M68HC11
REFERENCE MANUAL
ON-CHIP MEMORY
MOTOROLA
4-23
NOTE
Because the user has no way of knowing what is in the upper four
bits of the CONFIG register EEPROM location, the calculated value
for the selective-write method cannot be determined. Thus, the
selective-write method cannot be used for the CONFIG location.
Since the CONFIG location is only changed a few times in the lifetime
of a product, there is no motivation to use any method other than
erase-before-write to change the CONFIG location.
An examination of the electrical conditions during each of the programming methods
explains why the combination of write-more-zeros method and selective-write method
fails as it did in case A. This analysis also presents possible advantages and disad-
vantages of these programming methods. The basic operation of the floating-gate EE-
PROM is discussed in
4.3.2 Basic Operation of the EEPROM
.
Figure 4-8
demonstrates the erase-before-write method. One disadvantage of this method is that
it requires a time-consuming erase step prior to the programming step. One advantage
of this method is that it can be used to change any data pattern to any other data pat-
tern. Another advantage is that this method is the same as that used for rating the
write-erase life expectancy; thus, much characterization data exists to validate this
method.
The goal of the other two methods is to achieve a longer life expectancy without com-
promising data retention or programming integrity. The program-more-zeros method
appears to have no risks, and some experimental evidence shows that data retention
and program integrity are not compromised. The selective-write method appears to
have some theoretical problems, but experimental data has not confirmed any practi-
cal problem. Due to the theoretical risks of the selective-write method, that method
should probably not be used without a complete understanding of the risks.
Figure 4-9
shows the program-more-zeros method being used to change an EE-
PROM location from $F0 to $C0. In this example, the low-order four bits were previ-
ously programmed, and the current programming operation will change bits 4 and 5 to
zeros.
It has been suggested that it might be undesirable to program some bits longer than
other bits. Since charge transfer during programming occurs at an exponentially de-
caying rate, it seems unlikely that the additional programming time would result in any
significant difference in floating-gate charge. Only considerable characterization data
can prove or disprove these theories, but preliminary data supports the suggestion
that the extra programming time on some bits has no detrimental effects.