The disk drive design no one's ever made...

[Luposian]

Anyone here interested in seeing if we could come up with a way to create a disk drive (even just a prototype that proves a few concepts) that defies the standard that ALL disk drives have been based around since the beginning?

In all drives, the head floats over the platters, which spin at 4,500 - 15,000RPM. And, the reason they need to spin so fast, is so that the bits of data that the disk is trying to read/write are under the head at any given moment. This reduces rotational latency but turns up the HEAT!

But... what if... the data was ALWAYS in the same place? Not X# of times per second, but ALWAYS.

What if you could reduce drive heat to near nothing?

What if accessing data was only a matter of how fast you could move the head there?

These and other ideas are what generated this thread...

[keith]

This is kind of interesting, but we need the technology if we are to come up with a new design of hard disk drive :P

[Luposian]

Funny thing is... it's existed since the 70's!

Let's assume the magnetic media in a hard drive was a flat rectanguar "plate" (let's say it's single-sided, for simplicity sake). It never moves. How would you move the head to access (read/write) any area of the magnetic media?

Since the media never moves, the data at any one place is always there. Thus, there is no "rotational latency". There is no "spin up delay" (when you turn on your computer). And, no heat from spinning platters. The only "latency" (seek and track to track) that exists is "how fast you can get the head to that location?".

[TechnoMancer]

Have you tought of the fact that the head has to be moved over the medium to read the data, the rotating disk accomplishes this and now the rotational latency is probably almost insignificant because the circuit can only read the data at a certain rate anyway. There are also hybrid drives that use flash memory and hard drive components the disk stores stuff in a flash memory buffer which has the high speed, low latency, low temperature and othe radvatages and the disk is only used when data is read which is not in the buffer or when the buffer is full. These drives are pretty efficient! Your design would have higer seek times than rotational drives and would be ineffiecient as when you got to the end of the plate you would have to move down and start going the other way or go back to the other side and move down this is less effiecient than a disk shaped drive also you would still have to have a contant movement to read the data and trying to reread data that is already passed under the head would take far more time than the rotational latency of a conventional drive!!:cool:

[Luposian]

If the data the computer wants can be grabbed by the hard drive, in one pass, then that's EXCELLENT efficiency. But... the platter is constantly spinning. If it's not able to grab all of the data it needs, it needs to wait for the data to come by again... and again... and again. Until it gets it all. But because the data is constantly moving, the head has to wait until it gets it all and then move on to the next bit and so on...

That's why you need to spin the platters faster (up to 15,000RPM). That's the only way to compensate for this "inefficient design".

In my scenario, the data never moves. You move the head there and grab all the data you need and move on. Once.

Think about it this way... which is easier to grab... an apple on a spinning plate or an apple sitting on a plate, that isn't moving at all?

A "data plate", the size/dimensions of a 3.5" HD, also has more surface area, thus more data on it, than a 3.5" platter. The circular design of a HD platter shaves off data "in the corners".

If you could get a full stroke (from 0,0 to (say) 255,255; which is a diagonal shot... and thus longer than a full stroke in either the X or Y axis alone) down to a quarter of a second or less... I think the data takeup rate would easily compensate for the slower access rate.

The motors required to move the head wouldn't be large and they wouldn't take many volts/watts to power. There is a lot of (relative) power required to get a hard drive up to full speed. If the spindle seizes or the motor dies... your data ain't going anywhere. And all that spinning generates a lot of heat... and heat KILLS electronics!

And, something else to consider, you never have to worry about crashing the heads... EVER. Because they'd be permantly suspended above the plate at exactly the right distance... all the time. You could flick the power on and off a dozen times and the heads would just sit there, suspended. There is no "cushion of air" keeping the heads afloat. There is no "parking spot" necessary. Because the heads can NEVER actually touch the plate! EVER. Bump it. Drop it. While it's running! Who cares! There is no air flow to disrupt!

And, if the head's "home position" was always in the middle of the disk plate, accessing any data from any direction is that much closer! And, because of the head suspension design, you can do that!

I just wish there were some people willing to test out this design theory of mine to see just how good (or bad) it really is.

Hmm?

[TechnoMancer]

but the if you flick the power off and on the danger is prom the em pulse from the head. it is this that causes data corruption not only the head touching the disk!

also if the head does not move fast enough over the disk surface the head will not be able to detect the changes in magnetic field correctly! and if you want to change the data you would have to move the head back and pass over the disk again in your design!

another thing your drive could not be called a disk drive because it has no disks only a rectangular plate!!

by the way just the right height = so close to the disk surface that if theapparatus is bumped it might touch the disk also dropping your drive would still result in data loss as the jolt to the plat will cause demagnatization!!!

[Luposian]

but the if you flick the power off and on the danger is prom the em pulse from the head. it is this that causes data corruption not only the head touching the disk!
In my design, the drive doesn't apply power to the head(s) until after the BIOS is done booting. Flicking the power on and off would have no consequence, as the BIOS has to first ask for the boot sector, for the drive to actually power on the heads and start reading data.

also if the head does not move fast enough over the disk surface the head will not be able to detect the changes in magnetic field correctly! and if you want to change the data you would have to move the head back and pass over the disk again in your design!
If this is true, you may have found my concept's flaw. But my assumption was that data was a specific arrangement of magnetic particles and formatted (empty) space was all "organized" magnetic particles. The drive only reads from "arranged particles" and writes to "organized" ones.

Hmm?

another thing your drive could not be called a disk drive because it has no disks only a rectangular plate!!
This is true... we'll call it a XYPT Drive then! :-)

by the way just the right height = so close to the disk surface that if theapparatus is bumped it might touch the disk also dropping your drive would still result in data loss as the jolt to the plat will cause demagnatization!!!
The way the head is suspended over the "plate", is (theoretically, since it doesn't exist yet)) so rigid, it couldn't hit the plate. And how does a jolt to the plate cause demagnatization? Does dropping a magnet demagnatize it? Does dropping a floppy disk suddenly make it unreadable? I don't understand...

[TechnoMancer]

if u are reading the disk then if you flick the power off there may be a voltage surge and this will cause corruption of data

[Luposian]

if u are reading the disk then if you flick the power off there may be a voltage surge and this will cause corruption of data
This could happen with a regular Hard Drive, as well, so it's not specifically applicable as a "flaw" in my particular design. Thus, your argument is invalid.

[keith]

I have merged some of the posts together, as I feel that it improves readability. If there is a need to add comments, edit your post if it's still the last post of the thread, rather than posting new posts.