Rather than me seperatly e-mail those who sent me private e-mail about the
12C508 & their trouble with the JW version's not taking a cal. constant, I
wanted to post something that might help you out.

Please notice that the trouble is not that the value doesn't work when
OSCCAL is loaded from with the main part of the program; changing it 'on the
fly' within the main prog loop seems to work fine.  The problem is that (at
least for some of us), setting it with the first two bytes executed after
powerup does not work.

I have isolated it down to two trivial cases:

This DOES NOT WORK on some JW 12C508's:
---------------------------------------

1FFE   0C40   movlw 0x40
0000   0025   movwf 0x5
------ the two above are the important ones, the code following is other init.
0001          clrf 0x4
0002          movlf 0x9
0003          movlw 0xf
0004          moxwf 0x7
              etc.


The following code DOES correctly set OSCCAL (notice two duplicate lines added)
-------------------------------------------------------------------------------

1FFE   0C40   movlw 0x40
0000   0025   movwf 0x5

0001          clrf 0x4
0002          movlf 0x9

0003   0C40   movlw 0x40     \
0004   0025   movlw 0x5      / Repeating these two lines DOES load OSCCAL

0005          movlw 0xf
0006          moxwf 0x7
              etc.


This points to a powerup problem, at least in my mind.  Also, notice that to
make this woek, you need to KNOW the calibration constant, so it won't work
well for mass-programming OTP's.  Why would the value stay in OSCCAL when
written with instructions several cycles after powerup, but not when written
on the first two instructions?

        -Todd Peterson