My first drive was a 5mb winchester (named after the IBM system 3030, a new sealed mechanism as opposed to the removeable platter types)

It cost me 3,800 in 1980 dollars

At a fixed 17 sector per track and 3600 rpm, it layed out the magnetic regions as n/s pairs stuck end to end, and untill the tech pointed out by this article becomes common, all the disks on the shelf today are the same, but there have been changes !

The media in that first disk was painted on, the paint would flake off if the head touched the disk at speed, and the resulting pothole would often tear the head off the disk the next time around wen it bounced enough to dig into the aluminum underneath.

Today, the material is a vacuum deposit of energetic hot gasified metal that merges with the platter when it slams into it, on top of this they layer on a coat of synthetic diamond.

To say this is more sturdy, beside the thiner coat easer to magnetize faster are weak word to describe the difference.

All of todays disks now use this kind of media

(And the synthetic deposition of diamond is the next revolution to diamond based, instead of silicon, elecronics, and a 20X speed boost to tomorrows CPUs, be on the lookout for that, OK?)

Larger disks simply used more platters and more heads, well,,, untill RLL came out.

The first disks used MFM, Modified Frequency Modulation, meaning around the disk there was a clock bit inbetween every data bit.

RLL, Run Length Limited, disks was based on the idea of using the data as it own clock, a clock bit is only inserted if both the preceeding and trailing data bits are 0

(This is the same scheme that gave us double density floppies)

With less bits total using this scheme, the sectors (blocks of data) became smaller and we saw a jump in the sectors per track.

Still there was more to get, with fixed sectors per track, this means the bits are packed more tightly near the spindle, where a circle is less surface area, than at the rim, where the surface is closer to the distance around the outside edge.

If you push the bit rate up and bits overlap near the spindle, you cant go in as far, but it means at the rim of the disk capacity is wasted.

Enter ZBR, Zone Bit Recording

Here they changed the clock data rate at the head depending on how far in the head was.

At the outside, the clock rate is highest, to make use of all the maximum linear space per revolution you have there.

As the head steps inward, the controler steps down the data rate on the head, and record fewer sectors per track, so that the physical bit density around a track was held to a near constant.

All disks use this method today.

That still isnt all, enter the IBM Magnito resistive head.

Up to now, the read/write head was little different than the old reel to real tape decks, they had gone to ceramic substrates with deposited magnetic core material, but it was still an electromagnet with the gap flying over the disk.

A magnetic coil includes an inherent frequency limit, reduce the turns to go faster, and you lose coupling to the core, so the head was now the speed Problem to upping the data rate, (so you can get more bits in a given space)

the IBM Mr head dont use a coil for reading, but a resistor material that changes value when affected by a mag field, it could go faster (hardly any speed limit) than the old coil, so the read speed problem was solved, for writing they still used the old gap, but they simply tolerated the loss in a one turn coil (the write coil is now just a write coil, it cant read its own bits, and used a lot more current)

All of todays disks use IBM MR heads today, and untill now it was the latest innovation.

There was a lot of talk about how they could record by standing the magnets up on the disk, to get a tighter packing, and they might yet still ( to get more tracks on the disk, as well as more bits on a track) but the problems to overcome that still seem difficult to say the least.

But turning them sideways is just as neat, so just when you think they cant possibly take another leap, surprize !

I hope you found this interesting.

Raymond, I do believe we've found your real specialty & focus.

To those that say this is indeed verticle recording, well no, both poles of every bit are still seen at the disk surface, so it is still double the linear space per bit of true verticle recording.

But its still a win. two verticle poles per bit are still less space than the old string of bar magnet shaped bits laying flat method.

GaH

Its true verticle recording ! OMG.

The secret is the Geometry of the write head.

On one side of the gap, the pole is a knife edge.

(perhaps a thiner edge than you could ever put on a real knife)

The mag field is concentrated at that sharp knife edge, and thats where the bit is created.

The other side of the gap, the pole is deliberatly spread over a large area, covering several bits to weaken the magnetic concentration, allowing it to pass the opposite pole of the field past them without flipping them.

Gah ! they have solved the problem, hit be with a brick, whip me take my charge card.

I worked on a perpendicular recording hard drive way back in the late 80's. It was going to be the next big thing then, too. We did not succeed.

The main problem, as I understood it, was the return path (the big, deliberately spread out pole). While you can reduce the write field strength quite a bit by increasing the area, you are still passing a lot of flux through media you don't want to change. We had big problems with 'self erasing' by the return pole.

I think there was also an analogous read problem. With an inductive head you are really reading a superposition of the one strong bit under the little pole along with and all the weaker bits spread out under the big pole. For that matter, you start seeing low-frequency patterns from the deep layers of the media that you don't want. It was a big source of noise. I'm not sure how this would manifest itself with magneto-resistive read heads.

I hope they have really solved all these problems.

ArtD0dger

In this case, it dont matter, the media direction is so it passes under the knife edge last, and sectors are written whole block at a time.

The magnito resisive head is inert magnetically during reading, and so the noise you get from the other pole isnt a factor.

The Mag-resistive bit sensor part of the head is narrow enough to only see one bit at a time, in fact the metalised ecapuslation around it is a sheld so that the active material is only exposed to one bit at a time

Taken all together, it indeed looks like they have solved all these issues, and Yes!,, Vertical recording has been the holy grail for a long time.

This is another large jump, both in capacity, and data tranfer rates.

Raymond - I'm not sure I buy that the knife edge always passes last. That would mean that (a) the return pole is a skinny thing that rides only along the current track in front of the knife edge, or (b) the return pole is offset to the side so it hits only tracks 'yet to be written.' I don't think (a) would work very well, and (b) would have nasty ramifications for random access performance (multi-track sectors...). I think it's more likely that they have just perfected the media so that it is consistently immune to the low flux, yet consistently susceptible to the high flux. This is actually very hard -- we were always fighting variations in coating thickness, susceptibility, and regions where the domains would tilt away from perpendicular in one direction or another.

I could believe that MR heads solve the read problems, though. Like I said, I hope they really have solved all these really hard engineering problems.

It will sink in if you continue looking, as you can see above, it really took me a bit to overcome my own resistance to what i was looking at .

This has been an impossible problem for so long im afraid we have built up a bias, a resistance to reality, on what they have done here.

Todays, sputter etched vac deposited thin film media is only atoms thick, and very uniform, the problems from undersurface material generating noise isnt the problem it once was ...

And the knife edge creates a stood up magnet formed in a sideways stripe at the knife edge, like a domino stood up on a long edge, and will involve 100s of grains across.

The media is being pushed today because laid flat, the magnets formed are as short as they can make them, attempting to make them even shorter than the media is thick

What you see here, is that they can actually create better magnets on the film this way, better than things are now, as hard as they are pushing it currently.

How long we been working on this, 30 years ?

This idea is "spark from god" type stuff, i sure felt it... when it dawned om me, what i was looking at ...

With an easy double/tripple on bits per inch, that 420 gig disk is now over a terabyte.

Disks measured in terabytes ? it took only months for the IBM's GMR head invention to show up on store shelves from publish to product, that produced the jump from 9 gig disks to 90 gig disks. the tech we are using now .

This is not that large a jump, but its still big, because what we are double/trippling, is already big.

I expect to see this on store shelves before the year is out.