1 Million endurance Flash chip

Hello,

can anyone suggest flash device which has high erase/program endurance about 1 million or greater??

most probably it will be greater than 1 million.....

guru2kiot said:

most probably it will be greater than 1 million.....

Click to expand...

Most probably?

There are very few flash devices with an erase endurance greater then 100,000 cycles. I know there were a few Spansion parts (now Cypress) but I can't remember which part to go look at (there are a lot of them), besides Cypress may have discontinued those parts. Don't recall if Micron had any with such a high endurance rating.

You could go with an EEPROM device like this: http://www.atmel.com/Images/Atmel-8720-SEEPROM-AT24C512C-Datasheet.pdf

Flash endurance is specified under the worst case conditions, i.e maximum voltage and temperature, if you can ensure that both the voltage and temperature is low then flash endurance is very much higher and may meet your one million requirement.

pjmelect said:

Flash endurance is specified under the worst case conditions, i.e maximum voltage and temperature, if you can ensure that both the voltage and temperature is low then flash endurance is very much higher and may meet your one million requirement.

Click to expand...

Have any data to back up this claim? I could believe perhaps 10-15% increase in the cycle count but a 10x improvement, I really doubt that.

We are discussing commercial commodity ICs that are supposed to die so you will have to buy a new one every 5 years. Besides a military qualified part has the exact same specification as a standard commercial part from what I've seen, they just may be a lower speed grade device. If you look at the EEPROM (which are a much older device type and are mostly used in embedded and industrial applications as opposed to the consumer market) datasheet in my earlier post you'll see that part is spec'd for -55 to 125C which is well beyond the typical 0-80C of a commercial flash device yet it still has a >1,000,000 cycle endurance.

Failure in a flash device is a direct result of the breakdown of the oxide in the floating gate which gets subjected to a charge that damages the oxide. Yes this may accelerate slightly at higher temperatures, but I highly doubt that you could go from 100,000 guaranteed cycles to 1,000,000 guaranteed cycles even running the device at 0C. If I'm wrong then supply the documents that prove that you can run a 100K cycle device reliably to 1M cycles without issue.

You might find higher-reliability flash or EEPROM memories
in the "HiRel" market space along with some newer types
like chalcogenide memories (CRAM) and ferroelectrics
(FRAM). However you are certain not to like the price.

Speculating on acceleration factors is probably not any
good, unless you have the manufacturer's reports for the
product you plan to use you can't draw useful conclusions.
Design of the memory element and its programming cycle
are tradeoffs involving multiple dimensions (die cost, retention
time, write cycle endurance (these last two play off against
each other pretty directly, application supply and temperature
ranges, yield.

If you picked a gentler programming cycle and were willing to
accept a worse low-retention-time bit-population-tail you
might well get to a million write cycles. But are you willing
to do your own reliability work? Probably not.

dick_freebird said:

If you picked a gentler programming cycle and were willing to
accept a worse low-retention-time bit-population-tail you
might well get to a million write cycles. But are you willing
to do your own reliability work? Probably not.

Click to expand...

Along with having to buy the parts from a single lot, then testing some of the parts (to death) then having to return the bulk of the parts if they fail your test and buy new ones...rinse and repeat until you have the parts you need.

Quote Originally Posted by pjmelect View Post
Flash endurance is specified under the worst case conditions, i.e maximum voltage and temperature, if you can ensure that both the voltage and temperature is low then flash endurance is very much higher and may meet your one million requirement.

Click to expand...

Have any data to back up this claim? I could believe perhaps 10-15% increase in the cycle count but a 10x improvement, I really doubt that.

Click to expand...

I have no information on the improvement reduced voltage and temperature makes to the endurance of flash memory, but I remember reading a Microchip data sheet that said that reducing temperature and voltage makes a large improvement in flash endurance. I would be interested if anyone could give any further information on this as the manufacturers seem tight lipped about this.