Is it possible for HD platters to be remagnetized with a higher capacity?

[avi85]

Is it possible for HD platters to be remagnetized\remapped\etc.. with a higher capacity?
For example the HD in my sig is a 250GB model, but the 320GB of the same series has the same number of platters, just that the bits are closer together so technically shouldn't there be a way to increase the capacity?
It would be a great way to get more bang for your $$$ if this sort of thing were possible... kinda like oc'ing your HD...

 

[Mark R]

Maybe possible - but definitely not practical.

Even if the platters are made to the same specification (same magnetic coating, same thickness, etc.), the problem is the format.

You'd need to perform a low-level format on the platters to pack the sectors more closely. Unfortunately, it's impossible to low-level format a hard drive; the drive can't rewrite the low-level format, because the only way it has of positioning the heads is to track the low-level format.

The only practical way of doing it, is to remove the platters from the drive and reformat them on the manufacturer's low-level formatting machine. The modifiy the firmware on the drive after reassembling it.

 

[avi85]

Originally posted by: Mark R
Maybe possible - but definitely not practical.

Even if the platters are made to the same specification (same magnetic coating, same thickness, etc.), the problem is the format.

You'd need to perform a low-level format on the platters to pack the sectors more closely. Unfortunately, it's impossible to low-level format a hard drive; the drive can't rewrite the low-level format, because the only way it has of positioning the heads is to track the low-level format.

The only practical way of doing it, is to remove the platters from the drive and reformat them on the manufacturer's low-level formatting machine. The modifiy the firmware on the drive after reassembling it.

So, if a way were found to manipulate a drive to perform a low level format then it might be possible?

 

[johnpombrio]

Interesting question! Bottom line is no. If the drive HAD the capacity for more storage, the manufacturer would put it in! After all, it would not cost them any more. The limit is set by the read/write head and the firmware on the drive. The drive itself is servo coded by the R/W heads at the factory. But if the firmware says there are 20,000 sectors, then you better have 20,000 sectors.

 

[SuperFungus]

Wouldn't you also need higher resolution heads than what comes with your harddrive or are those standard across all drives?

 

[patentman]

In short, no.

Hard disks are shipped with a magnetic head that is calibrated to read media having a certain bit density. Could that be changed with a firmware update? Possibly, but generally the heads are designed to work with maximum accuracy at the density at which the platter is designed. Thus, it is theoreticaly possible to increase the read/write resolution of a magnetic head, but you would likely run into ll sorts of read/write errors and corrupted data. In addition, there are physical limitations that come into play which limit the resolution of the head, i.e. the soft magnetic materials used.

As for increasing the density of the platter itself, that depends on two things. First there is the coercivity of the magentic layer, which is essentially the strength at which magnetic domains can be oriented in the platter. Coercivity is material dependant, and cannot be changed without altering the alloy used to make the magnetic recording layer. Second is signal to noise ratio, which must be higher than that which the magentic head can discern (the head has to be able to discern the signal from the noise). Signal to noise ratio is also highly dependant on the materials used for the magnetic recording layer. Bit density and signal to noise ratio are usually inversely related, as bit density increases, signal to noise ratio decreases. this is due to the fact that as bit density increases, an effect called intergranular exchange coupling comes into play. I've written about this effect a number of times already on this forum (search for posts under the sig "klaviernista" in the archives). Essentially, interganular exchange coupling is an effect wherein the magnetic field of a grain within the media is interfered with by the magnetic fields of surrounding grains. As a result, the magentic field of each grain in the media is destabilized, which results in a corresponding decrease in signal to noise ratio.

The magnetic recording layers of a hard disk are generally designed so as to minimize the effects of intergranular exchange coupling while at the same time allowing for substantial recording density. There are a number of things that can be done to reduce intergranular exchange coupling (e.g. material selection, underlayer selection, the use of pinning layers, the addition of alloying elements which segregate out into the grain boundaries etc..), but currently all of these techniques are done while the magnetic layer is being deposited on the platter.

Interesting question, but the answer is a resounding no.

See my posts (klaviernista) under these threads for more info:

http://forums.anandtech.com/messageview...hreadid=1338904&enterthread=y&arctab=y

http://forums.anandtech.com/messageview...hreadid=1462426&enterthread=y&arctab=y

http://forums.anandtech.com/messageview...hreadid=1444815&enterthread=y&arctab=y

THIS ONE IN PARTICULAR IS PERTINENT:
http://forums.anandtech.com/messageview...hreadid=1422946&enterthread=y&arctab=y

 

[avi85]

Originally posted by: patentman
In short, no.

Hard disks are shipped with a magnetic head that is calibrated to read media having a certain bit density. Could that be changed with a firmware update? Possibly, but generally the heads are designed to work with maximum accuracy at the density at which the platter is designed. Thus, it is theoreticaly possible to increase the read/write resolution of a magnetic head, but you would likely run into ll sorts of read/write errors and corrupted data. In addition, there are physical limitations that come into play which limit the resolution of the head, i.e. the soft magnetic materials used.

As for increasing the density of the platter itself, that depends on two things. First there is the coercivity of the magentic layer, which is essentially the strength at which magnetic domains can be oriented in the platter. Coercivity is material dependant, and cannot be changed without altering the alloy used to make the magnetic recording layer. Second is signal to noise ratio, which must be higher than that which the magentic head can discern (the head has to be able to discern the signal from the noise). Signal to noise ratio is also highly dependant on the materials used for the magnetic recording layer. Bit density and signal to noise ratio are usually inversely related, as bit density increases, signal to noise ratio decreases. this is due to the fact that as bit density increases, an effect called intergranular exchange coupling comes into play. I've written about this effect a number of times already on this forum (search for posts under the sig "klaviernista" in the archives). Essentially, interganular exchange coupling is an effect wherein the magnetic field of a grain within the media is interfered with by the magnetic fields of surrounding grains. As a result, the magentic field of each grain in the media is destabilized, which results in a corresponding decrease in signal to noise ratio.

The magnetic recording layers of a hard disk are generally designed so as to minimize the effects of intergranular exchange coupling while at the same time allowing for substantial recording density. There are a number of things that can be done to reduce intergranular exchange coupling (e.g. material selection, underlayer selection, the use of pinning layers, the addition of alloying elements which segregate out into the grain boundaries etc..), but currently all of these techniques are done while the magnetic layer is being deposited on the platter.

Interesting question, but the answer is a resounding no.

See my posts (klaviernista) under these threads for more info:

http://forums.anandtech.com/messageview...hreadid=1338904&enterthread=y&arctab=y

http://forums.anandtech.com/messageview...hreadid=1462426&enterthread=y&arctab=y

http://forums.anandtech.com/messageview...hreadid=1444815&enterthread=y&arctab=y

THIS ONE IN PARTICULAR IS PERTINENT:
http://forums.anandtech.com/messageview...hreadid=1422946&enterthread=y&arctab=y

WOW! thanks for all the info, if you know all this stuff about HDs, I've always wondered why they don't make a 5.25" HD (these would mainly be drives for storage, not the OS ), at today's densities it could probably hold a couple of terabytes (at least), I know that it probably couldn't spin at 7200rpm cause of the noise it would make (right?) but 2 of them in raid at 5400RPM would have decent throughput (on the outer part it would probably be faster than a 3.5"@7200) or another solution to the speed problem could be: (and this brings us to another question) is there such a thing as raid built into a single HD? while the heads are on 2 (or more) different platters but in the same position they could double (or triple or quadruple etc...) the speed by both reading/writing on there respective platters instead of one reading/writing and the other one just sitting there. (sort of an integrated raid 0 solution), now if 2 or more of these (integrated raid 0) drives were put into raid 0 the HD would probably cease to be a bottleneck (to some extent) as we know it...

 

[patentman]

Originally posted by: avi85
WOW! thanks for all the info, if you know all this stuff about HDs, I've always wondered why they don't make a 5.25" HD (these would mainly be drives for storage, not the OS ), at today's densities it could probably hold a couple of terabytes (at least), I know that it probably couldn't spin at 7200rpm cause of the noise it would make (right?) but 2 of them in raid at 5400RPM would have decent throughput (on the outer part it would probably be faster than a 3.5"@7200) or another solution to the speed problem could be: (and this brings us to another question) is there such a thing as raid built into a single HD? while the heads are on 2 (or more) different platters but in the same position they could double (or triple or quadruple etc...) the speed by both reading/writing on there respective platters instead of one reading/writing and the other one just sitting there. (sort of an integrated raid 0 solution), now if 2 or more of these (integrated raid 0) drives were put into raid 0 the HD would probably cease to be a bottleneck (to some extent) as we know it...

No problem, thats why my sig is what it is. As for making a 5.25" hard drive, that has to with the cost of making: a) a large platter flat; and b) making a large platter strong enough to withstand the forces exerted on it when spun at a realtively high rate. As to a), you are probably aware that in HDD's, the read/write head in a hard disk floats on a cushion of air a very small distance away from the platter. As a result, the weight distribution, surface roughness and microwaviness (this is the real term, I didn;t make it up) of the platter have to be tightly controlled so as to prevent wobbling while spinning at thousands of RPM. This is important for two reasons. First, less wobbling means that the head is less likely to crash into the surface of the platter, and second, less wobbling means less noise, because the distance between the head and the platter remains relatively constant (remember, magnetic field intensity decreases with increases in distance). However, the larger a platter is, the harder it is to control the weight distribution, surface roughness and microwaviness over the entire platter. In fact, in may cost a manufacturer more to make one 5.25" platter than several 3.5" platters.

As for having an integrated raid 0 setup, that is an interesting question. I can't say as I have ever seen a HDD advertised as being set up this way, but I don't see why that couldn't be done. Probably would just take a firmware update and require a raid driver. As for an increase in performance/removing bottlenecking, I doubt that you would see the kind of increases that you are stating. Anand wrote a good article a while back that basically affirmed my position that having a raid setup doesn't necessarily increase performance. The link to that writeup is below.

http://www.anandtech.com/showdoc.aspx?i=2101&p=10

As for having an internal Raid 1 setup, that kind of defeats the point of raid 1, but I gather you knew that seeing as how you didn't suggest it.

 

[patentman]

Originally posted by: SuperFungus
Wouldn't you also need higher resolution heads than what comes with your harddrive or are those standard across all drives?

You would need a higher resolution head, and I can assure you that read/write heads are not standardzed across all hard drives. Indeed, they aren't even standardized within the same product line.

 

[avi85]

Originally posted by: patentman

Originally posted by: avi85
WOW! thanks for all the info, if you know all this stuff about HDs, I've always wondered why they don't make a 5.25" HD (these would mainly be drives for storage, not the OS ), at today's densities it could probably hold a couple of terabytes (at least), I know that it probably couldn't spin at 7200rpm cause of the noise it would make (right?) but 2 of them in raid at 5400RPM would have decent throughput (on the outer part it would probably be faster than a 3.5"@7200) or another solution to the speed problem could be: (and this brings us to another question) is there such a thing as raid built into a single HD? while the heads are on 2 (or more) different platters but in the same position they could double (or triple or quadruple etc...) the speed by both reading/writing on there respective platters instead of one reading/writing and the other one just sitting there. (sort of an integrated raid 0 solution), now if 2 or more of these (integrated raid 0) drives were put into raid 0 the HD would probably cease to be a bottleneck (to some extent) as we know it...

No problem, thats why my sig is what it is. As for making a 5.25" hard drive, that has to with the cost of making: a) a large platter flat; and b) making a large platter strong enough to withstand the forces exerted on it when spun at a realtively high rate. As to a), you are probably aware that in HDD's, the read/write head in a hard disk floats on a cushion of air a very small distance away from the platter. As a result, the weight distribution, surface roughness and microwaviness (this is the real term, I didn;t make it up) of the platter have to be tightly controlled so as to prevent wobbling while spinning at thousands of RPM. This is important for two reasons. First, less wobbling means that the head is less likely to crash into the surface of the platter, and second, less wobbling means less noise, because the distance between the head and the platter remains relatively constant (remember, magnetic field intensity decreases with increases in distance). However, the larger a platter is, the harder it is to control the weight distribution, surface roughness and microwaviness over the entire platter. In fact, in may cost a manufacturer more to make one 5.25" platter than several 3.5" platters.

As for having an integrated raid 0 setup, that is an interesting question. I can't say as I have ever seen a HDD advertised as being set up this way, but I don't see why that couldn't be done. Probably would just take a firmware update and require a raid driver. As for an increase in performance/removing bottlenecking, I doubt that you would see the kind of increases that you are stating. Anand wrote a good article a while back that basically affirmed my position that having a raid setup doesn't necessarily increase performance. The link to that writeup is below.

http://www.anandtech.com/showdoc.aspx?i=2101&p=10

As for having an internal Raid 1 setup, that kind of defeats the point of raid 1, but I gather you knew that seeing as how you didn't suggest it.

If raiding (hey look at that, I turned it into a verb ) 2 hd's doesn't have a real-world effect on performance then the bottleneck is the seek time. right? so for the best HD performance I'd be best off getting a 15k.5 cheetah?

 

[patentman]

Originally posted by: avi85
If raiding (hey look at that, I turned it into a verb ) 2 hd's doesn't have a real-world effect on performance then the bottleneck is the seek time. right? so for the best HD performance I'd be best off getting a 15k.5 cheetah?

Probably yes, although you might want to get some ear protection, the cheetah is LOUD

 

[Ruptga]

So if seek time is the biggest problem (which doesn't surprise me), what are your thoughts on hybrid drives. Even with industrial strength flash, do you expect the flash to be too wimpy and simply die after three years, etc.

 

[kpb]

Originally posted by: avi85

Originally posted by: patentman
In short, no.

Hard disks are shipped with a magnetic head that is calibrated to read media having a certain bit density. Could that be changed with a firmware update? Possibly, but generally the heads are designed to work with maximum accuracy at the density at which the platter is designed. Thus, it is theoreticaly possible to increase the read/write resolution of a magnetic head, but you would likely run into ll sorts of read/write errors and corrupted data. In addition, there are physical limitations that come into play which limit the resolution of the head, i.e. the soft magnetic materials used.

As for increasing the density of the platter itself, that depends on two things. First there is the coercivity of the magentic layer, which is essentially the strength at which magnetic domains can be oriented in the platter. Coercivity is material dependant, and cannot be changed without altering the alloy used to make the magnetic recording layer. Second is signal to noise ratio, which must be higher than that which the magentic head can discern (the head has to be able to discern the signal from the noise). Signal to noise ratio is also highly dependant on the materials used for the magnetic recording layer. Bit density and signal to noise ratio are usually inversely related, as bit density increases, signal to noise ratio decreases. this is due to the fact that as bit density increases, an effect called intergranular exchange coupling comes into play. I've written about this effect a number of times already on this forum (search for posts under the sig "klaviernista" in the archives). Essentially, interganular exchange coupling is an effect wherein the magnetic field of a grain within the media is interfered with by the magnetic fields of surrounding grains. As a result, the magentic field of each grain in the media is destabilized, which results in a corresponding decrease in signal to noise ratio.

The magnetic recording layers of a hard disk are generally designed so as to minimize the effects of intergranular exchange coupling while at the same time allowing for substantial recording density. There are a number of things that can be done to reduce intergranular exchange coupling (e.g. material selection, underlayer selection, the use of pinning layers, the addition of alloying elements which segregate out into the grain boundaries etc..), but currently all of these techniques are done while the magnetic layer is being deposited on the platter.

Interesting question, but the answer is a resounding no.

See my posts (klaviernista) under these threads for more info:

http://forums.anandtech.com/messageview...hreadid=1338904&enterthread=y&arctab=y

http://forums.anandtech.com/messageview...hreadid=1462426&enterthread=y&arctab=y

http://forums.anandtech.com/messageview...hreadid=1444815&enterthread=y&arctab=y

THIS ONE IN PARTICULAR IS PERTINENT:
http://forums.anandtech.com/messageview...hreadid=1422946&enterthread=y&arctab=y

WOW! thanks for all the info, if you know all this stuff about HDs, I've always wondered why they don't make a 5.25" HD (these would mainly be drives for storage, not the OS ), at today's densities it could probably hold a couple of terabytes (at least), I know that it probably couldn't spin at 7200rpm cause of the noise it would make (right?) but 2 of them in raid at 5400RPM would have decent throughput (on the outer part it would probably be faster than a 3.5"@7200) or another solution to the speed problem could be: (and this brings us to another question) is there such a thing as raid built into a single HD? while the heads are on 2 (or more) different platters but in the same position they could double (or triple or quadruple etc...) the speed by both reading/writing on there respective platters instead of one reading/writing and the other one just sitting there. (sort of an integrated raid 0 solution), now if 2 or more of these (integrated raid 0) drives were put into raid 0 the HD would probably cease to be a bottleneck (to some extent) as we know it...

I beleive it was seagate who relatively rescently tried to do 5 1/4 hds again with thier bigfoot line. As far as i know they performed pretty poorly, largely due to slow seek time from the large disk, and didn't off that much more capacity at the price point. As far as i know they stopped producing them after a year or 2 of trying.

I beleive seagate tried a single drive raid again in an old baracuda drive. The added complexity apparently add much more cost than the increase in performance. I beleive they only had one generation of baracuda's that did that and the next generate was back to normal and still faster.

 

[krotchy]

Originally posted by: johnpombrio
Interesting question! Bottom line is no. If the drive HAD the capacity for more storage, the manufacturer would put it in! After all, it would not cost them any more. The limit is set by the read/write head and the firmware on the drive. The drive itself is servo coded by the R/W heads at the factory. But if the firmware says there are 20,000 sectors, then you better have 20,000 sectors.

This logic, while it should be true is kind of flawed. In fact manufacturers can make a lot more money by taking a full featured device and disabling parts and selling things in tiers.

This can be seen fairly often in video cards and processors. I dont remember the exact model, but I think the 6800GT could be unlocked to a 6800 Ultra, since the GT was just an Ultra with some pipes turned off.

This is also done often with cellular phones. A manufacturers might release 3 phone models in 3 different cases. The lowest end one might have blue-tooth disabled, even if it is inside the phone. The mid tier has something else disable and the top tier is fully functional. Even though the three phones all have the same features internally, it is almost always cheaper for the manufacturer to make 3 million of one PCA than to make 1 million of 3 different ones.

 

[patentman]

Originally posted by: kpb
I beleive it was seagate who relatively rescently tried to do 5 1/4 hds again with thier bigfoot line. As far as i know they performed pretty poorly, largely due to slow seek time from the large disk, and didn't off that much more capacity at the price point. As far as i know they stopped producing them after a year or 2 of trying.

Seagate did put out a bigfoot line of 5.25 hard drives, but I wouldn't call that effort "recent." As I recall, the bigofoots were popular 8-10 years ago, which is an eternity for hard drive technology.

 

[dkozloski]

Originally posted by: patentman

Originally posted by: kpb
I beleive it was seagate who relatively rescently tried to do 5 1/4 hds again with thier bigfoot line. As far as i know they performed pretty poorly, largely due to slow seek time from the large disk, and didn't off that much more capacity at the price point. As far as i know they stopped producing them after a year or 2 of trying.

Seagate did put out a bigfoot line of 5.25 hard drives, but I wouldn't call that effort "recent." As I recall, the bigofoots were popular 8-10 years ago, which is an eternity for hard drive technology.

Quantum put out the Bigfoot drive and it didn't become Seagate until the Seagate/Quantum/Maxtor merger. Bigfoot was the start of the downfall of Quantum.

 

[Matthias99]

Originally posted by: krotchy

Originally posted by: johnpombrio
Interesting question! Bottom line is no. If the drive HAD the capacity for more storage, the manufacturer would put it in! After all, it would not cost them any more. The limit is set by the read/write head and the firmware on the drive. The drive itself is servo coded by the R/W heads at the factory. But if the firmware says there are 20,000 sectors, then you better have 20,000 sectors.

This logic, while it should be true is kind of flawed. In fact manufacturers can make a lot more money by taking a full featured device and disabling parts and selling things in tiers.

This can be seen fairly often in video cards and processors. I dont remember the exact model, but I think the 6800GT could be unlocked to a 6800 Ultra, since the GT was just an Ultra with some pipes turned off.

This is also done often with cellular phones. A manufacturers might release 3 phone models in 3 different cases. The lowest end one might have blue-tooth disabled, even if it is inside the phone. The mid tier has something else disable and the top tier is fully functional. Even though the three phones all have the same features internally, it is almost always cheaper for the manufacturer to make 3 million of one PCA than to make 1 million of 3 different ones.

Well... sometimes.

Those video cards you mentioned? At first the cheaper GeForce6 models were cut-down versions of the high-end cards (and so could be modded into the faster version by changing the BIOS). Sometimes this was done because the GPUs actually had problems in one of the pipelines and couldn't be used as the high-end model, but other cards were disabled simply because there was more demand for the less expensive version. But then later on NVIDIA started making a version of that GPU that had fewer hardware pipelines (so they could get better yields and more chips per wafer, since there were fewer transistors in the GPU). Clearly they thought it was more economical to actually make two different versions than to make the big, low-yield chips and use them for everything.

In terms of your cellphone example... it might be the same 'phone' and PCB, but they might also do things like not putting in the bluetooth radio hardware for the low-end model (assuming it actually costs some non-trivial amount of money). Basically, if there's a big price difference to provide the hardware functionality that differentiates multiple products, the manufacturer will likely develop separate hardware for each product.

I don't know enough about the low-level details of HD technology to say whether modern platters are 'binned' the same way as memory or processor chips, or if there are different platter specifications used for different densities of drive within the same product line. I work on things that use hard drives; I don't make 'em.

 

[avi85]

Originally posted by: patentman

Originally posted by: kpb
I beleive it was seagate who relatively rescently tried to do 5 1/4 hds again with thier bigfoot line. As far as i know they performed pretty poorly, largely due to slow seek time from the large disk, and didn't off that much more capacity at the price point. As far as i know they stopped producing them after a year or 2 of trying.

Seagate did put out a bigfoot line of 5.25 hard drives, but I wouldn't call that effort "recent." As I recall, the bigofoots were popular 8-10 years ago, which is an eternity for hard drive technology.

When SSD's come out, won't 5.25" HD's be something to consider again, cause it will be able to hold a huge amount of memory chips, and all the other problems will be moot. See here and notice how the capacity increases with the size, if a 2.5" HD can hold 128 GB then a 5.25" drive can hold way more, cause it's like 10 times as big(if you factor in the height too).

 

[bobsmith1492]

Originally posted by: avi85

Originally posted by: patentman

Originally posted by: kpb
I beleive it was seagate who relatively rescently tried to do 5 1/4 hds again with thier bigfoot line. As far as i know they performed pretty poorly, largely due to slow seek time from the large disk, and didn't off that much more capacity at the price point. As far as i know they stopped producing them after a year or 2 of trying.

Seagate did put out a bigfoot line of 5.25 hard drives, but I wouldn't call that effort "recent." As I recall, the bigofoots were popular 8-10 years ago, which is an eternity for hard drive technology.

When SSD's come out, won't 5.25" HD's be something to consider again, cause it will be able to hold a huge amount of memory chips, and all the other problems will be moot. See here and notice how the capacity increases with the size, if a 2.5" HD can hold 128 GB then a 5.25" drive can hold way more, cause it's like 10 times as big(if you factor in the height too).

FLASH is extremely small, though; I don't think it would it need more space except maybe for cooling concerns, though they say the use hardly any power.