Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
How to Build a Computer 25: The Magnetoresistive Effect
Our story starts with Lord Kelvin, one of the great old school physicists. You can read about his career from anonymous’s old Saturday Night Science. Actually, at the point he enters this story I don’t think Kelvin had made lord yet; he was just some bloke named Thompson. This Thompson fellow was playing around with magnets and electricity and that sort of thing. What he discovered is that you can change the resistance of a wire with a magnetic field. And furthermore that that change in resistance depends on the angle between the wire and the magnetic field.
Let’s take that a little more slowly. Change in resistance when you’re in a magnetic field? Okay, I can buy that; there’s all this nonsense about wires and magnets and whatnot that I’ve been blathering about up until this point. Angle? The resistance in your wire will vary a great deal whether it’s parallel or antiparallel to the magnetic field on your disk. (Antiparallel means parallel, but facing the other direction. The northbound lane on a highway is antiparallel to the southbound lane.) If your wire is running current right-to-left and your magnetic field is pointed left-to-right then your wire’s resistance is at it’s highest because of your antiparallel configuration.
The benefit to the magnetoresistive read head is that it allows you to separate the reading head from the writing head. That does a couple things for you. For example, you want to write to a wide track to ensure the data is readable, but you want to read from a narrow track to ensure you don’t pull any data from the next one over. You don’t have to compromise on width if you’ve got separate heads. The write head writes to a wide area, the read head reads from a narrow area. Note that those are relative terms; as @Saxonburg was kind enough to note, on modern drives these tracks are only about three hundred atoms wide.
Modern hard drives will actually stack a column of these heads together to increase the data transfer rate. Here, this is what your write head looks like:
Actually, that’s not that different from the kielbasa configuration; you’ve got the coil and you’ve got the gap. This is what your read head looks like:
The wire going through the middle (and connecting the two pencil-wires coming in from the top) is where you get your magnetoresistance to measure. We don’t use straight magnetoresistance these days; you can get it a lot more sensitive if you take advantage of some quantum effects. In my college physics days I actually sat through a colloquium on the subject, but between the presenter’s thick Indian accent and my unfamiliarity with the subject I got almost nothing out of it. Except for the cookies.
You can get ‘giant’ magnetoresistance (that’s a technical term mind you) by layering materials. Start with an electrode. Then put down a permanent magnet facing one direction. Put a thin layer of non-magnetic material on top of that. Then put your platter surface where you’re going to be writing stuff on top of that. The magnetic layer below reinforces the magnetoresistance effect in the top layer (something to do with quantum spin coupling; I don’t understand it either.)
You can actually do one better; if you epitaxially (uh, I should have explained that word before now, and probably will in the future. No spoilers!) epitaxially add a layer of magnesium oxide in between magnetic layers you get a quantum tunneling interaction between your magnets. This really boosts the signal, from like +/- 5% with normal magnetoresistance to about +/- 110% with Tunneling Magnetoresistance (commonly abbreviated TMR). All modern hard disk drives use TMR.
Since TMR boosts your signal so much you can start with a smaller signal. That is, you can make your spots smaller, get a higher areal density, and store more gigabytes of cat pictures. As with the older heads, it’s still a good idea to boost your signal in other ways. Current hard drives float the heads about three nanometers above the surface of the platter, using a cushion of air to keep it from scraping. (Don’t get dust on your hard drive platter; bad things happen if you do.)
How small is it theoretically possible to make your spaces? A discrete unit of these consists of something like eighteen grains. If you make ’em much smaller than that the magnetism can randomly flip directions (you’d be small enough that quantum effects come into play), which obviously messes with your ability to accurately store and retrieve data.
We’ve covered the topic of reading and writing information to your disk, and we’ve covered it pretty thoroughly. I’ll be backing out a bit to focus on the mechanical aspects of an HDD, how you’re spinning the disk and whatnot. Join us next week for “A Closer Look at what’s Heaping on your Platter” or “Rust or Bust!”
This is part twenty-five of my ongoing series on building a computer, the New Model Army way. You may find previous parts under the tag How to Build a Computer. This week’s post has been brought to you by Ricochet Silent Radio! Join us next week for a thrilling story of future as we follow the unlikely career of Judge Mental: Lord Protector of the United States! Remember; three chimes mean good times on Ricochet Silent Radio!
[First – Silicon] [Previous – Reading and Writing] [Next – Spindles and Platters and so Forth]
Published in Science & Technology
Is it one of the bad things when somebody confuses four pieces of information with three possible states? Or am I missing something from Information Science?
I’m looking forward to the final part, when he puts all the pieces together: cheese cubes, smoked meats, spaghetti, ice cream sandwiches, toast… and makes it all work.
For eighty years, since it’s analog, video technology has had a concept that long before BLM is called “blacker than black”. Back in the analog TV world, roughly 1939-2009, there’s never supposed to be a moment when there’s no video signal coming down the line. Even a flat, dead, black picture would indicate (let’s just make it up) voltage of, say, .0032 microvolts. A great big bright sunny picture might generate (relative to this hypothetical example) 1 volts. So signals coming in at (let’s say) .0002 to .00010 microvolts, well below even the darkest of dark signals, will be filtered out of the video picture but used as invisible guides, as control pulses.
And not even a cookie to soothe. If it’s any consolation the next post will be talking about spinning disks and that sort of thing, much easier to follow.
Nothing so complicated; I’m assuming it’s also got an option for “YesYesYesYesYes” when it wants to signal strong agreement.
Navy ships have (probably had by now) long used analog computers for their fire control systems, as described in BUSHIPS Ordinance Publication 1303. They worked on the principle that magnetic fields tend to align themselves to one another. They didn’t need no stinkin’ bits.
You are so binary.
All or nothing baby!
Wikipedia has a really great picture of a read/write head. Not one that my company made, but still a pretty great picture.
I’ll drop one of ours below so you can compare it. Mine still has photoresist on it (the purply stuff), and it hasn’t had the stainless steel etched away for most of it yet. You can see on the Seagate picture the solder droplets on the bond pad, and some of the etching. The way they have the holes around the head? That’s a spring; it gives more flexibility to the, ah, flexure to help maintain the correct height off the platter.