New Old FRAM memory chips reliability

Go To Last Post
36 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have a bunch of 10 years old now FRAM chips (2KB I2C 5V, 24C16 type) Ramtron original cool

They have 10 years retention spec.

My question is, for any experienced user of these things: Can I use them in new designs? Can I expect 10 years from now? Or are they just... expired?

The chips are new frown.

Thanks!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

To the best of my understanding. when you write to it, that "10 year data retention" clock starts over. The chip, itself, does not disintegrate into dust after 10 years. It has to do with the odds of electrons escaping the wells that they are put into, or stray electrons coming into wells that were "empty". Those kinds of thing can happen as the result of cosmic rays, xrays, and just the randomness of electron behavior. I would personally not worry about it. 

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This is not a "General Electronics" question, this is a specific question about a particular (non-Atmel) part - so you need to contact the manufacturer for support.

 

rammon wrote:
Ramtron

were acquired by Cypress in 2015:

 

http://www.cypress.com/documentation/other-resources/cypress-semiconductor-has-acquired-ramtron-international-corporation

 

so you should contact Cypress for support:

 

http://www.cypress.com/products/f-ram-nonvolatile-ferroelectric-ram

 

 

Top Tips:

  1. How to properly post source code - see: https://www.avrfreaks.net/comment... - also how to properly include images/pictures
  2. "Garbage" characters on a serial terminal are (almost?) invariably due to wrong baud rate - see: https://learn.sparkfun.com/tutorials/serial-communication
  3. Wrong baud rate is usually due to not running at the speed you thought; check by blinking a LED to see if you get the speed you expected
  4. Difference between a crystal, and a crystal oscillatorhttps://www.avrfreaks.net/comment...
  5. When your question is resolved, mark the solution: https://www.avrfreaks.net/comment...
  6. Beginner's "Getting Started" tips: https://www.avrfreaks.net/comment...
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

awneil wrote:

This is not a "General Electronics" question, this is a specific question about a particular (non-Atmel) part - so you need to contact the manufacturer for support.

 

rammon wrote:
Ramtron

were acquired by Cypress in 2015:

 

http://www.cypress.com/documentation/other-resources/cypress-semiconductor-has-acquired-ramtron-international-corporation

 

so you should contact Cypress for support:

 

http://www.cypress.com/products/f-ram-nonvolatile-ferroelectric-ram

 

 

Well, forget "Ramtron". Is "FRAM memory" general electronics enough?

I've put a honest question. If I'm lucky enough to find here someone with a direct or good experience of this particular issue, so be it.

I don't need to be redirected to the chip manufacturer,

thank you.

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:

To the best of my understanding. when you write to it, that "10 year data retention" clock starts over. The chip, itself, does not disintegrate into dust after 10 years. It has to do with the odds of electrons escaping the wells that they are put into, or stray electrons coming into wells that were "empty". Those kinds of thing can happen as the result of cosmic rays, xrays, and just the randomness of electron behavior. I would personally not worry about it. 

 

Jim

Thank you, it seems logical.

What I can be sure is that I didn't consume any cycle from their 10^14 cycles...

But I was worried about retention time.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

rammon wrote:
But I was worried about retention time.

Unless you baked these chips in an oven, you should not be worried about any lasting effects of depolarization or imprint.

 

If you are concerned about it, a write of ones, followed by a write of zeros, and then a read with its accompanying write-back should sufficiently cancel any effects of depolarization.  New chips new again.  Just start using them.

 

EDIT:  If the chips were kept at a moderate temperature and are indeed new then imprint should not have occurred, all bets are off if they were stored in a hot location.

"I may make you feel but I can't make you think" - Jethro Tull - Thick As A Brick

"void transmigratus(void) {transmigratus();} // recursio infinitus" - larryvc

"It's much more practical to rely on the processing powers of the real debugger, i.e. the one between the keyboard and chair." - JW wek3

"When you arise in the morning think of what a privilege it is to be alive: to breathe, to think, to enjoy, to love." -  Marcus Aurelius

Last Edited: Thu. Mar 1, 2018 - 09:42 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

larryvc wrote:

rammon wrote:
But I was worried about retention time.

Unless you baked these chips in an oven, you should not be worried about any lasting effects of depolarization or imprint.

 

If you are concerned about it, a write of ones, followed by a write of zeros, and then a read with its accompanying write-back should sufficiently cancel any effects of depolarization.  New chips new again.  Just start using them.

 

EDIT:  If the chips were kept at a moderate temperature and are indeed new then imprint should not have occurred, all bets are off if they were stored in a hot location.

Hey, thank you, didn't know that.

No hot location. Maybe 30 degrees Celsius in some hot summers blush ...

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

rammon wrote:
No hot location. Maybe 30 degrees Celsius in some hot summers blush ...
What, did you have them in your pocket?

"I may make you feel but I can't make you think" - Jethro Tull - Thick As A Brick

"void transmigratus(void) {transmigratus();} // recursio infinitus" - larryvc

"It's much more practical to rely on the processing powers of the real debugger, i.e. the one between the keyboard and chair." - JW wek3

"When you arise in the morning think of what a privilege it is to be alive: to breathe, to think, to enjoy, to love." -  Marcus Aurelius

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

larryvc wrote:

rammon wrote:
No hot location. Maybe 30 degrees Celsius in some hot summers blush ...
What, did you have them in your pocket?

LOL, no. I keep them in my office room, which is also some kind of a warehouse for my little components stock.

I don't have air conditioning and a few days in a summer the temperature can rise up to 30-32 celsius.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
It has to do with the odds of electrons escaping the wells that they are put into, or stray electr ...
  Huh? That's for Flash. I believe Ramtron uses some ferro magnetic process that uses a completely different technology underneath. That is also why they get so many write scycles. It does not "wear" like flash.

 

As to the original question:

I haven't looked too deep into Fram technologly

It seems logical that the 10 year data retention timebomb is reset after each fresh write, but that is no solid engineering.

Your datasheet should have that info. Read it carefully.

 

But why trust it?

It is pretty easy to put your data in the chip twice and use a CRC to check your data.

If the CRC fails upon read, then use the other copy.

 

There are some nice hand optimised asm CRC routines in avr/util/crc.h

 

If you as "Rookie" have never used (or heard off) CRC, then do some background reading on it.

CRC's are tried and proven (and standarized, (sort of, there are lots of different standards)) to check the reliability of data.

Paul van der Hoeven.
Bunch of old projects with AVR's:
http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Paulvdh wrote:
I believe Ramtron uses some ferro magnetic process

Yes - that's what the 'F' in "FRAM" stands for!

 

Top Tips:

  1. How to properly post source code - see: https://www.avrfreaks.net/comment... - also how to properly include images/pictures
  2. "Garbage" characters on a serial terminal are (almost?) invariably due to wrong baud rate - see: https://learn.sparkfun.com/tutorials/serial-communication
  3. Wrong baud rate is usually due to not running at the speed you thought; check by blinking a LED to see if you get the speed you expected
  4. Difference between a crystal, and a crystal oscillatorhttps://www.avrfreaks.net/comment...
  5. When your question is resolved, mark the solution: https://www.avrfreaks.net/comment...
  6. Beginner's "Getting Started" tips: https://www.avrfreaks.net/comment...
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

FRAM, at least RAMTRON's, were dual technology. There is an SRAM "layer" on top and the nonvolatile layer is below that. The SRAM layer gives it write speed, as the copy to the NV layer is independent and is at least page-wise parallel. 

 

Yes, the NV part uses magnetic spin states instead of electrons in wells, My bad on that. But, the underlying degradation basics are very similar. Still susceptible to cosmic rays, X rays, and such. Just no wells for electrons to leak into or out of. 

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Be careful of one thing with FRAM's - every read is destructive and it goes back to automatically write after a read.  That shouldn't be a problem, but I had a bad experience with VCC being just slightly too low and ended up losing all the FRAM contents because of it.  Make sure your VCC is SOLID.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MRAM read is not destructive though its Vdd BOD is 3.0V for write-inhibit with 2ms to recover from brownout.

 

Everspin Technologies

Toggle MRAM Technology

https://www.everspin.com/toggle-mram-technology

...

During a read, the pass transistor is activated and data is read by comparing the resistance of the cell to a reference device. 

...

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

alank2 wrote:

Be careful of one thing with FRAM's - every read is destructive and it goes back to automatically write after a read.  That shouldn't be a problem, but I had a bad experience with VCC being just slightly too low and ended up losing all the FRAM contents because of it.  Make sure your VCC is SOLID.

I know that reads are destructive like writes. I'm using FRAM for more than 10 years. And I know about the VCC too (speaking of that, this applies to eeprom and flash too). My BOD level for the cpu is always higher than from the nonvolatile memories.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Paulvdh wrote:

ka7ehk wrote:
It has to do with the odds of electrons escaping the wells that they are put into, or stray electr ...
  Huh? That's for Flash. I believe Ramtron uses some ferro magnetic process that uses a completely different technology underneath. That is also why they get so many write scycles. It does not "wear" like flash.

 

As to the original question:

I haven't looked too deep into Fram technologly

It seems logical that the 10 year data retention timebomb is reset after each fresh write, but that is no solid engineering.

Your datasheet should have that info. Read it carefully.

 

But why trust it?

It is pretty easy to put your data in the chip twice and use a CRC to check your data.

If the CRC fails upon read, then use the other copy.

 

There are some nice hand optimised asm CRC routines in avr/util/crc.h

 

If you as "Rookie" have never used (or heard off) CRC, then do some background reading on it.

CRC's are tried and proven (and standarized, (sort of, there are lots of different standards)) to check the reliability of data.

 

"The other copy" might be just meaningless or plain wrong after 6 months, one year, two or three, etc. When you spec your device for ten years (I spec it for 5 years, but you got the idea).

With FRAM you even don't have the excuse "battery low, sorry, please change the battery).

And about the CRC, do I look like "Sir, I don't know what CRC is. Please explain"? Or you just looked at my join date?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

rammon wrote:
"The other copy" might be just meaningless or plain wrong after 6 months
  ??? If any data changes you of course update both copies.

If both copies (with their checksum) go"plain wrong within half a year" there is no hope for your hardware. (But at least you know it because of the CRC's.)

 

rammon wrote:
And about the CRC, do I look like "Sir, I don't ...
No need to feel offended. "Rookie" level is assigned to you by avrfreaks, based on your post count. I did not do an exhaustive search on your other posts to get a clear picture of your experieance level. I'm just staring at a monitor here.

 

==============

But when thinking about it a bit more. Why bother at all?

If it's for a hobby project, then you'll find out in a few years if data retention is good enough.

If it's for a commercial project, then why bother with a few peseto's/rand/buck/eur for chips?

And if you're paranoid you can have multiple copies in multiple chips. Preferably both Fram & Flash. and you can compare them.

 

You can also use a timer and re-write your data once a year (month, week, day, whatever).

Paul van der Hoeven.
Bunch of old projects with AVR's:
http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Headline in today's EE Times (link takes you to the actual article):

 

Cypress Sees a Future for FRAM

Despite FRAM's inability to cost-effectively scale, its fast write speed and unlimited endurance make it ideal for event logging in autonomous vehicles and industrial Internet of Things applications.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Last Edited: Fri. Mar 2, 2018 - 05:16 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
unlimited (sic) endurance

Note that they are always very careful to say, "almost unlimited endurance" - certainly in print.

Top Tips:

  1. How to properly post source code - see: https://www.avrfreaks.net/comment... - also how to properly include images/pictures
  2. "Garbage" characters on a serial terminal are (almost?) invariably due to wrong baud rate - see: https://learn.sparkfun.com/tutorials/serial-communication
  3. Wrong baud rate is usually due to not running at the speed you thought; check by blinking a LED to see if you get the speed you expected
  4. Difference between a crystal, and a crystal oscillatorhttps://www.avrfreaks.net/comment...
  5. When your question is resolved, mark the solution: https://www.avrfreaks.net/comment...
  6. Beginner's "Getting Started" tips: https://www.avrfreaks.net/comment...
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I did not write that. Some EETimes writer did.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The article does say, "almost unlimited"

 

Top Tips:

  1. How to properly post source code - see: https://www.avrfreaks.net/comment... - also how to properly include images/pictures
  2. "Garbage" characters on a serial terminal are (almost?) invariably due to wrong baud rate - see: https://learn.sparkfun.com/tutorials/serial-communication
  3. Wrong baud rate is usually due to not running at the speed you thought; check by blinking a LED to see if you get the speed you expected
  4. Difference between a crystal, and a crystal oscillatorhttps://www.avrfreaks.net/comment...
  5. When your question is resolved, mark the solution: https://www.avrfreaks.net/comment...
  6. Beginner's "Getting Started" tips: https://www.avrfreaks.net/comment...
Last Edited: Fri. Mar 2, 2018 - 06:07 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Every now and then rumors keep popping up that there is some none volatile static PC memory coming. No more dynamic ram or refresh cycles.

Just pull the plug from your PC when you're done with it for the day and put it back to continue where you were the day before.

 

Wikipedia has a list of some different technologies for NVram:

https://en.wikipedia.org/wiki/Non-volatile_random-access_memory

 

But Flash is so cheap that it's very hard to compete with it.

It's multiple orders of magnitude cheaper than other technologies. You can use a 2GB uSD card (eMMC) and write each cell only once in the lifetime of your product. Who cares if it only has 100 write cycles (for very bad quality Flash chips).

 

On a youtube video of some hacker gatherering presentation they went a bit further into this.

It seems that some of the uSD cards rely so much on error correction that they don't even properly store data on the first write. They have so many dead cells that they have to rely on error correction to get any data out without errors. It's degraded to a statistic probability to be able to get some data back.

On the other end of the spectrum there are also companies (some Swiss one pops up) that claim very high write cycle counts & reliability for each memory cell.

Paul van der Hoeven.
Bunch of old projects with AVR's:
http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Paulvdh wrote:

rammon wrote:
"The other copy" might be just meaningless or plain wrong after 6 months
  ??? If any data changes you of course update both copies.

If both copies (with their checksum) go"plain wrong within half a year" there is no hope for your hardware. (But at least you know it because of the CRC's.)

 

rammon wrote:
And about the CRC, do I look like "Sir, I don't ...
No need to feel offended. "Rookie" level is assigned to you by avrfreaks, based on your post count. I did not do an exhaustive search on your other posts to get a clear picture of your experieance level. I'm just staring at a monitor here.

 

==============

But when thinking about it a bit more. Why bother at all?

If it's for a hobby project, then you'll find out in a few years if data retention is good enough.

If it's for a commercial project, then why bother with a few peseto's/rand/buck/eur for chips?

And if you're paranoid you can have multiple copies in multiple chips. Preferably both Fram & Flash. and you can compare them.

 

You can also use a timer and re-write your data once a year (month, week, day, whatever).

What if I buy new chips and their date code is 2008? Not possible?

So, you say things based on avrfreaks post count. At least, it seems that you realized that your first post was just wrong, and CRCs can't help in this particular case.

You have to trust the hardware. Data retention is data retention. When the user turns off the device with a parameter A=123, after five years when he turns it on again, A should be 123 too. Maybe the user have this written on paper too, who knows? Worse, it may be a calibration value and suddenly after five years the device turns to not be calibrated anymore... I said five years, but what if this will happen after three months?

 

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 2

The first rule of a robust system is to be able to detect an error. CRCs give you this. ECC would be better. Unfortunately, in the real world strange stuff happens that can’t be easily anticipated. Whilst the manufacturers might tell you they guarantee retention for 20 years, that doesn’t mean it is immune to the likes of cosmic and other high energy rays. Error detection and redundancy is your friend. Have you looked into how unreliable a MLC nand flash chip is?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Data Retention Time is an "expected value" with a "confidence level". The manufacturer may tell you "10 years" but what said manufacturer MIGHT mean is "10 years with 99.9% odds of no failure". But only the manufacturer knows. In fact, they have to provide such estimates based on accelerated failure testing because they cannot wait 20 years (or even 10) to verify before releasing the product. So, YOU have to have confidence that the manufacturer has chosen meaningful acceleration factors (its usually but not always temperature). 

 

It appears from on-line comments that the cells do not "wear out" (e.g. loose their ability to function) significantly with age. Rather, it seems mostly to be a matter of pure data retention. That is, a bit NOT retained does not mean that the cell no longer functions. 

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

rammon wrote:
At least, it seems that you realized that your first post was just wrong, and CRCs can't help in this particular case.
Of course CRC's do help "in this particular case" . Upvoted Kartman.

Paul van der Hoeven.
Bunch of old projects with AVR's:
http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Paulvdh wrote:
Of course CRC's do help "in this particular case" . Upvoted Kartman.
+ 1

 

The nice thing about using CRC on FRAM is that the data has to be read to calculate the CRC and compare to the expected value.  For every read there is a write-back performed that freshens the state of the memory location.

 

Paulvdh wrote:

It is pretty easy to put your data in the chip twice and use a CRC to check your data.

If the CRC fails upon read, then use the other copy.

This is a great idea, and as long as a CRC is performed once in a while on this extra copy you can be assured that the retained data is correct.  Any CRC errors should be handled appropriately.

"I may make you feel but I can't make you think" - Jethro Tull - Thick As A Brick

"void transmigratus(void) {transmigratus();} // recursio infinitus" - larryvc

"It's much more practical to rely on the processing powers of the real debugger, i.e. the one between the keyboard and chair." - JW wek3

"When you arise in the morning think of what a privilege it is to be alive: to breathe, to think, to enjoy, to love." -  Marcus Aurelius

Last Edited: Sun. Mar 4, 2018 - 12:49 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

At least, it seems that you realized that your first post was just wrong, and CRCs can't help in this particular case.

What an odd thing to say.

 

Is it a full moon or something?  More than a few posts filled with righteous indignations these last couple of days...

"Experience is what enables you to recognise a mistake the second time you make it."

"Good judgement comes from experience.  Experience comes from bad judgement."

"Wisdom is always wont to arrive late, and to be a little approximate on first possession."

"When you hear hoofbeats, think horses, not unicorns."

"Fast.  Cheap.  Good.  Pick two."

"Read a lot.  Write a lot."

"We see a lot of arses on handlebars around here." - [J Ekdahl]

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There was a full moon on Thursday, March 1.  laugh

Letting the smoke out since 1978

 

 

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

joeymorin wrote:
Is it a full moon or something?  More than a few posts filled with righteous indignations these last couple of days...
digitalDan wrote:

There was a full moon on Thursday, March 1.  laugh

Well then, it seems like a lot of people got mooned then!cheeky

"I may make you feel but I can't make you think" - Jethro Tull - Thick As A Brick

"void transmigratus(void) {transmigratus();} // recursio infinitus" - larryvc

"It's much more practical to rely on the processing powers of the real debugger, i.e. the one between the keyboard and chair." - JW wek3

"When you arise in the morning think of what a privilege it is to be alive: to breathe, to think, to enjoy, to love." -  Marcus Aurelius

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ok, once again, this is about data retention in FRAM memories as a function of TIME.

The particular case is brand new chips older than the minimum retention time specified by the manufacturer.

 

For those who insist on the CRC thing, lets make an experiment: take a chip/board with some SRAM (internal or external, doesn't matter). Treat the SRAM as it is FRAM, or just non-volatile. Make your best to save one value between power cycles. This is just a simulation of an depleted battery (like your SRAM was on battery but the battery just died), or just an accelerated time simulation on FRAM (your seconds may be years on a real FRAM).

Then return here and tell us how the CRC helped.

 

I really want to thank larryvc who gived a very pertinent response, and ka7ehk who seems to understand my particular case.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How about a double moon?

Paul van der Hoeven.
Bunch of old projects with AVR's:
http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How did Shroedinger’s cat fare?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:

How did Shroedinger’s cat fare?

In this universe or another?  Schroedinger's Cat.

 

Rammon, did the chips work or not?

"I may make you feel but I can't make you think" - Jethro Tull - Thick As A Brick

"void transmigratus(void) {transmigratus();} // recursio infinitus" - larryvc

"It's much more practical to rely on the processing powers of the real debugger, i.e. the one between the keyboard and chair." - JW wek3

"When you arise in the morning think of what a privilege it is to be alive: to breathe, to think, to enjoy, to love." -  Marcus Aurelius

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:
How did Shroedinger’s cat fare?
That was an aguably intersting experiment in 1935. The cat is long dead. Even if it had regular meals & drinks.

If it was a young kitten back then and if it had 9 lives I'll give it a small chance.

Paul van der Hoeven.
Bunch of old projects with AVR's:
http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

larryvc wrote:

Kartman wrote:

How did Shroedinger’s cat fare?

In this universe or another?  Schroedinger's Cat.

 

Rammon, did the chips work or not?

The chips work, but FRAM is like battery backed SRAM, when you loose your "battery", you still has SRAM functionality, only without data retention.

So, I only hope you are right. The devices are already in the field by now. They are a ten years old design that got a new order recently. I made them,

ship them, and after that I realized that the FRAM chips that I put inside are already ten years old...