Re: Few questions on embedded stuff

From: Thad Smith (thadsmith_at_acm.org)
Date: 06/30/04 

Date: Wed, 30 Jun 2004 09:25:19 -0600

Jim Granville wrote:

> This has some interesting info on the content-reliability of SRAM:
>
> http://www.physorg.com/printnews.php?newsid=227

To quote from that page:

> An approximately 4-digit improvement in the soft error rate compared with
> Renesas Technology's previous 0.13 µm process 16M-bit low-power SRAM
> (without ECC circuitry ) has been achieved by providing a cylindrical capacitor,
> like the type used for DRAM cells, at each memory cell storage node. Alpha ray
> radiation experiments have confirmed for the first time that the SRAM
> performance is free of bit defects due to soft errors.

What does "4-digit improvement in soft error rate" mean? Failure rate
1/3000 of comparison RAM? What is the approximate failure rate of
conventional uncorrected SRAM? of the new technology? At what flux is
SRAM immune to radiation errors? What is the safety margin with respect
to normal background radiation? How does this compare to memories with
larger processes (not just previous 0.13 um)?

Thad