max # of eeprom read cycles

[terry michaels]

Message text written by Rob
>Ok, I realize that after a while, the data stored within an EEPROM goes
"soft", where it supplies an intermittent data state..
I saw a modern chip that stated 10 million cycles... Now, in the period
of 14 yrs and with everyday use, this could easily be exceeded, right?
I think I *might* have found a culprit..
So, this means that many pieces of older gear with older and less stable
EEPROMs should all be going "soft" about now..
What do you think? Any thoughts about this theory?
Rob
<

Hi Rob:

This is a subject I've thought about a few times.  There is a huge number
of microprocessor based products out there that have their firmware stored
in a EPROM, which like an EEPROM, stores its data as charges on an aluminum
gate buried within a layer of silicon dioxide, which is very pure glass. 
The gate is a small capacitor, and you can imagine it would take an
extremely small amount of leakage to alter the charge over a timescale of
years.  This arrangement seems to work well, but I seem to remember that
EPROMS are only guaranteed to retain data for 10 years, after which charge
degradation due to defects in the silicon dioxide and tunneling effects
might begin to discharge some of the gates below the "high bit" threshold. 
Since these devices have been used in large numbers for around 20 years
now, you would think there would be a large number of MPU based products
that would be crashing about now.

BTW, I fired up a Z80 based computer that was built in 1980, with its CP/M
operating system stored entirely in EPROMs, about a month ago.  It still
worked.

Terry Michaels