hi
if 3.5" 5400rpm hd from samsung and wd is good for sound and lower power consumption as a storage device, would 4500rpm be even better for those purposes?
obviously this should be coupled to a high performance sd drive
if 5400rpm is good, is 4500rpm better?
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
-
NeilBlanchard
- Moderator
- Posts: 7681
- Joined: Mon Dec 09, 2002 7:11 pm
- Location: Maynard, MA, Eaarth
- Contact:
Typical HDD rpms are:
3600, 4200, 5400, 7200, 10000, 15000
Of course there's some odd rpms like Seagate 545XE at 3811rpm and "7200rpm" WDs a few years back were actually around 7000rpm. Seagate Low-Power is 5900rpm. And Hitachi HDD spin at 4500rpm when in non-operational low-rpm mode. There was some 20000, 22000rpm experiments for the enterprise sector.
But still the closest common rpm point is 4200rpm. That's typically what they publish, even if the HDD actually spun at 4000 or 4400.
But yes, regardless of whether it's 4200 or 4500rpm, it would be quieter than 5400rpm if all else would be identical: i.e no cut costs in cast, top cover, bearings, motor, etc. Typically we've seen cost optimizations with 1...2 platter drives compared to flagships. Sometimes the low-capacity drives have actually become more noisy than bigger variants but usually the noise drop is still present (even though smaller than one might expect). In some cases, no costs are cut where noise is considered. Samsung downscales for example have been as sturdy as their flagships, meaning that they scale down is noise as well.
3600, 4200, 5400, 7200, 10000, 15000
Of course there's some odd rpms like Seagate 545XE at 3811rpm and "7200rpm" WDs a few years back were actually around 7000rpm. Seagate Low-Power is 5900rpm. And Hitachi HDD spin at 4500rpm when in non-operational low-rpm mode. There was some 20000, 22000rpm experiments for the enterprise sector.
But still the closest common rpm point is 4200rpm. That's typically what they publish, even if the HDD actually spun at 4000 or 4400.
But yes, regardless of whether it's 4200 or 4500rpm, it would be quieter than 5400rpm if all else would be identical: i.e no cut costs in cast, top cover, bearings, motor, etc. Typically we've seen cost optimizations with 1...2 platter drives compared to flagships. Sometimes the low-capacity drives have actually become more noisy than bigger variants but usually the noise drop is still present (even though smaller than one might expect). In some cases, no costs are cut where noise is considered. Samsung downscales for example have been as sturdy as their flagships, meaning that they scale down is noise as well.
whiic wrote:Typical HDD rpms are:
3600, 4200, 5400, 7200, 10000, 15000
Of course there's some odd rpms like Seagate 545XE at 3811rpm and "7200rpm" WDs a few years back were actually around 7000rpm. Seagate Low-Power is 5900rpm. And Hitachi HDD spin at 4500rpm when in non-operational low-rpm mode. There was some 20000, 22000rpm experiments for the enterprise sector.
But still the closest common rpm point is 4200rpm. That's typically what they publish, even if the HDD actually spun at 4000 or 4400.
But yes, regardless of whether it's 4200 or 4500rpm, it would be quieter than 5400rpm if all else would be identical: i.e no cut costs in cast, top cover, bearings, motor, etc. Typically we've seen cost optimizations with 1...2 platter drives compared to flagships. Sometimes the low-capacity drives have actually become more noisy than bigger variants but usually the noise drop is still present (even though smaller than one might expect). In some cases, no costs are cut where noise is considered. Samsung downscales for example have been as sturdy as their flagships, meaning that they scale down is noise as well.
thanks.
the reason I ask is this:
if this a 4200rpm 3.5" 2tb drive were coupled with say a fast SSD 32GB, all else equal, would the overall experience be acceptable?
could it be software selectable (i.e 4200rpm for quiet, 5400rpm or 7200rpm for speed)
?
As far as I know, currently no hard disk support variable platter rotation speed. While 4500/4200rpm indeed may be better than 5400rpm acoustically, given all other factors are equal, I personally would not probably notice the difference if the hard disk is suspended. Maybe you want to look into a NAS?
what's NAS?sonic6k wrote:As far as I know, currently no hard disk support variable platter rotation speed. While 4500/4200rpm indeed may be better than 5400rpm acoustically, given all other factors are equal, I personally would not probably notice the difference if the hard disk is suspended. Maybe you want to look into a NAS?
Perhaps HD in the future should offer a software selectable 7200/5400/4200 speed depending on user preference. How hard is it to "undervolt" the spindle? 4200rpm could use lower power and run cooler and last longer (and be quieter).
-
NeilBlanchard
- Moderator
- Posts: 7681
- Joined: Mon Dec 09, 2002 7:11 pm
- Location: Maynard, MA, Eaarth
- Contact:
variable rpm
"if this a 4200rpm 3.5" 2tb drive were coupled with say a fast SSD 32GB, all else equal, would the overall experience be acceptable?"
If you used that 4200rpm drive only for storing mp3s, movies, etc. then yes. If you did video-editing, then it depends on what resolutions you're working with and whether you have enough CPU power to bottleneck the IO system. Also, for video-editing 2x or 3x 4200rpm is better than 1x 15000rpm... it all about preventing seeking by sourcing files from different locations and writing the product to a third one.
But anyway, there's no 4200rpm 3.5" drives around... except those vintage HDDs that some of us have lying around. Those aren't quiet by the way. 4200rpm or even 3600rpm doesn't matter much if the drive has ball-bearings and is 10+ years old. Or 20+ years old. My Miniscribe had it's 20th birthday last June.
"could it be software selectable (i.e 4200rpm for quiet, 5400rpm or 7200rpm for speed)"
Hitachi has two rpm modes but only higher mode is operational. While spinning slowly, heads are unloaded off the platters.
So - no. It's not simple to make a variable rpm HDD... if you want to keep it operational during all the rpms. The biggest problem is fly height. Read/write heads fly over the platters by use of a slider... it's a ceramic "wing" that produces lift that keeps it from touching (and damaging) platter surface but also prevents it from flying so high that read/write capability would suffer.
Fly height remains quite stable only on a small range of airflow speed. Since the slider flies at a height of ~1...10 nanometers (those are millionths of a millimeter) the airflow speed is pretty much determined by speed of the platter. If the inside radio is half of the outside radius, so are the circumferences and also airflow speeds. So there's already 50% variation between start of LBA space (outer tracks) to slow tracks on the inner end. Surely you might ask if there's possibility to add 20% more variation to airflow speeds but this might mean the fly height becoming unstable at the inner tracks.
If it simply became lower at inner tracks, it could be possible to design the slider to have minimum acceptable fly height at inner track at lowers selectable rpm, and as head moves outwards or if speed is increase, the fly height is adjusted by a twin-metal support that is specifically heated to make it bend towards the platter, nullifying the increase in fly height.
The problem is not just varying fly height but the fact that to keep head at moderately constant fly height over typical variation of airspeed, the relationship between speed and fly height it not linear or even always positive (i.e at some point, increasing speed may decrease lift). The worst-case would be a bi-stable fly-height where slider might end up flying on several heights and stabilize there, or possibly oscillate and bang the slider against the platter until HDD breaks down due to rapid wear and tear to magnetic medium.
If the sliders were redesigned NOT to try maintaing constant fly-height these unstable or bistable modes would be more rare, but fly-height control by heatable head suspension would become even more necessary as it is today. Nowadays it's used to lower fly-height during operation, where as during idling the fly-height is allowed to increase (as head doesn't require to be able to do anything while it's idlin). And some HDD that don't use a twin-metal suspension and use head attached to the slider, use the heater only when reading, as writing lowers fly-height anyway as it heats the head even without additional heating power. These work by heating the frontmost tip of the slider only. But either way, as it's used today, fly-height control is not applied to compensate varying airspeeds.
Then there's of course some tweaking to be done to allow electronics to read bits at different speeds. This probably ain't as big a problem as the inner tracks already have different bitrates as outer tracks.
"How hard is it to "undervolt" the spindle?"
Loooool. First, like described before, dropping spindle speed is the least problematic aspect of... well, dropping spindle speed.
If you mean a ghetto mod... no, your HDD would become unoperational if designed rpm is not met. HDD controller would refuse to load heads to media if rpm doesn't increasy enough to match the tolerance range.
Undervolting specifically doesn't reduce spindle speed. rpm is controlled accurately. If you lower the voltage to spindle, the spindle controller just keeps supplying longer current pulses to motor... until at some point (10V? 11V?) it would probably notice the undervoltage and refuse to even try to spin up. Or if it somehow didn't check for undervoltage condition, it would just A) be unable to make the first few revolutions (stiction), compare to trying to spin-up a fan with too much undervoltage or B) spin too slowly for a while, until it aborts and retries. Or stays there at mid-rpm, never becoming operational, since it knows it's own rpm, it won't even try to become operational.
Constant speed 4200rpm would be easy to make.
Also, since airspeed varies 50% from outer to inner tracks, you could allow 30% rpm variation if you short stroked the platter by 30% of diameter (0.7*0.7 =0.49, thus quite close to 50%).
NOTE: 30% of INNER diameter. Since the circumference at inner diameter is half of that on the outer, the capacity loss is quite a bit smaller than 30%.
Total area is pi*R(outer)^2 minus pi*R(inner)^2. In my example it would be around pi*R^2 minus pi*(0.5R)^2. (Using just R as synonym for R(outer) to make it easier to read.)
That equal to pi*(R^2-0.25R^2) = pi*0.75R^2
Short-stroked would be pi*R^2 minus pi*(0.65R)^2 = pi*(R^2-0,4225R^2) = pi*0,5775R^2
Dividing pi*0,5775R^2 by pi*0.75R^2 you get 0.77. That is, short-stroking 30% of distance you only lose 23% of capacity.
So, sacrificing 23% of capacity by short-stroking, you could achieve stable HDD operation with a HDD that allows 30% of spindle speed variation WITHOUT redesigning the slider and WITHOUT worrying about the need to apply fly-height control to actuator position.
This got quite hardcore. I think I'll copy-paste it to StorageReview as well. There's a few HDD engineers that like to discuss stuff like this. I'd like to hear their opinion on the matter.
If you used that 4200rpm drive only for storing mp3s, movies, etc. then yes. If you did video-editing, then it depends on what resolutions you're working with and whether you have enough CPU power to bottleneck the IO system. Also, for video-editing 2x or 3x 4200rpm is better than 1x 15000rpm... it all about preventing seeking by sourcing files from different locations and writing the product to a third one.
But anyway, there's no 4200rpm 3.5" drives around... except those vintage HDDs that some of us have lying around. Those aren't quiet by the way. 4200rpm or even 3600rpm doesn't matter much if the drive has ball-bearings and is 10+ years old. Or 20+ years old. My Miniscribe had it's 20th birthday last June.
"could it be software selectable (i.e 4200rpm for quiet, 5400rpm or 7200rpm for speed)"
Hitachi has two rpm modes but only higher mode is operational. While spinning slowly, heads are unloaded off the platters.
So - no. It's not simple to make a variable rpm HDD... if you want to keep it operational during all the rpms. The biggest problem is fly height. Read/write heads fly over the platters by use of a slider... it's a ceramic "wing" that produces lift that keeps it from touching (and damaging) platter surface but also prevents it from flying so high that read/write capability would suffer.
Fly height remains quite stable only on a small range of airflow speed. Since the slider flies at a height of ~1...10 nanometers (those are millionths of a millimeter) the airflow speed is pretty much determined by speed of the platter. If the inside radio is half of the outside radius, so are the circumferences and also airflow speeds. So there's already 50% variation between start of LBA space (outer tracks) to slow tracks on the inner end. Surely you might ask if there's possibility to add 20% more variation to airflow speeds but this might mean the fly height becoming unstable at the inner tracks.
If it simply became lower at inner tracks, it could be possible to design the slider to have minimum acceptable fly height at inner track at lowers selectable rpm, and as head moves outwards or if speed is increase, the fly height is adjusted by a twin-metal support that is specifically heated to make it bend towards the platter, nullifying the increase in fly height.
The problem is not just varying fly height but the fact that to keep head at moderately constant fly height over typical variation of airspeed, the relationship between speed and fly height it not linear or even always positive (i.e at some point, increasing speed may decrease lift). The worst-case would be a bi-stable fly-height where slider might end up flying on several heights and stabilize there, or possibly oscillate and bang the slider against the platter until HDD breaks down due to rapid wear and tear to magnetic medium.
If the sliders were redesigned NOT to try maintaing constant fly-height these unstable or bistable modes would be more rare, but fly-height control by heatable head suspension would become even more necessary as it is today. Nowadays it's used to lower fly-height during operation, where as during idling the fly-height is allowed to increase (as head doesn't require to be able to do anything while it's idlin). And some HDD that don't use a twin-metal suspension and use head attached to the slider, use the heater only when reading, as writing lowers fly-height anyway as it heats the head even without additional heating power. These work by heating the frontmost tip of the slider only. But either way, as it's used today, fly-height control is not applied to compensate varying airspeeds.
Then there's of course some tweaking to be done to allow electronics to read bits at different speeds. This probably ain't as big a problem as the inner tracks already have different bitrates as outer tracks.
"How hard is it to "undervolt" the spindle?"
Loooool. First, like described before, dropping spindle speed is the least problematic aspect of... well, dropping spindle speed.
If you mean a ghetto mod... no, your HDD would become unoperational if designed rpm is not met. HDD controller would refuse to load heads to media if rpm doesn't increasy enough to match the tolerance range.
Undervolting specifically doesn't reduce spindle speed. rpm is controlled accurately. If you lower the voltage to spindle, the spindle controller just keeps supplying longer current pulses to motor... until at some point (10V? 11V?) it would probably notice the undervoltage and refuse to even try to spin up. Or if it somehow didn't check for undervoltage condition, it would just A) be unable to make the first few revolutions (stiction), compare to trying to spin-up a fan with too much undervoltage or B) spin too slowly for a while, until it aborts and retries. Or stays there at mid-rpm, never becoming operational, since it knows it's own rpm, it won't even try to become operational.
Constant speed 4200rpm would be easy to make.
Also, since airspeed varies 50% from outer to inner tracks, you could allow 30% rpm variation if you short stroked the platter by 30% of diameter (0.7*0.7 =0.49, thus quite close to 50%).
NOTE: 30% of INNER diameter. Since the circumference at inner diameter is half of that on the outer, the capacity loss is quite a bit smaller than 30%.
Total area is pi*R(outer)^2 minus pi*R(inner)^2. In my example it would be around pi*R^2 minus pi*(0.5R)^2. (Using just R as synonym for R(outer) to make it easier to read.)
That equal to pi*(R^2-0.25R^2) = pi*0.75R^2
Short-stroked would be pi*R^2 minus pi*(0.65R)^2 = pi*(R^2-0,4225R^2) = pi*0,5775R^2
Dividing pi*0,5775R^2 by pi*0.75R^2 you get 0.77. That is, short-stroking 30% of distance you only lose 23% of capacity.
So, sacrificing 23% of capacity by short-stroking, you could achieve stable HDD operation with a HDD that allows 30% of spindle speed variation WITHOUT redesigning the slider and WITHOUT worrying about the need to apply fly-height control to actuator position.
This got quite hardcore. I think I'll copy-paste it to StorageReview as well. There's a few HDD engineers that like to discuss stuff like this. I'd like to hear their opinion on the matter.
Bigfoots <3
I'd like to see one with fluid-dynamic bearings, proper areal density and 3 to 4 platters (half-height) and 3600rpm. They'd be awesome and record-breaking big capacities.
However, Bigfoot didn't exist in it's day to offer huge capacities. They basically existed to cut costs. Instead of 4+ platters, they only had 1 or 2. They cut costs in components that way. Their performance sucked (even though theoretically the large diameter should at least have offered high STR on outer tracks). Their reliability was so-and-so.
Most of the problems were attributed to firmware and lack of caching. Odd PIO4 connectivity issue, corruption, etc. earned them name "Bugfoot". I call them "Home Security Devices". If it's not screwed into any computer, you can lift it up and use it as a weapon against burglars. Bigfoot's unreliability and lack of performance are both probably a bit exagerated, though. There was some issues, that's for certain, and naturally it couldn't keep up with 5400rpm 3.5 inchers, but it wasn't outright horrible. It's just that something that stands out from the mass is guaranteed to either be crowned as a god... or bashed repeatedly by every nerd. What actually happens from PR point of view, may be just sheer luck or lack of luck.
Bigfoots could have been extremely high capacity drives back then. There's of course difficulties in keeping more platters spinning balanced but they could have done it with more expensive bearings. Instead, they chose to aim for value-segment, with low-profile (i.e typical 3.5" HDD height) and ultra-low-profile (half of CD-drive height). They're the thinnest of 5.25" HDDs. Other were so called "half-height" (CD-drive size) and "full-height" (2x CD-drive height).
As fun as it is to think about resurrection of Bigfoot, it's not going to happen. Sure, at a point it was almost certain that even just 5400rpm wouldn't happen. But thanks to WD being sneaky in their specs, they fooled some retailers to sell Greenpowers as 5400-7200rpm (or just as 7200rpm). A white lie was required to make retailers offer customers the possiblity to obtain quietness. Quietness wasn't considered to be a big feat back then. I guess not even now. Being "green" was and still is... and probably will be. However stating "5400rpm" was a definite suicide back then. Now it's less of a suicide as F2 Ecogreen sells pretty well. There was a F1 Ecogreen which also was openly 5400rpm (without WD-like BULLSHIT and hiding) but it never actually got any significant distribution in Finland. Even Google search returned nill, zilch, zit, zero. After GP, F2 started selling as was picked by many retailers.
To bring 3600rpm 5.25" to market... lol. I guess retro nerds might like one. Though half of them would puke at the very idea.
I'd like to see one with fluid-dynamic bearings, proper areal density and 3 to 4 platters (half-height) and 3600rpm. They'd be awesome and record-breaking big capacities.
However, Bigfoot didn't exist in it's day to offer huge capacities. They basically existed to cut costs. Instead of 4+ platters, they only had 1 or 2. They cut costs in components that way. Their performance sucked (even though theoretically the large diameter should at least have offered high STR on outer tracks). Their reliability was so-and-so.
Most of the problems were attributed to firmware and lack of caching. Odd PIO4 connectivity issue, corruption, etc. earned them name "Bugfoot". I call them "Home Security Devices". If it's not screwed into any computer, you can lift it up and use it as a weapon against burglars. Bigfoot's unreliability and lack of performance are both probably a bit exagerated, though. There was some issues, that's for certain, and naturally it couldn't keep up with 5400rpm 3.5 inchers, but it wasn't outright horrible. It's just that something that stands out from the mass is guaranteed to either be crowned as a god... or bashed repeatedly by every nerd. What actually happens from PR point of view, may be just sheer luck or lack of luck.
Bigfoots could have been extremely high capacity drives back then. There's of course difficulties in keeping more platters spinning balanced but they could have done it with more expensive bearings. Instead, they chose to aim for value-segment, with low-profile (i.e typical 3.5" HDD height) and ultra-low-profile (half of CD-drive height). They're the thinnest of 5.25" HDDs. Other were so called "half-height" (CD-drive size) and "full-height" (2x CD-drive height).
As fun as it is to think about resurrection of Bigfoot, it's not going to happen. Sure, at a point it was almost certain that even just 5400rpm wouldn't happen. But thanks to WD being sneaky in their specs, they fooled some retailers to sell Greenpowers as 5400-7200rpm (or just as 7200rpm). A white lie was required to make retailers offer customers the possiblity to obtain quietness. Quietness wasn't considered to be a big feat back then. I guess not even now. Being "green" was and still is... and probably will be. However stating "5400rpm" was a definite suicide back then. Now it's less of a suicide as F2 Ecogreen sells pretty well. There was a F1 Ecogreen which also was openly 5400rpm (without WD-like BULLSHIT and hiding) but it never actually got any significant distribution in Finland. Even Google search returned nill, zilch, zit, zero. After GP, F2 started selling as was picked by many retailers.
To bring 3600rpm 5.25" to market... lol. I guess retro nerds might like one. Though half of them would puke at the very idea.
I had few of them and installed/serviced hundreds of PC's with the Bigfoot drive. One of them is no match for the power of the .357 magnum. The bare aluminum cast makes for a great target. 
The one thing they did was use that second/third/fourth unused 5.25" drive bay that many tower cases come with. No rails to mess with - pop in and go. It was also the first drive that had the clear "do not remove" view port over the platter. Never quite figured out why though.
I've still got one somewhere - it's a TX version that is relatively quiet. When I first saw the drive come out the thing that popped into my head was the ATA version of the 5mb ST-506.
Larger overall size combined with lower spindle speed and higher areal density could create an efficient, low heat, no vibration drive. What else do you put in those empty 5.25" bays? One can only ponder the possibility.
The one thing they did was use that second/third/fourth unused 5.25" drive bay that many tower cases come with. No rails to mess with - pop in and go. It was also the first drive that had the clear "do not remove" view port over the platter. Never quite figured out why though.
I've still got one somewhere - it's a TX version that is relatively quiet. When I first saw the drive come out the thing that popped into my head was the ATA version of the 5mb ST-506.
Larger overall size combined with lower spindle speed and higher areal density could create an efficient, low heat, no vibration drive. What else do you put in those empty 5.25" bays? One can only ponder the possibility.
"It was also the first drive that had the clear "do not remove" view port over the platter. Never quite figured out why though."
To keep the dust out, obviously.
You meant the hole? It's needed for low-level formating the drive at factory. Old MFM and RLL HDDs needed to be low-level formated in the system. You could even decide what density you wanted to use in your HDD!
Modern HDD are low-level formated at the factory. This means you can't do the same trick on new drives and re-low-level format it to higher capacity like you can format 1.44MB floppy disk to 2+MB or stuff like that.
To write the servo tracks when no tracks exist, it is needed that an external machine moves the actuator arm. The robot moves the actuator arm through a hole in the cast and after low-level format is done, the hole is stickered. The sticker is usually silvery gray but I'm happy that some Bigfoots (though not all of them) used clear, see-through plastic. Looking through the window, you can see it's right above actuator arm... where the cast aluminun part of arm ends and the thin steel suspension begins... there should be a hole in the arm... this hole is where the robot's "arm" attaches itself to while low-level formating.
One can also move the actuator arm from above. Hitachi drives have a sticker on front cover. Seagate, Maxtor and Samsung have two-layered top covers which may mean that the hole is beneath the top layer (though some of them may have a hole in bottom cast instead).
Some old HDD might have been low-level formated before top cover was screwed on. I don't think this has been possible for years, as after low-level format is done, the actuator assembly must remain in exactly the same postition. And as the actuator arm has a screw through top cover... tightening it would screw up low-level format as tracks would misalign as actuator axis is shifted even a few micrometers, maybe even less. Some HDDs also have top screw to support platter assembly in addition to actuator assembly.
With 10 year old HDDs, you can pop the top cover off and you'd still probably be able to read the contents of the HDD... that is, until dust gets in and destroys the drive. The track width of those drives is way wider than those of modern drives so removing screw above actuator arm probably wouldn't cause immediate inoperability of the HDD.
"I've still got one somewhere - it's a TX version that is relatively quiet."
I have one Bigfoot BF (ultra-low-profile, 2.5GB) and one Bigfoot CY (low-profile, 4.3GB). Latter has the see-through peek window on bottom cast. Both are 3600rpm. According to Total Hardware 99 (TH99) BF and CY were 3600rpm but TS was 4000rpm. Eah. It was not 4200rpm.
Also, at that point in history there actually was 4500rpm drives like IBM Trailblazer and Seagate Medalist. So I guess I was kinda wrong by declaring it an odd capacity. 4500rpm was outphased before 4200rpm though, as laptop HDDs used usually 4200rpm.
But the era of odd rpm points is not yet over with 5900rpm Seagate Low-power. And those 7000rpm WDs. I dunno what rpm modern WDs spin but just a few years ago 7000rpm was a reality (despite them being listed as 7200rpm).
To keep the dust out, obviously.
You meant the hole? It's needed for low-level formating the drive at factory. Old MFM and RLL HDDs needed to be low-level formated in the system. You could even decide what density you wanted to use in your HDD!
Modern HDD are low-level formated at the factory. This means you can't do the same trick on new drives and re-low-level format it to higher capacity like you can format 1.44MB floppy disk to 2+MB or stuff like that.
To write the servo tracks when no tracks exist, it is needed that an external machine moves the actuator arm. The robot moves the actuator arm through a hole in the cast and after low-level format is done, the hole is stickered. The sticker is usually silvery gray but I'm happy that some Bigfoots (though not all of them) used clear, see-through plastic. Looking through the window, you can see it's right above actuator arm... where the cast aluminun part of arm ends and the thin steel suspension begins... there should be a hole in the arm... this hole is where the robot's "arm" attaches itself to while low-level formating.
One can also move the actuator arm from above. Hitachi drives have a sticker on front cover. Seagate, Maxtor and Samsung have two-layered top covers which may mean that the hole is beneath the top layer (though some of them may have a hole in bottom cast instead).
Some old HDD might have been low-level formated before top cover was screwed on. I don't think this has been possible for years, as after low-level format is done, the actuator assembly must remain in exactly the same postition. And as the actuator arm has a screw through top cover... tightening it would screw up low-level format as tracks would misalign as actuator axis is shifted even a few micrometers, maybe even less. Some HDDs also have top screw to support platter assembly in addition to actuator assembly.
With 10 year old HDDs, you can pop the top cover off and you'd still probably be able to read the contents of the HDD... that is, until dust gets in and destroys the drive. The track width of those drives is way wider than those of modern drives so removing screw above actuator arm probably wouldn't cause immediate inoperability of the HDD.
"I've still got one somewhere - it's a TX version that is relatively quiet."
I have one Bigfoot BF (ultra-low-profile, 2.5GB) and one Bigfoot CY (low-profile, 4.3GB). Latter has the see-through peek window on bottom cast. Both are 3600rpm. According to Total Hardware 99 (TH99) BF and CY were 3600rpm but TS was 4000rpm. Eah. It was not 4200rpm.
Also, at that point in history there actually was 4500rpm drives like IBM Trailblazer and Seagate Medalist. So I guess I was kinda wrong by declaring it an odd capacity. 4500rpm was outphased before 4200rpm though, as laptop HDDs used usually 4200rpm.
But the era of odd rpm points is not yet over with 5900rpm Seagate Low-power. And those 7000rpm WDs. I dunno what rpm modern WDs spin but just a few years ago 7000rpm was a reality (despite them being listed as 7200rpm).
Now that I think about it, when we disregard 4200rpm because it's used mostly just in laptop HDDs we have:
3600rpm
4500rpm
5400rpm
-
7200rpm
Note: exact 900rpm intervals with the exception that 6300rpm is missing.
_____
"Modern" range of rpms is like this:
4200rpm
5400rpm
7200rpm
10000rpm
15000rpm
And instead of being a constant jump to next rpm point, it looks rather exponential: 28.5%, 33.3%, 38.9%, 50%.
Should we add 3600rpm to the low-end, between 4200 and 3600 would be 16.7%. But I don't think 3600rpm really exist... I think the last time we saw that was in 1 inch form-factor IBM Microdrives. I haven't heard anything of them for ages... I guess they're gone by now. Should at least be, as in 1 inch size, it's much cheaper to just make it flash. One could also get more capacity out of it with solid state.
Heck, even 1.8 inch has been completely lost to SSD. The only 1.8 inch HDD I manage to find from retailers are old stock IDE. No-one bothers to manufacture 1.8 inch HDD anymore as SSD take over PDAs, mobile phones, etc.
I have a hunch 2.5 inch is the next to fall. It won't take many years. 3.5" will most likely be the last stronghold for mechanical mass storage... unless they flee back to 5.25" which they probably don't want to do unless they actually have to.
It may also be that when facing peril against SSD, HDD manufacturer's will again go on capacity frenzy. It's been a long time since HDD capacities climbed 100+ % every year. It's been increasingly difficult to do that but if facing external threat that cannot be fought against by cartel-like silent agreement, they might have to do something similar again (and possibly we might see 75GXP like horror stories in the process).
____
Also, this competition with SSDs might result in something good to silencers. Before SSD became a threat to HDD industry, all they did was compete each other: speed, capacity, speed, capacity. This mean that industry was slowly introducing new rpm point higher and higher, and while the technology matured and became cheaper, the high rpm points became mainstream.
Now, with competition with SSDs, they know they can never beat their speed. It would be futile. So, all they have is capacity and price. They may not be able to match 0 dB of SSD like they can't match <1ms random access time of SSD either but with 5400rpm (or even 4200rpm) they can get VERY close to observed silence... and with the help of any fan or coil buzz, become inaudible over background noise.
So, SSDs are not only hope to silencers due to their quietness, they also force HDD manufacturer to accept the reality that higher speed is unnecessary... especially when people start to install OSes on SSDs and use HDDs only for mp3s and movies. Then it wouldn't matter if HDD spun <3600rpm. I really couldn't care less. As long as it's able to send 2 MB/s, it shouldn't bottleneck HDTV experience.
I wouldn't want to boot my OS from one of those drives, though. I still remember how long it took to boot from Seagate U5 5400rpm... though it wasn't because rpm that it sucked. 3600rpm Bigfoots with their heavy actuators, lower rpm and longer seek distances had the same access time as U5 total-piece-of-shit. Luckily I used to boot Win98SE from it. 98 is light to boot. I wonder how many minutes it would take to boot XP or Vista from U5. 10 minutes?
But surely, if a 5.25" 3600rpm can match 5400rpm 3.5" drive in random access performance and acoustics, surely a <3000rpm 3.5" drive could also do it. And with modern areal densities they would be three times faster even with half the spindle speed compared to U5... not that I think 2700rpm drive would ever become reality.
Seagate U5 has ~21...22ms random access time. Latency is 5.5ms so seek is 15ms (compared to 8...10ms of most 3.5" regardless of rpm. For example, Samsung F2 is at 15ms, giving 9.5ms seek time.)
So, if we take 10ms seek time but slow the rpm until latency increases to a point where random access time is as bad as U5, a HDD with a fast actuator can achieve same performance at... 2730rpm. And have the triple STR bonus just for using modern density platters.
My point? For a data drive going 4200 or 3600 doesn't really matter. It's only when you need to boot from it that one might get pissed due to how low it's performance is... that, or trying to edit HDTV material using lossless video sources.
3600rpm
4500rpm
5400rpm
-
7200rpm
Note: exact 900rpm intervals with the exception that 6300rpm is missing.
_____
"Modern" range of rpms is like this:
4200rpm
5400rpm
7200rpm
10000rpm
15000rpm
And instead of being a constant jump to next rpm point, it looks rather exponential: 28.5%, 33.3%, 38.9%, 50%.
Should we add 3600rpm to the low-end, between 4200 and 3600 would be 16.7%. But I don't think 3600rpm really exist... I think the last time we saw that was in 1 inch form-factor IBM Microdrives. I haven't heard anything of them for ages... I guess they're gone by now. Should at least be, as in 1 inch size, it's much cheaper to just make it flash. One could also get more capacity out of it with solid state.
Heck, even 1.8 inch has been completely lost to SSD. The only 1.8 inch HDD I manage to find from retailers are old stock IDE. No-one bothers to manufacture 1.8 inch HDD anymore as SSD take over PDAs, mobile phones, etc.
I have a hunch 2.5 inch is the next to fall. It won't take many years. 3.5" will most likely be the last stronghold for mechanical mass storage... unless they flee back to 5.25" which they probably don't want to do unless they actually have to.
It may also be that when facing peril against SSD, HDD manufacturer's will again go on capacity frenzy. It's been a long time since HDD capacities climbed 100+ % every year. It's been increasingly difficult to do that but if facing external threat that cannot be fought against by cartel-like silent agreement, they might have to do something similar again (and possibly we might see 75GXP like horror stories in the process).
____
Also, this competition with SSDs might result in something good to silencers. Before SSD became a threat to HDD industry, all they did was compete each other: speed, capacity, speed, capacity. This mean that industry was slowly introducing new rpm point higher and higher, and while the technology matured and became cheaper, the high rpm points became mainstream.
Now, with competition with SSDs, they know they can never beat their speed. It would be futile. So, all they have is capacity and price. They may not be able to match 0 dB of SSD like they can't match <1ms random access time of SSD either but with 5400rpm (or even 4200rpm) they can get VERY close to observed silence... and with the help of any fan or coil buzz, become inaudible over background noise.
So, SSDs are not only hope to silencers due to their quietness, they also force HDD manufacturer to accept the reality that higher speed is unnecessary... especially when people start to install OSes on SSDs and use HDDs only for mp3s and movies. Then it wouldn't matter if HDD spun <3600rpm. I really couldn't care less. As long as it's able to send 2 MB/s, it shouldn't bottleneck HDTV experience.
I wouldn't want to boot my OS from one of those drives, though. I still remember how long it took to boot from Seagate U5 5400rpm... though it wasn't because rpm that it sucked. 3600rpm Bigfoots with their heavy actuators, lower rpm and longer seek distances had the same access time as U5 total-piece-of-shit. Luckily I used to boot Win98SE from it. 98 is light to boot. I wonder how many minutes it would take to boot XP or Vista from U5. 10 minutes?
But surely, if a 5.25" 3600rpm can match 5400rpm 3.5" drive in random access performance and acoustics, surely a <3000rpm 3.5" drive could also do it. And with modern areal densities they would be three times faster even with half the spindle speed compared to U5... not that I think 2700rpm drive would ever become reality.
Seagate U5 has ~21...22ms random access time. Latency is 5.5ms so seek is 15ms (compared to 8...10ms of most 3.5" regardless of rpm. For example, Samsung F2 is at 15ms, giving 9.5ms seek time.)
So, if we take 10ms seek time but slow the rpm until latency increases to a point where random access time is as bad as U5, a HDD with a fast actuator can achieve same performance at... 2730rpm. And have the triple STR bonus just for using modern density platters.
My point? For a data drive going 4200 or 3600 doesn't really matter. It's only when you need to boot from it that one might get pissed due to how low it's performance is... that, or trying to edit HDTV material using lossless video sources.
whiic i'm impressed,
not heard of bigfoot but why not make drives 5.25" platter size? why not seal these platters in a vaccuum?
i guess then 5400 rpm it is.
current platter is 500gb, how long will it take to double?
not heard of bigfoot but why not make drives 5.25" platter size? why not seal these platters in a vaccuum?
i guess then 5400 rpm it is.
current platter is 500gb, how long will it take to double?
Last edited by dan on Sun Aug 09, 2009 6:50 pm, edited 1 time in total.
To clarify - my point of the early "viewport" was a comparitive to the special Raptor drives of a couple of years ago.
Diminishing returns comes into play with drive castings and platter size. The margins are so thin I'd hate to be in the PC component industry at the moment - especially hard drives. It's difficult to differentiate your product and selling price is king.
Diminishing returns comes into play with drive castings and platter size. The margins are so thin I'd hate to be in the PC component industry at the moment - especially hard drives. It's difficult to differentiate your product and selling price is king.
-
jessekopelman
- Posts: 1406
- Joined: Tue Feb 13, 2007 7:28 pm
- Location: USA
That's why manufacturing capability trumps design. If you do innovate, you can get it to market faster/cheaper; and if you don't, you can get your copycat to market faster/cheaper.lemmy wrote:The margins are so thin I'd hate to be in the PC component industry at the moment - especially hard drives. It's difficult to differentiate your product and selling price is king.