Lpc2000

Hello to all, this is my first post in this group.

I'm thinking of using the LPC2138 in a design that must have a 20
year life cycle. The literature says the flash has only 10 year 
retention, but in other flash devices that is affected by
temperature, write cycles etc. 

Is there any more detailed info on this? For example, if I only write 
to the flash a few times and keep the temperature below 60C, can I 
expect longer retention?

Cheers
Dave

At 02:22 AM 6/19/05 +0000, foreigner_ca wrote:
>I'm thinking of using the LPC2138 in a design that must have a 20
>year life cycle. The literature says the flash has only 10 year
>retention, but in other flash devices that is affected by
>temperature, write cycles etc.

The 2104 data sheet claims a 20 year retention time.  The 2138 datasheet 
claims 20 years in the text but doesn't list retention in the 
characteristics section.

>Is there any more detailed info on this? For example, if I only write
>to the flash a few times and keep the temperature below 60C, can I
>expect longer retention?

I would expect it to be more related to temperature than programming 
cycles.  At least until you have a significant number anyway, say a few 
thousand or so.   In fact I would expect it to be quite non-linear with 
temperature.  It would be interesting to have the information.  Perhaps 
philips_apps can point us to an appnote or suggest one be written?

Robert

" 'Freedom' has no meaning of itself.  There are always restrictions,   be 
they legal, genetic, or physical.  If you don't believe me, try to chew a 
radio signal. "  -- Kelvin Throop, III
http://www.aeolusdevelopment.com/

Hi Dave,

let me try to explain. The 10 years are specified after all 10k 
reprogramming cycles done. It is all about how many ppm (or for that 
matter fits) the device will have. The target is always to have 0 
failures over the specified time. Statistically thats not really 
possible but every vendor tries the best to get as close as possible. 
With this is mind the typical retention is much more along the value of 
100 years (no typo) than 10 years. Typical being that even at 100 years 
less than 1 in 100 may be less than 1 in 1000 devices is expected to 
fail.

So, in an attempt to answer your question what happens after ten years? 
The probability of one bit in the whole flash memory failing increases 
day by day, and after ten years the fit rate is just a little higher 
than after 1 year and just a little lower than after 20 or even 50 or 
100 years. To the closest approach what will happen, nothing bad to the 
flash.

There was an answer from Robert stating that reprogramming cycles have 
a stronger effect on the data retention than temperature, well both 
have an effect because high temperature is used to simulate fast aging, 
so at higher temperatures, the years go by faster ;-) 

Given your less than 60C and just a few programming cycles, though 
shall not fear any failures in more than 20 years. 

Hope this help, Robert


--- In lpc2000@yahoogroups.com, "foreigner_ca" <va3dp@r...> wrote:

> Hello to all, this is my first post in this group.
> 
> I'm thinking of using the LPC2138 in a design that must have a 20
> year life cycle. The literature says the flash has only 10 year 
> retention, but in other flash devices that is affected by
> temperature, write cycles etc. 
> 
> Is there any more detailed info on this? For example, if I only write 
> to the flash a few times and keep the temperature below 60C, can I 
> expect longer retention?
> 
> Cheers
> Dave

Yes, thank you. I'm quite excited to start working with this device.

Dave


--- In lpc2000@yahoogroups.com, "philips_apps" <philips_apps@y...> 
wrote:
> Hi Dave,
> 
> let me try to explain. The 10 years are specified after all 10k 
> reprogramming cycles done. It is all about how many ppm (or for 
that 
> matter fits) the device will have. The target is always to have 0 
> failures over the specified time. Statistically thats not really 
> possible but every vendor tries the best to get as close as 
possible. 
> With this is mind the typical retention is much more along the 
value of 
> 100 years (no typo) than 10 years. Typical being that even at 100 
years 
> less than 1 in 100 may be less than 1 in 1000 devices is expected 
to 
> fail.
> 
> So, in an attempt to answer your question what happens after ten 
years? 
> The probability of one bit in the whole flash memory failing 
increases 
> day by day, and after ten years the fit rate is just a little 
higher 
> than after 1 year and just a little lower than after 20 or even 50 
or 
> 100 years. To the closest approach what will happen, nothing bad to 
the 
> flash.
> 
> There was an answer from Robert stating that reprogramming cycles 
have 
> a stronger effect on the data retention than temperature, well both 
> have an effect because high temperature is used to simulate fast 
aging, 
> so at higher temperatures, the years go by faster ;-) 
> 
> Given your less than 60C and just a few programming cycles, though 
> shall not fear any failures in more than 20 years. 
> 
> Hope this help, Robert
> 
> 
> --- In lpc2000@yahoogroups.com, "foreigner_ca" <va3dp@r...> wrote:
> > Hello to all, this is my first post in this group.
> > 
> > I'm thinking of using the LPC2138 in a design that must have a 20
> > year life cycle. The literature says the flash has only 10 year 
> > retention, but in other flash devices that is affected by
> > temperature, write cycles etc. 
> > 
> > Is there any more detailed info on this? For example, if I only 
write 
> > to the flash a few times and keep the temperature below 60C, can 
I 

> > expect longer retention?
> > 
> > Cheers
> > Dave

At 11:21 PM 6/19/05 +0000, philips_apps wrote:
>So, in an attempt to answer your question what happens after ten years?
>The probability of one bit in the whole flash memory failing increases
>day by day, and after ten years the fit rate is just a little higher
>than after 1 year and just a little lower than after 20 or even 50 or
>100 years. To the closest approach what will happen, nothing bad to the
>flash.
>
>There was an answer from Robert stating that reprogramming cycles have
>a stronger effect on the data retention than temperature,

Actually Robert I suggested the reverse.  I meant to anyway.  The 
assumption being that programming lifetime was limited by the strength of 
the oxide layer so that if you only used a small percentage of the rated 
cycles you would not have much effect on the oxide layer and therefore not 
perturb the retention time significantly.  Temperature on the other hand 
raises the thermal energy of the trapped electrons making tunneling more 
likely resulting in an increase in the leakage current and a drop in the 
retention time.  Also I would expect the tunneling probability and thus 
retention time to probably vary exponentially with temperature.

There seem to be relatively little literature covering this.  After I wrote 
my initial reply I did a quick search and the only thing I found was a 
Freescale application note that essentially provided the same thermal 
explanation, although it didn't cover programming cycles.  Does Philips 
have any public information on this?

Robert

" 'Freedom' has no meaning of itself.  There are always restrictions,   be 
they legal, genetic, or physical.  If you don't believe me, try to chew a 
radio signal. "  -- Kelvin Throop, III
http://www.aeolusdevelopment.com/