bit-tech.net

IBM and TDK partner up to make MRAM

IBM and TDK partner up to make MRAM

A render by IBM of what a MRAM memory cell looks like.

Magnetoresistive Random Access Memory (MRAM) could be a replacement for flash memory a few years ahead in the future. The nonvolatile memory uses magnetic storage elements to store data instead of storing it as an electric charge like flash memory does.

The whole process revolves around two ferromagnetic plates sandwiched together with thin insulating layer between them which creates magnetic cells. One of the plates holds a permanent magnetic polarity while the other is changed by an external field.

The bits are then figured up by measuring the electrical resistance of the cells. If both cells have the same polarity, a low resistance is measured and the bit is read as a "0". If both cells have opposite polarities, then a high resistance is measured and the bit is read as a "1."

The major advantages that MRAM has over flash memories is that its read and write times are significantly faster and it has a near limitless write cycle life. MRAM also will not deteriorate over time like flash memory does due to transistors wearing out over time in flash memories.

The big downside to MRAM right now is that it is currently available in capacities only up to the megabit range. A commercialised MRAM chip was released last year by Freescale Semiconductor, but it didn't really take the market by storm with a 4Mbit (yes, that's bit, not byte - Ed) module costing around £15.

All of this is about to change though as IBM and TDK have launched a joint R&D team to build their own high capacity MRAM chips.

If the joint team succeeds in developing higher capacity chips, then you can expect to find it among all sorts of consumer electronics. The uses for the memory are varied from instant-on desktop computers and printers that never lose their settings to buffer memory in a disk storage system.

"We are planning to work together for about four years to really get this to a level of maturity where we can demonstrate the technology," said Bill Gallagher, the senior manager of exploratory non-volatile memory.

IBM and TDK are the only companies trying to push MRAM onto the market, though. Freescale semiconductor has been putting heavily into its R&D and even licensed out the technology to Honeywell for use in military and space applications.

What types of applications do you guys think could benefit from MRAM? Tell us your ideas over in the forums or in the comments section below.

12 Comments

Discuss in the forums Reply
mmorgue 23rd August 2007, 14:15 Quote
Sounds very promising, once the storage capacities increase to more real-world requirements.
Almightyrastus 23rd August 2007, 14:18 Quote
will be interesting to see how the power usage turns out as well esp if there are possible plans to replace hard drives at some point in the future. Of course power usage will also affect heat and as a result access speeds.
Bluephoenix 23rd August 2007, 17:51 Quote
I want to see how they plan to sheild against severe EMI and outside sources of magnetism.

if they can successfully do both then it might replace ECC memory in number crunching stations for CAD and the like, where the improved speed would lead to faster analysis times.
pendragon 23rd August 2007, 18:50 Quote
Quote:
Originally Posted by Bluephoenix
I want to see how they plan to sheild against severe EMI and outside sources of magnetism.

if they can successfully do both then it might replace ECC memory in number crunching stations for CAD and the like, where the improved speed would lead to faster analysis times.

I agree. This issue immediately lept to mind when I heard it was magnetic.
-Xp- 23rd August 2007, 19:32 Quote
Interesting stuff! Good to see more alternatives to the old hard disk :)

PS: noticed a typo in the last paragraph.
Quote:
IMB and TDK are the only companies trying to push MRAM onto the market...
Glider 23rd August 2007, 19:38 Quote
Quote:
Originally Posted by pendragon
I agree. This issue immediately lept to mind when I heard it was magnetic.
Current HDs are magnetic too, and they seem to do fine?

Faradays Cage to the rescue ;)
completemadness 23rd August 2007, 19:57 Quote
I'm guessing that huge antiferromagnet (or whatever) has something to do with it not getting destroyed by EMR
Bluephoenix 24th August 2007, 16:10 Quote
Quote:
Originally Posted by Glider
Current HDs are magnetic too, and they seem to do fine?

Faradays Cage to the rescue ;)

a Harddrive is not that sensitive to magnetism when on or off because you have to drag a magnet across the disc (or across the side to it for perpendicular recording) to realign the cells and create read errors. with the MRAM, any severe EMI could cause a bit change (and subsequent read error) based upon the design shown.
completemadness 26th August 2007, 06:29 Quote
Quote:
Originally Posted by Bluephoenix
because you have to drag a magnet across the disc
Correct
Quote:
or across the side to it for perpendicular recording
I'm pretty sure that's wrong - you just have to move the particles through the magnetic field to destroy them, where you put the magnet (or the alignment of the cells) doesn't really matter
Bluephoenix 26th August 2007, 08:15 Quote
Quote:
Originally Posted by completemadness

I'm pretty sure that's wrong - you just have to move the particles through the magnetic field to destroy them, where you put the magnet (or the alignment of the cells) doesn't really matter

personally tried it with a half-dead disc, and it is indeed correct.
completemadness 26th August 2007, 11:47 Quote
Quote:
Originally Posted by Bluephoenix
personally tried it with a half-dead disc, and it is indeed correct.
whos right ? lol ;)
Bluephoenix 26th August 2007, 19:46 Quote
Quote:
Originally Posted by completemadness
whos right ? lol ;)

I meant that I tried running a magnet at 90 degrees to the disc, on a half dead perpendicular recording drive, and more sectors turned up with errors.

so that would be me who's right.


but you did have a point, which is why I resorted to physical testing in the first place! :D
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums