Talk:Ferroelectric RAM

Definition for "near-unlimited"
Can we have a definition for "near-unlimited"? Or maybe just a real number? :-) - Omegatron 01:25, Jun 20, 2005 (UTC)


 * Better now? :) --Shaddack 02:21, 20 Jun 2005 (UTC)


 * near-infinitely better. :-) - Omegatron 02:40, Jun 20, 2005 (UTC)


 * Cripes, has this article really been around for 8 years now... anyway, if this is in terms of read/write cycles, it looks like it's 10^16 (or roughly 2^52), which is pretty good. Continual read-writes at about 25MHz (which seems like about what's realistically achievable right now - or maybe rather less?) would give it a lifetime of 12.7 years, which is pretty damn good for a flash memory competitor. Still not comparable to yer typical hard disk, but for removable storage that won't be getting continually hammered, that's effectively "completely stable". In other words, a 25 year lifespan at 50% duty cycle. One from 1988 which had seen fairly hard service would still be operable. Regular DRAM can fail in less time than that. 193.63.174.211 (talk) 17:58, 7 October 2013 (UTC)

Ramtron?
What's the scoop on the shameless vendor plug?


 * Are you referring to the purely factual information that Ramtron is the main (perhaps only?) vendor of FRAM chips? --Brouhaha 00:37, 9 November 2005 (UTC)


 * One must assume this is the case. Perhaps this is a shameless vendor plug from one of FRAM's competitors? Maury 13:06, 15 November 2005 (UTC)


 * Maybe we should update this page, since the statement about Ramtron certainly isn't true now: "Fujitsu started volume production of FRAM products in 1999, and has reached worldwide sales of approximately 500 million chips, including discrete memory chips and chips with embedded FRAM memory. ... Fujitsu doubled the memory capacity from previous products and brought the MB85R2001 and MB85R2002 products to market as early as possible. These 2 Mbit FRAM products are the largest capacity FRAM products at volume production level in the world." From Recent press announcement)--Wildwillywright 17:15, 9 May 2007 (UTC)


 * Ramtron has been in FRAM forever so I'd cut them some slack. But also Chrysalis was doing FRAM out of Albuquerque in the mid-1980s.  When I was at Aerospace Corporation we signed off on USAF-SD seed funding for quite a number of NVM technologies including FRAM (Ramtron and Chrysalis both got funding IRC) and MRAM at Honeywell.  This was in 1986.  I was the technical guru and sign-off on all NVM technology for the USAF at the time.


 * Ramtron and Chrysalis provided us with samples of their FRAMs back then which we did "special" testing on for space applications. You have to remember the 20-year rule, however.  Commercial viability is very different but it's also somewhat arbitrary as a definition.


 * The JPL link is somewhat tangential compared to what we were already working on with various vendors. We funded to eventually have parts for SDI and put quite a bit of money into various technologies.


 * A special example of the 20-year rule of innovation: it 20 years to within a month between the first MRAM (using AMR rather than GMR) and when Freescale came out with their first commercial MRAM. The Freescale effort directly ties to design and people from the mid-80s Honeywell work. Jsgski (talk) 23:50, 16 June 2009 (UTC)

Magnetic properties
Is this memory sensitive to magnetic fields?


 * No, it's electric fields that would be an issue, not magnetic ones. It's not clear whether or not this is a problem in practice -- stray magnetic fields do not cause problems with MRAM for instance, as the field needs to be extremely targeted in order for a write to occur. Maury

Proposal for Featured article
I think it deserves it. Introduction and conclusion as well as a good explanation of its way of working. Who wants to nominate it? David Latapie (✒ | @) 13:26, 26 September 2006 (UTC)
 * Please follow the directions at WP:FAC more carefully if you want to nominate the article. &mdash;Bunchofgrapes (talk) 05:10, 12 November 2006 (UTC)

Figure?
The text mentions a figure that I can't see. Where is it? --pgimeno 16:19, 27 May 2007 (UTC)
 * I've removed the reference (it would be good to see the figure though). 82.3.234.155 (talk) 13:06, 3 August 2008 (UTC)

This type of ram has been developed
This type of RAM has been developed. http://octopart.com/info/Ramtron/FM24C256-G?c=25&d=0 XU-engineer 01:55, 22 October 2007 (UTC)

Flash write time?
There is a line:

Flash memories commonly need about 1 ms to write a bit, whereas even current FeRAMs are at least 100 times that speed.

Am I missing something here? My experience with flash is that it's at least 1000x faster than this.... 125 bytes/sec write time? C'mon!!!! —Preceding unsigned comment added by 76.166.187.216 (talk) 00:15, 24 January 2008 (UTC)


 * Agreed, I think there are a few mistakes regarding the relative read/write speeds of FRAM. I have it on pretty good authority (a solid state physicist working on multiferroics) that FRAM reading involves going through a hysteresis cycle every time a bit is read and/or written, making the whole process quite slow. I inserted a citation needed tag in the opening paragraph. Larryisgood (talk) 17:32, 29 January 2011 (UTC)

More early history
I hesitate to put this in the first paragraph of the article - perhaps it belongs elsewhere.

June 1955 Scientific American Magazine article "Computer Memories" by Dr. Louis N. Ridenour has a photograph of a ferroelectric RAM on page 92, built by Bell Telephone Laboratories. It is not clear to me whether or not this memory ever entered production. AlanDewey (talk) 23:08, 13 July 2010 (UTC)

The accepted generic acronym for ferroelectric random-access memory: FeRAM or F-RAM ?
Firstly, I'm inclined to avoid using the acronym FRAM because that leads to a disambiguation, although some companies such as Texas Instruments use it. The Fram was an arctic exploration vessel from Norway.

In a footnote, Ferroelectric_RAM says FeRAM is the accepted generic acronym for ferroelectric random-access memory.

In a footnote, Non-volatile_memory says F-RAM is the accepted generic acronym for ferroelectric random-access memory.

So which is it? Wikipedia should be consistent with itself. Google search returns:
 * About 391,000 results for FeRAM, and surely nearly, if not all of them are for ferroelectric RAM. Some use Feram: —that's also the name of a hero in the Dragon Scroll video game
 * About 251,000,000 results for F-RAM, headed by FRAM filters and the ship. Hard to say how many of those hits are for ferroelectric RAM.  There are also hits for Fleet Rehabilitation and Modernization.
 * About 1,490,000 results for "F-RAM", with mostly all being for ferroelectric RAM. This is the acronym used by Ramtron International, the main supplier of F-RAM chips, according to Wikipedia.  Our Ramtron International article favors this acronym, as does their website.

Fujitsu comes up second after Ramtron in Google searches, but I don't see mention of ferroelectric RAM in their Wikipedia article. Their website uses FeRAM but seems to favor the more generic FRAM. They claim, ''Fujitsu has delivered more than 1 billion FRAM devices. We lead the world in developing and manufacturing FRAM.''

On an electronics vendor site the only ferroelectric RAM I could find was F-RAM (made by Ramtron, of course). This strikes me as a battle for the de facto standard market leadership and acronym. I'm not sure we can declare a winner yet. Maybe we just report which acronyms are favored by which companies. Wbm1058 (talk) 21:34, 11 February 2012 (UTC)


 * Footnotes referenced above have been removed. There is no single accepted generic acronym for ferroelectric random-access memory. Wbm1058 (talk) 16:39, 18 February 2012 (UTC)

Alright, so what's the typical data density?
Both per regular size chip (say, maybe per square mm/square centimetre/square inch, and/or per something that would fit into a microSD space, and per a regular SD space, both of the latter of which are also about the size of typical permanently surface-mounted storage chips on system PCBs), and per US dollar? There's absolutely zip in the article right now, dated or otherwise, to give us a pointer on this other than a vague note that "it's a lot less than flash". Which, yknow, is about accurate for regular DRAM anyway; you can get 64GB into the space of a uSD with flash, I haven't seen any DRAM devices that offer that kind of density yet. 193.63.174.211 (talk) —Preceding undated comment added 18:01, 7 October 2013 (UTC)


 * Anyone? It's been almost 3 years now. 193.63.174.115 (talk) 08:16, 28 June 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 4 external links on Ferroelectric RAM. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20090414013337/http://trs-new.jpl.nasa.gov/dspace/bitstream/2014/33777/1/94-0506.pdf to http://trs-new.jpl.nasa.gov/dspace/bitstream/2014/33777/1/94-0506.pdf
 * Added archive https://web.archive.org/web/20100615155151/http://www.materials.drexel.edu/mml/pubs/Spanier_Kolpak_etal_NL06.pdf to http://www.materials.drexel.edu/mml/pubs/Spanier_Kolpak_etal_NL06.pdf
 * Added archive https://web.archive.org/web/20060920032305/http://download.micron.com/pdf/technotes/ddr2/TN4716.pdf to http://download.micron.com/pdf/technotes/ddr2/TN4716.pdf
 * Added archive https://web.archive.org/web/20090228193906/http://www.ramtron-online.cn/ to http://www.ramtron-online.cn/

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 04:26, 30 September 2017 (UTC)

wrong statement: "has the unusual technical disadvantage of a destructive read process"
All DRAM is destructive for read. Dram is the main PC memory of the last 40 years. So, nothing is unusual here. — Preceding unsigned comment added by 87.176.133.106 (talk) 20:33, 4 May 2018 (UTC)

The allegation that "FRAM read is destructive" is simply misleading. There is no such behaviour upon a read of FRAM data from outside the chip. The only true thing about this is, that FRAM chips implement a "write after read" _architecture_ "inside". And this is invisible for the application and completely transparent to the user, and thus never requires a user-side "write after read". So after a read, data is still there just as it was before.

Far too many weasel words
The "Comparison with other memory types" sections of this article is filled with weasel words. For example (emphasis added):


 * ...it seems reasonable to suggest that the benchmark...
 * The destructive read aspect of FeRAM may put it at a disadvantage...
 * This means that FeRAM could be expected to be lower power...
 * For a "mostly-read" device the difference might be slight, but for devices with more balanced read and write the difference could be expected to be much higher.
 * ... FeRAM could be much faster than DRAM...
 * ... the electrical and switching delays would likely be similar...

... and on and on and on (at least as many more as those I've cited above). It reads like someone is hypothesizing and reasoning and guessing about all this, rather than basing their writing on real sources. Sure, the author may be correct, but being likely to be true shouldn't be good enough to be included in Wikipedia.

I'm tempted to spray tags all over the place, but before I do I wanted to check here. Any thoughts on the appropriate way to handle this? -- Dan Griscom (talk) 01:47, 28 November 2018 (UTC)

Earliest known commercial product ?
The article claims that "The earliest known commercial product to use FeRAM is Sony's PlayStation 2 Memory Card (8 MB), released in 2000". However, FeRAM has also been used in Sonic 3 cartridges for Genesis (1994). See https://info.sonicretro.org/Sonic_the_Hedgehog_3. — Preceding unsigned comment added by DarkMagus237 (talk • contribs) 07:44, 10 October 2022 (UTC)

Conclusion
Please refresh 223.237.95.102 (talk) 12:43, 22 November 2022 (UTC)