Lpc2000

--- In lpc2000@yahoogroups.com, Jayasooriah <jayasooriah@...> wrote:
>
> Dave,
> 
> An alternative scenario (that was suggested by a foundry expert I
spoke to 
> when I first discovered this) is more plausible for the following
reasons:
> 
> 1/  The boot loader does not select good segment(s) for RAM or ROM.  It 
> only limits their sizes by reducing the respective bound.  In other
words 
> the boot loader does not set or change the base.


That particular assertion would be much more assuring if it was coming
from Phillip's.   As for size limiting, of course the FLASH address
decoder requires some remapping, but if that is implemented in FLASH
itself that is really trivial.


> 
> 2/  It is very unusual to use this method (of selecting using bound
only) 
> to improve yield.  

Selecting bounds only would be practically useless, since error
segments are randomly located. The FLASH address decoder also needs
adjustment. However, remapping FLASH segments during prepackaging
tests is a very routine operation.

> He thought that for the sizes of RAM and ROM involved 
> here, it is unlikely there are yield problems to the extent software is 
> used to adjust the bound post production.

GIGO

> 
> Jaya
> 
> --- In lpc2000@yahoogroups.com, "derbaier" <dershu@> wrote:
>  > It is fairly common to "bin" parts during prepackaging testing, so
>  > that chips that have failed some part of the testing can still be
>  > used.  If a segment of memory fails in testing the part can simply be
>  > downgraded to the next smaller memory specification and sold at a
>  > reduced price. In other kinds of ASIC chips that is usually done as a
>  > bond out option, but in this case it was apparently done when the
boot
>  > loader was programmed prior to packaging. Since the tests are
>  > performed prior to packaging, it is possible that the chips you have
>  > were simply mispackaged if you only have a very small quantity of
>  > chips exibiting the problem that you describe. If you can, it would
>  > probably be good if you could return the chips to your
distributor for
>  > replacement. That way, Phillips would be able to verify the problem
>  > and possibly eliminate future occurances. In other words, it is
fairly
>  > likely that you have an LPC2131 silicon in a LPC2136 package.
>  >
>  > That's just my guess!
>  > -- Dave
> 
> Send instant messages to your online friends
http://au.messenger.yahoo.com
>