Todd Peterson wrote:


>Thanks to all who answered my 12C508 JW question.  There have been a
>large
>number of posts concerning the matter, so let me summarize the answer:
>you
>must read the calibration constant out of the last location (0x01ff in
>the
>12C508) and write it down somewhere PRIOR to the first UV erase cycle
>or you
>will loose it.
>
>
>My problem seems to be different; In the last location I have 'movlw
>0xcd',
>and in the first location I have 'movwf 0x5'.  As the 12c508 has a 5x
>core,
>it should vector on power-up to the last loaction, which will place
>the
>calibration constant in W.  Then the PC should roll over and write
>this
>value into OSCCAL because of the instruction in the first byte of
>memory.
>An increment (or decrament) of the calibration constant by one is
>supposed
>to change the period by 4ns, according to the specs.  However, I have
>tried
>many different calibration values (FF, E0, CD), and none of them seem
>to
>change the period of a pin I am toggling on an output port and
>watching with
>a scope.  A one ms signal is .8 ms long.
>
>I am using a CCS compiler, and am beginning to think the problem might
>lie
>within its delay routine, as I cannot see what else might be wrong.
>However, the compiler's delay routines have worked well before, and it
>is
>odd that changing the cal. constant wouldn't change the pin toggle
>rate.
>
>Suggestions, anyone?
>
>
>Thanks to all who responded previously; this list truly is effective!
>
>
>-Todd Peterson
>


Just something to think about:

A scope is only about 3% accurate, maybe less depening on

model.  That's about + or - 4.17us or 240khz, if it's in

calibration.  When was the scope last calibrated?  It

might be better to use a frequency counter. And again,

check the calibration.


Carl Watley
WB2ZSG