to manage bending stress on the arm: Firing only a single actuator would cause MUCH less than half the head motion as the non-operating actuator acts as a solid opposing block (the single actuator forces against two fixed points).
Yep, I speculated an asymmetrical drive in the video. Would be silly if it's just one side driven. But if movement is visible then you should be able to see something driving just one I suspect. But the distances are just to small, so it's all moot it seems. But might do some more experiments which would entail taking out one arm completely for access.
Not sure you can manufacture thin flat transducers like that that contract in lenth?
Yeah. That's what I was thinking too. An opposing pair to move from side to side. And the little "springs" on either side provide some stabilizing force.
documents.westerndigital.com/content/dam/doc-library/en_us/assets/public/western-digital/collateral/white-paper/white-paper-dual-stage-actuator.pdf like that?
@@EEVblog from the white paper that evergreen linked, the geometry shown in your video, and what I remember from Smart Structures in grad school, they're only extending and contracting the length of the PZT elements by ~100nm. The voltage applied to the tops and bottoms of the transducers makes them thicker and thinner. Since they're trying to maintain constant volume, the length contracts or extends to compensate for the thickness change.
This is actually great for this application because it essentially gives a sort of mechanical advantage to the actuation, requiring much less voltage on the PZT element to pull very hard on the read head. The stiffness of the polymer mounting the PZT elements to the read head would be very precisely accounted for in calculating how much motion transfers to the read head and the material could even theoretically be tuned to adjust down the amount the read head moves with PZT contraction.
Bloody brilliant stuff.
I would think is is a length change in the ceramic. As stated above a piezo ceramic can be engineered to change length. I once used length changing piezo ceramic in a micro fluid pump.
Hook up an oscilloscope to the piezo and see if tapping the tip generates power...
that's an amazing idea
You can use a 4-10khz signal for the actuator to hear the actuator. Also you should touch the head with something like a pingpong ball to adapt the mechanical vibration to air.
You may need 50V-100V to operate the piezo actuator
Not only mechanically but also the software and algorithm to maintain data consistency and keep data uncorrupted is remarkable.
partial response maximum likelihood is witchcraft!
When you said "you dont need much current to move these micro actuators" I for some reason imagined the guy over at Photonicinduction pushing 1000 amps through it just for lulz.
Awwww… He *POPPED* it 🤦
😝
it is not a knife ;)
..no more harddrive..it´s popped! LOL . All what left is a molten blob of metall on his burned carpet...where is my beer?
That would be fun to see. I'd love to see how the heads would react to that much current. If they'd just pop or if they'd move some before popping
@@thomasvlaskampiii6850 The assembly would most likely just vaporize.
I fondly remember the satisfying clunk-thump-clunk of my first 5MB hard drive as my programs read & wrote data. Hats off to the engineers of modern disk drive designs.
And the satisfying clunk as you sent it the head park command. On some drives this made the entire table shudder.
Then there were the old Shugart drives. The disks were huge, at least 8 inch possibly larger, and the head assembly did indeed weigh more than a pound. The actuator was a huge coil acting on a big magnet that moved several inches for a full stroke. Yes the magnet was a part of the head assembly and the coil was stationary. Given the huge mass involved the access times was pretty amazing. Running a butterfly seek test the actuator moved so fast it looked like a spooky half transparent image of it, but you didn't want to get your fingers anywhere close to that as it made a full stroke about 25 times per second. It had to accelerate the assembly and then stop it after it had moved about three inches 25 times per second. With the heavy head assembly that takes some serious power and it made the whole building shudder. It sounded amazing when you ran calibration tests. The first time I heard it I thought someone was using a jackhammer on the floor.
People was coming from several floors to see what was making such a racket. I think that drive was some where between 5 and 20 MB...
I remember hearing "high-tech sci-fi computery squeaky sounds" from tech scenery, playing through first location of Unreal.
They seem to be, in fact, made up from an ancient MFM drive sounds, seeking repeatedly. The one that had a stepper motor for an actuator :)
I reckon. My very first HDD was an ACTUAL Seagate ST-504. It weighed nearly 10kg for a 10MB drive.
Fascinating to see this stuff up close and realize that they make these in extreme quantities - fully automated. 24/7.
The in-process QC challenge is mind boggling.
I can’t believe humans being build these things. Period.
@@andyapple9 yes, it almost feels like this stuff comes from another planet... it was all made by aliens.
@@xissburg no, it doesn't feel like that when you have the precise history of development and the thing getting better each stage since the time those things used to weight half a ton
@@xissburg if the limit of your understanding is “hurr hurr, light switch goes click” it probably does, yes. But then some people are dumb enough to think the pyramids were built by aliens… who used only local materials and tooling to cut the blocks and deliberately made the simplest possible shape for a stable tall structure.
I've seen hard drives since the beginning. Those old things where the heads looked like magnetic tape heads and the platters were 18" in diameter. I worked with ones were you needed to do a low level format before you actually formatted the disk. And now these where the low level format is done by a precision device at the factory. You can really see all the years of science that has gone into these things.
One place where I worked in the late 1970s, they had a WANG computer. It had a 5MB fixed hard drive and a 5MB removable hard drive pack, about 18" in diameter. It was VERY noisy. When we started it, it sounded like a jet engine starting. Funny thing, when they upgraded from 5MB to 10 MB, they backed up, removed a resistor, reformatted, and restored. That was a very expensive resistor.
Yeah, I've got a DEC RK05 disk pack drive that weighs 110-lbs (~50-kg) with heads/arms nearly 6-inches (~15-cm) long, if I remember correctly. I think the storage capacity is 1.6 (12-bit) megawords, or 14.5K (12-bit) words/pound (32K words/kg). ;) Damn, HDD technology improvement over the years has been amazing!
I know a guy who worked with *really* early ones. The platters were huge, made of Cadmium, and held a whopping 1 Megabyte.
@@dylantowers9367 I saw a collection of old platters that a Professor had. The largest I remember was between 3 & 4 feet (0.9-1.2-m) in diameter and about 1/4-inch (~6-mm) thick, but I don't recall whether he told what its capacity had been.
@@dylantowers9367 I worked att a radar surveillance mountain once and even if everything was state of the art they had put up some "thread memories" on a wall from way back in the time. It was just a long very thin coiled up thread of metal that when "hit" on one end whit piezo material it held that information for certain time before it reflected back. 1 bit fast memory back in the early -50.
Interesting stuff! In my experience piezo actuators operate with higher voltages.. Given that there's no specs I would start with a 30Vpp (some piezo actuators can even go up to 600-3000Vpp but I don't think this is the case) also, for an actuator application the frequency needs to be under 200~300Hz.. I'm curious to see their driver schematics given that the voltage has to be boosted, driven differentially, drive multiple actuators
I think most of you got it wrong. The piezzo mechanisms is to dampen the vibrations of
1) final position of arm. Without the this mechanism, the system control mechanism can never stop the shaking of the arm by itself since the arm is long. It is considered a secondary system of fine tuning, much the same as multiple stages in power supply design. In PS design, if you want to have a very low ripple noise, you must use multiple stages and filters.
2) step motor (spindle) which constantly shakes the hard disk. The piezo electric mechanism can vibrate in the higher frequency whereas the arms step motor cannot.
3) other small higher frequency noises like fans in the motherboard and chassis.
This has not been a new idea. It has been around since 2005.
Besides, the heads have multiple read and write sensors much like a video camera or scanners. They can read multiple tracks at once. If track 8 needs to be read and the arm goes near there, one of the heads detects the right track and the arm may not need to move to get track 8.
Hard disks are one of the most sophisticated electromechanical systems ever invented in such a small space. I'm surprised how cheap they are regardless of number of them being made.
Its incredible the level of engineering that's in Hard Drives, you buy one for $40 and just dont realise the crazy development that's been involved. Amazing engineering for the masses.
Anyone who remembers the Philips VCC series videorecorders: they had piezo actuators connected to the tiny heads on the rotating drum and those video heads could follow the videotrack much more accurate than the VHS and Betamax recorders could. It could for example display perfect stills and stripeless fast forward/reverse searching. So not so new technology just on a much smaller and more accurate scale in harddrives now.
I have repaired some, so I do remember they were a pain in the b... to calibrate. But quite neat that you could use both sides of the tape, like a compact cassette.
Videotape recorders had piezo actuators for track following starting in the 70's. Ampex won an Emmy Award for its Automatic Scan Tracking. If the AST servo was misaligned, you could hear the head tips sing.
First seen in a mass produced VCR in the Philips V2000 format machines and then later on became pretty well standard in most high end VHS machines
That's a big advance from about 100TPI on 14" platters on drives I worked on in the seventies to >400,000TPI in today's drives.
From what I read the displacement of the micro actuators is in the order of 1µm and below. So not a huge surprise we couldn't see any movement when you applied a few volts ;)
That’s what I was going to say. Probably too small to actually see the movement.
How come your comment is 2 days old when Video came out today Sir.
@@abhijithanilkumar4959 probably patreon
@@garretr4488 Obviously magic.
@@abhijithanilkumar4959 Oh, in Australia is it tomorrow already or something like that, so the comment could have been made somewhere near the international date line, going around in the wrong direction and arriving two days before it was made, instead of a day after, if it makes any sense. :)
Visible light is usually defined as having wavelengths in the range of 400-700 nm. You can not expect to see movements in range of 50 nm. Sometimes I did not manage to get sharp images of 50 nm structures with the SEM.
Well, with interferece you could
@@kramnecknerf I know Daves office light is not monochromatic and I guess he hasn't a Michelson interferometer at his bench.
All this talking about harddrives made get the "Get perpendicular" song stuck in my head. Thanks Dave! ;)
Hahahhaa good vibes! Old school :-)
It's like image stabilization, but for your data! That is pretty neat.
Great video! I have had success observing motion of piezo devices using a poor man's strobe, e.g. a blinking LED. If you can excite the actuator at mechanical resonance, it should deviate more (perhaps greater than the design calls for). You can find the resonance frequency electrically with a 1-port network analyzer or sig gen and scope. If the strobe frequency is slightly different from the excitation frequency by a few Hz, you might see a slow oscillation of the arm. You can do some back of the envelope math to determine how much deviation you can measure with a given magnification. Just a thought...
@EEVblog Assuming the HDD drive spins at 7200rpm I would expect it to wiggle at and above 120Hz to keep the head on an elliptic / excentric track, so you might try it with a higher frequency. This would allow it to be heard in operation. Reminds me on the movement of a Laser pickup lens in a CD drive and its movement on an excentric CD.
The silver glob of glue is conductive and grounds the back side of the transducer.
These videos made be realize how cool hard drives are
11:49 I remember about 30 years ago, Tim Hunkin said something similar about a VCR head chip on "The secret life of machines" episode :)
Look up Tim Hunkin's YT channel. He's posting remastered Secret Life of Machines videos with added commentaries. Enjoy! :-)
@@WilliamDye-willdye Exactly what I did - what a nice surprise!
Shame about the 3rd season, existing recordings were mildly corrupted at the ends as far as I can remember.
The control loop strategy is also very interesting, using a macro-micro topology with complementary paths. The low pass response is actuated by the arm and the high-pass part by the micro actuators. Neat complex system equations 🤯
High end disks use 3 actuators now. Whole arm and two smaller joints
@@cannesahs I had a huge 20MB hard disk and it was highly educative to see the anatomy --- you do not need a microscope (and I was younger then)
Hey! you got a nice channel there! I subscribed. Keep making those videos...more power to you!
@@seeker4430 Thank you!! 😄😄
Dude you have more energy right now than I had in that part of my teens after I discovered Red Bull, and before they made them age restricted.
Awesome. I love the video!, there's some awesome technology in these things and it's easy to take them for granted!
7:25 - "A thing of beauty is a joy for ever" - John Keats, Endymion
Oh jeez, just imagining the sort of block allocation algorithms you'd need to be able to take advantage of being able to independently micro-actuate each head while having them still not totally independent at the main arm.
My sharpshooter friend actually adds mass to his gun. This helps him keep it steady.
The mass of the arm would help hold the heads steady while the fine tuning is accomplished by the micro actuator.
The micro actuators are not simply for locating a track with high precision, they are actually there to overcome oscillations in the main part of the arm. When the arm moves from one track to another, it can't simply stop when it reaches the desired track. The arm will flex and if the drive coil simply stops when the head is over the track, the head will continue past the track and vibrate back and forth. Feedback to the drive coil is used to dampen this vibration, but given the extremely small distances between the tracks, the latency necessary to dampen this vibration and settle on the track using only the primary drive coil is significant. By using the micro actuators, the head can remain over the track even as the main arm continues to oscillate slightly. So the primary purpose of the micro actuators is to reduce latency.
But how is the feedback to the drive coil picked up ? I thought to see piëzo sensors in them.
@@erikdenhouter I'm quite sure the feedback is readfrom the actual track, some kind of data encoding that will ensure not all zeros or ones written for any significant length of track, so the intensity of the signal can be used to calculate the offset (in radial direction) Ols harddisks used 'dithering'to be able to know the sign of the erroro (am I moving in or out of the track), Not sure how it is done here,maybe two slightle offset heads?
Truly amassing & taking for granted 10/10
Not only does the mass of the arm contribute, but at speed, the elastic nature of the whole arm starts to become dominant in position error. You get lots of ringing and in order to reduce settling time, which is inversely related to read speed, the piezo actuators help keep the head inside the desired track. Its counteracting the large read arm vibrating like a rubber band. The voice coil motor has a lower bandwidth typically too. The piezo can be driven at a much higher control rate, which will help when chasing higher transfer speeds.
9:15 The "spring arm" on the back is one electrode, and the bridge across the red insulator on this side grounds this side to the frame of the arm. Piezoelectric crystals shorten along the current path, lengthening along the plane perpendicular to that path. When current is applied, the piezo-actuator wedges both red insulators apart, causing the "I" shaped flexure ( en.wikipedia.org/wiki/Flexure ) to deform at the weakest point which would be where there is a hole punched through seemingly for no reason between the actuators.
I perceived this as shifting the magnetic field up and down to vertically access the surfaces of magnetic plates laminated to a single plate. Am I wrong.
I would say that it is actually addictive to follow you in trying to understand the beast.
Apply 40kHz to the piezos for ultrasonic platter cleaning!
HECK yeah a Whole Dedicated Video about these! Absolutely Wonderful!
This is really cool I never knew HDD heads had this kind of micro-actuators at the tip. It is so ironic with all this technology in spinning HDD and it’s all virtually obsolete technology except for the largest multi TB drives used in data center servers and large workstation PCs. Everything else has already transitioned to flash.
Next stage after flash already exists.. But flash needs to hit it's limits before next switch.. mram...
Yeah, it has recently occurred to me that we are probably now in the golden age of magnetic media. Pretty soon, economies of scale are going to be working against magnetic storage and the incredibly low cost-per-TB we see now is going to go away. Hopefully that only happens when the alternative media are all at least as dense. (But the sentimentalist in me will miss spinning disks when they vanish for good.)
@@nickwallette6201 No, magnetic is the future, but in solid state.. MRAM and things like STT-MRAM.. But yes, these mechanical beasts we have today will eventually vanish... Many thought they would be gone by now, but I think they will survive flash...
One has to be thinking what kind of storage medium is best for long storage archiving. Mechanical spinner HDD may be the best for archiving, as long as it’s powered down for the archived period of time. Or it may be the worst form of storage I don’t know
@@ericcindycrowder7482 DNA is honestly pretty damn good if it's temperature controlled. You get 2 bits of data per about 30 atoms if I've imagined it correctly (it won't be far off if it is, certainly not an order of magnitude)
Read speed is kinda rubbish currently but stability is great
Good heavens, How do they even manage to get those exact and most precise movements. Just, really unbelievable😱😱😱😱
I can only imagine. Maybe the heads are tuned individually before the drive is marked as a PASS. This means each head will require a unique amount of residual voltage governed by hdd board. This could be regarded as part of the low low formatting of the drive ???
One of my co-workers used to be a QA dude at Hutchinson Tech (Hutchinson, MN) who specialized in making these little tiny hard disk parts. Sadly he doesn't have "engineer brain" so he didn't notice a lot of the technical stuff I know I would have, which limited his ability to tell fun stories about the amazing tolerances they had... but it still sounded like they had a pretty extensive test regime, to make sure nothing that didn't pass all specifications was shipped.
Hello those micro-actuators are for adjustment of head vertical distance from the disks, it usually is like this : 1 for GND ,2 for differential read element,2 for write element
and 1 for vertical z micro-actuator and other two are horizontal x-y micro actuators. total of 8 connection .
The helical scan head in a video cassette recorder used to be the most precisely engineered thing in the home back in the day.
Info from the Sony patent: The track width of the magnetoresistive head is 0.5 to 0.8 μm, the distance between shields is 0.13 to 0.145 μm.
Yes, and Video 2000 used piezo crystal to move the heads for trick playback. It worked quite well.
Perhaps try setting it up with a record player stylus against the head and try to see if you can get some audio to go through that.
Normal person: Sad by the loss of a hd and replacement cost. Tosses crashed disk in the bin.
Nerd: Curious about disk crash. Provides hours upon hours of exiting exploration time of the remains.
Don't pass up the magnets that are in there, you will be hard pressed to find anything stronger than you can get out of a hard drive.
@@JWH3 I collect those magnets, and they're so damn handy. You can hold a photo on the wall, mount a utility lamp to the wall (metal steel stud or frame) or a shelf, hang a space rack on the fridge, the possibilities are endless.
Thank you Dave for this technology introspection, of so called "mundane" things but for not entrained eye's.
I think this is a job for the Slo-Mo guys ;)
I worked at a factory that built those suspension heads was very different machine in the process to bond and install those actuators. Plus the white damper installed on the suspension arm it’s self.
I'd like to see if you could aim a laser at the head and observe the reflected beam - perhaps a couple of metres away. The principle is the same as a mirror galvanometer - presuming you could find a suitable point on the head.
I do have a laser displacement device, can't remember the resolution though. Most likely not good enoguh.
Hats Off science folks and engineers and the motivated wealthy industrials who believe and finance/invest in the researches to where it has become now! Joy Forever indeed! 👍👍
Both OxTools and AvE have done good discussions on flexures, like what you're seeing in this actuator arm.
The best practical approach to flexures that I've seen is from prototyping course for scientists by Dan Gelbart. It's on UA-cam, look it up, great stuff
@@shimmerite_ua Genius this guy!
That blows my mind actually. I love hard drives
Dave - the read and write heads dont hover tens of nm over the plater :) - they hover 3-6 nm since quite a very long time - there is even a speciaj heater in the write head that bumps the head and brings it closer to the surface ;) - Very nice you took that topic here in eevblog, thank you!
Thanks so much for sharing. 😎👌🏼
If you have two coplanar piezo elements, or a single element with two coplanar electrodes on the side opposite a common electrode, the area in between the two elements/two electrodes will twist if one side expands and one side contracts. This configuration can also be used to control up and down motion simultaneously. It could be that this is how the side to side motion is performed, but who knows given that you can't really tell how a piezo element moves just by looking at it.
There's a homemade scanning tunneling microscope design that does this using a piezo speaker element with the silver electrode cut into four quadrants. Each "diagonal" pair of electrodes forms the X or Y axis. The movement on the Z axis is controlled by a voltage uniformly applied to all four electrodes (making each one expand or contract identically) and the movement on the X and Y axes are controlled by the difference in voltage between each complementary pair (creating a "twist" in between them.) Commercial STMs use three linear elements, one for each axis, but this design works well enough that one hobbyist managed to image individual carbon atoms on the surface of graphene with it, and at a cheaper price for his whole setup than one single purpose-made STM actuator costs.
I don't know about being the most precise mechanism we own. Those phone Gyro and accelerometers are pretty precise and mechanical
Oh yeah….
MEMS gotta give credit where it's due
Good one! And a shout out to the sheet metal stamping, forming and punching die designers for those parts as well; I'd love to do a field trip to see those presses in operation making the parts for the read/write actuator.
This would be a cool topic for Applied Science. He's got an electron microscope.
I think the voltage applied is DC. piezos will expand or contract depending on the polarity of the DC. Also when a piezo is manufactured they are polarized in what ever plane is useful for the application. BobC and Egwene22 got it right.
I've got ideas for how to test it precisely...cast a shadow that make it bigger and easier to see small movements... Put a lever on it so it makes a small movement into a big one.... Lean a micrometrer onto the head so you can measure how much itt moves
First, they store data on top and bottom of the platters. Second, the position control has two loops: coarse control using electromagnetics and fine control using piezoelectrics. The micro actuators eliminate residual overshoot, ringing, vibration, etc. once the heads are close to decrease seek time. Third, the two piezoelectric elements are operated differentially so that one pushes while the other pulls and the head bracket rotates about an intermediate pivot. Check to see whether the opposite sides are connected to power for one and ground for the other. Also, piezoelectrics generally operate on tens or hundreds of Volts. Finally, the motion might not be visible without X500 or so magnification or an optical lever.
The range of motion in this system is around 1 um, and the voltages are around 10V. Optical lever is a good idea -- with a 5 meter arm, the movement of reflected beam should be clearly visible, provided the vibrations are kept low.
The Philips Video 2000 system from the 1980s used piezo elements to move the video heads for "dynamic track following". I seem to remember the voltage they used was pretty high, like 70V or something. Of course those piezo elements were much bigger than those in that hard disk.
If I understand correctly, putting a voltage over a piezo element makes it bend along the longest side, so I imagine the elements go ever so slightly U-shaped and pull their ends towards each other by just a bit. The springs on the side will keep the head assembly from bending with the piezos; they're not needed to make everything go back in place when the voltage goes off.
Works of Art..
I have always thought they were amazing and never heard anyone else talk about the actuators.
The little flexures allow for 'front-back- motion, to be converted at the head into 'left/right' motion
that microscope camera is beautiful
The multi-stage actuator seems to serve two purposes when I saw the animation: more precise head positioning, PLUS keep the azimuth perpendicular to the radius even though the head travels in an arc.
I don't think the arc movement matters. As long as you're reading at the same azimuth as you wrote at, everything should work right. I don't think there's any need to have a consistent azimuth from one track to the next.
Last time I played with one of these. I powered the acuator on accident with the diode range poking around.
If memory serves. The actuator energized will contract the head on the same side. It pivets in the metal between the two elements.
Also if memory serves under those conditions the movement was indeed visible even with the naked eye.
Absolutely incredible! Controlling something physical, at the nanometer scale, and it’s something most of us use everyday and have no idea.
der8auer recently did a few videos at a lab that has a SEM with probes small enough to test INDIVIDUAL TRANSISTORS on like 7nm process CPUs 🤯
It also uses piezoelectric elements to get such fine positioning.
I didn’t know about using piezoelectric elements for positioning until that video, now this shows me it’s a lot more common than I knew! Very cool.
HDDs are impressive mechanical beasts.
Though, sometimes I wonder why manufactures don't just go to a larger form factor. It would increase material costs a bit, but increasing the disk diameter by just 10% makes a fairly huge impact to disk area and hence capacity. And most of the cost of the drives isn't material cost, but rather assembly costs, and that should remain largely the same for a drive that is physically 10% larger.
Micro actuators like this would also help combat a lot of the issues with going to a physically larger disk.
I for one wouldn't mind seeing a 5.25" HDD if it has a good price per TB.
Though, the big issue with hard drives at current is that they are starting to suffer from their low read write speeds compared to their capacity.
In short, moving data away from it takes a considerable amount of time, and if one is repairing a raid array, then this time is time that another drive can die within and one's data ends up lost. So there is some concern in regards to write speed per TB as well.
Then there is SMR drives, something that could be handled a lot better by RAID systems. Though, a fair few drives are hiding the SMR operation away from the OS and drive controllers and such... Then there is the other extreme called Host Managed SMR, but that at least has advantages when rebuilding the array, since one can much more intentionally work with the shingled blocks and therefor not end up rewriting the same block x many times just because the RAID system didn't know it were the same block.
Mainly it's compatibility question. The big piece of hard drive market goes to datacenters and similar big commercial system, and nobody is going to replace all their equipment just to accommodate larger drive.
And also, engineers have to develop dense packaging techniques anyway, because companies want to sell hdds for laptops, and they won't get any more space there. So, if you are going to invent this thecnology anyway, why not use it in bigger drives?
I doubt its about material costs in the BOM, surely they stick with the form factor because there has to be continual physical compatibility with server and PC HDD slots. consider how many servers there are in the world now, all manufactured to accept HDDs of a certain size
Because of ISO compliance reasons.
@@pro100vald HDDs in laptops has rapidly fallen out of favor the last 5 years to the point that finding a laptop with a hard drive is hard.
And commercial systems in datacenters already get replaced on a 3-5 year cycle due to increasing power efficiency. Electricity, cooling and space aren't free things.
And to be fair, space is a more real reason for why HDD manufacturers don't move to physically bigger drives.
@@-yeme- And I never stated material cost is the reason. Just that a bigger drive wouldn't be all that much more expensive to manufacture.
Form factor is a bit more of a reason. But to be fair, going to a physically larger drive has some other downsides, mainly in backwards compatibility. This is however not a major requirement in the storage market where cost per TB is frankly more important.
All though, I did also point out one of the biggest reasons for why more capacity isn't actually a good thing unless one can also bring forth more read write speed.
The faster rotation of the disc, the smaller friction force on the head, it will ease the work of the coil/transducer, and maybe on the coil/transducer voltage is AC with high frequency controlled by PWM system.
If you probe the correct contacts, you can then manually actuate the heads to get a readout on the scope. Then amplify and use that signal profile to drive. #reverse_egineering
Wow!!! Thanks.
As broadcast videotape technology progressed the track widths on tape reduced. At some point 'dynamic tracking' (my phrase) was introduced where the head was mounted directly (or nearly) on a piezo transducer to enable precise tracking. Head and drum servos took care of the basic positioning but the dynamic tracking did the rest.
hard disk is always amazing to me, they have unbelievabe precision, yet still dirt cheap( you can get 1TB hdd for less than $50)
It's an amazing piece of technology we take so much for granted, yeah.
It’s amazing just how much formula one level engineering goes into the spinning rust.
Soon it won't be spinning rust with the read writer head split into two separate actuators. Seagate is getting hard drives of this type that are nearly saturating sata 3 speeds. SSD's will still have these hard drives beat in IOPS by a considerable amount. but hard drives will no longer be painfully slow. it will feel like a low end ssd but with way better overall lifespan than flash ever could.
Getting footage of the movement may be a project for the Applied Science Channel. Someone get a hold of Ben.
40 years ago I was quite sure 10 megabyte (10 mb) Shugart drives were magic and could not possibly increase in size.
Fascinating!!!
I believe that the "Flying " heads can be flown like a wing. The "flexing" of a micro movement is much like the moving the control surfaces to cause a bigger movement of the wing.
5:37 like course and fine 😊❤️👍👌.
Amazing stuff.
Now that's a real artwork.
It's actually pretty fascinating that we can store information this way. Just think about all the manhours put into the hardware, firmware and driver software.
So it is basically the same prinicple behind the way optical disk drive's read/write heads are mounted in a springy way, extremely precisely movable by electro magnets!
Because for optical disks drives, it is the same thing. The head assembly is moved by a spindle but the disk's tracks are too small and not as perfectly centered on the disk, so the head by itself needs to follow the tracks on a very fine scale.
Good analogy.
It's amazing how much technology has gone into producing these disks
It's amazing how ubiquitous and dirt cheap this level of precision has become.
Reminds me of the Drivetec floppy drive of the late 70s/early80s. Has two positioners--"coarse" and a "fine" ones. Used embedded servo recording to get around 3 MB on a 5.25" floppy. Disks had to be factory-formatted, which was a deal-breaker. Kodak eventually bought out the bankrupt company and marketed improved versions of the drive for a short time.
Afaik it's very likely that these go up and down to control azimuth due to pressure changes (for different altitudes) + IIRC they would try to retract if they detect a drop so that there's a lower risk for the heads to touch the platters.
If something is moving something has to be non-rigid in there. So I'll take your word for it. It looks like there are two layers, the dark, and the silver metal, that are free. The actuators are what connect the two layers, and would cause a left right movement on the plane of the platter relative to the arm.
Actually, another interesting question would be what is the resonance frequency of those actuators? (It could be found with a spectrum analyzer.)
Mwahahaha great question
They say "little things please little minds", but actually "the devil is in the detail" too. At one time the stack of heads would be tracking a servo track on one of the other platter surfaces. This no longer works because the alignment between heads can't be guaranteed, so fine are the track spacings e.g.with temperature changes. Having a spare surface is also throwing away capacity. Instead, one solution is to imprint data on the disk, like a chevron pattern that repeats and reverses itself. Using digital counters the head can be centred by noticing that when it is off centre, counting up and down changes the timings slightly and corrections can be made. It's unlikely that any of this movement can be shared between different heads, meaning that every head has to think for itself so that all the heads can be simultaneously reading on track all the time.
Tuning these must get fun when they make smaller tracks. Probably use a high gain but then use a dead band to reduce gains as it gets close to the demand position so you get fast movement and limit overshoot without having to use too much kI or kD. I work on much bigger, less precise equipment so if anyone can shed light on that I would be interested.
the piezo actuators expand and contract on the application of voltage giving side movement.
I perceived this as shifting the magnetic field up and down to vertically access the surfaces of magnetic plates laminated to a single plate. Am I wrong.
Cool. Didnt know those were transducers.
That is fucking incredible!
a shame we'll be losing this knowledge soon when it'll all be solid state memory
I used to recycle the entire drive without thinking until I got the idea of turning the platters (and keeping the magnets for other uses) into wind chimes. To date, i've done it countless times over.
Perfect, Dave!
Just *perfectly roight* , Mr. Jones.
Similar idea to what optical discs players have had since ever (IIRC), an stepper takes the head next to the read position, and then the lens is moved with two coils to get to the exact track, and keep the lens focused too.
I was thinking that there is probably a lot of cross-over technology from optical drives to hard drives. The Law of Accelerated Returns.
Great job Dave. Funny at 10:52, half a bee's WHAT? LOMAO.
NICE SCOPE! Got a link? Is there a delay when viewing and finger movement?
The amount of precision involved makes it amazing that I've seen LAPTOPS that are 10 years old with their original hard disks still going strong! When I took the one out of my Macbook, it was still in working order I was just wanting to upgrade the speed by putting in a solid state disk. By the way, it was a Toshiba hard disk, for what it's worth. :)
Could be the perfect topic for another Applied Science video!
Depending on the type of head those may actually have been the 2nd stage and controlled the tilt and up and down movement. Modern hdd have a 3rd stage attached to that ladder looking thing underneath (right where he cut the demo clip) and it moves the head side to side quite a bit but is also voice coil powered. Other heads only have 2 stages and use piezo side to side actuators with flexure springs.
New technology takes awhile to develop and get into actual disc drive production. One has old farts that say it will not work. One has issues of costs of making stuff never made before. :)
In development work Piezoelectric actuators were placed on disc drive head /slider flextures in the lab in 1990 to test concepts.
Rotary disc drive actuators with Neodymium magnets for 3.5 inch drives were production in the mid 1980s.
The design issues of adding one or two piezo actuators per slider/head is not trivial. The micro actuators were at first where they are in this video. Today the focus is having two actuators right by the head / slider.
For clarity the slider/head is the tiny rectangular block that flys floats above the rotating disc and has the read write gap or gaps.
The bulk of the cost of a rotating disc drive is the heads and media so design details are sweated to the Nth degree
Design details of what is disc drive skunk works might not be in production for eons. So taking apart a consumer drive that is 3 years old is like taking apart a car that is decades old.
An old disk drive engineer told me in 1979 that the only thing that is constant in this industry is change. That is when my business card said disc pack engineering. It did not say disk but disc. I mention this to irritate English majors. :)
30 years ago some twin city university Researcher contacted me in the heads and media dept of a drive company and was all excited about 8 inch dusc drives. That design was sold off to a European company 2 years before. There was no actual hardware in California just some ancient drawings of an obsolete product.. so the chap was happy to get all sorts of heads and media specifications on the fossil device. The drive was so old it used a linear actuator and rails. Plus old 300 and 450 Oersted "mud" ie brown tan platters (discs/disks). I was working on a rotary 2.5 inch drive then. With 3.5 size in full production and 5.25 being sun-setted.
NGK Insulators the spark plug ceramic company has been a big player in Peizo actuator design and production. Besides disk drives the micro actuators are used in inkjet printers as micro pumps.
Often tidbits of technology that are in the journals like magnetic society etc of IEEE are old and after in production since all was secret.
You can google us patent 5210455a to see NKG was into Piezoelectric actuators in 1990. It references older patents that go to 1951.
Bell telephone Labs US patent 2549872A is from 1951 and is about a focusing ultrasonic radiator. It was filed in 1948