Revision of default fuse settings

[GoogleCodeExporter]

The default fuse bits specifications in the device files may cause problems
with programmers which do not mask off fixed zero and fixed one bits of
specific devices.

Original issue reported on code.google.com by robhamerling on 10 Oct 2008 at 11:08

[GoogleCodeExporter]

With version 0.0.50 of the Dev2Jal script every device file has a PIC specific 
set of
default fuse settings. This should prevent verify errors with not so well 
behaving
PIC programmers. 

Original comment by robhamerling on 12 Oct 2008 at 12:08

• Changed state: Fixed

[GoogleCodeExporter]

Although 'fixed' in Oct 2008, I have re-opened this issue because several 
device files have default config bits settings which are not conform the 
datasheet defaults.

Background info: The default config bits settings in the device files come from 
a tools file devicespecific.cmd, which contains an entry for every PIC. When a 
new PIC arrives a new set of parameters has to be added. This is typically 
performed by 'copy-paste-modify', without an (automated) check with datasheet 
or MPLAB, thus a possible source of errors. 

It seems feasible to derive default config bits settings from the .dev files in 
MPLAB. Will investigate....

Original comment by robhamerling on 28 Mar 2012 at 6:55

• Changed state: Started
• Added labels: Component-Device

[GoogleCodeExporter]

For the record.

I'm working on an extension of the dev2jal script to collect the default config 
bits settings from the MPLAB .dev files, which seems to work out fine.

The bad news:
While checking the preliminary results with the datasheets I found 
contradictions in the datasheet itself. For example:
Take 39582B of the 16f87xa, page 144 (REGISTER 14-1).  Above the bar with the 
config bits you'll find 'U-0' for bit 12. At the bottom of the page you'll find 
in the legend the meaning of this: U = Unimplemented bit, read as '0'.
Now look up on the same page the description for bit 12: Unimplemented: Read as 
'1'.
So bit 12 is probably(!) unimplemented, but does it read as '0' or as '1'???  
Or does Microchip a different logic than I'm used to? 
And this is just an example, several other datasheets have similar 
'contradictions'.
Worse: in several MPLAB .dev files (latest version) I found several occasions 
where other bits are specified as Unimplemented (Unused) than the datasheet 
tells me... 

The good news:
My exercise shows that the extended script will correct many errors in the 
default config bits settings (at least for the implemented bits, but also for 
unimplemented bits!).

... to be continued ...

Original comment by robhamerling on 30 Mar 2012 at 1:10

[GoogleCodeExporter]

I added a procedure to the dev2jal script to derive default configuration bits 
settings from MPLAB .dev files.  Because of the issues described above in 
comment 3, I kept it possible to specify an alternative setting in 
devicespecific.json, which will override the derived settings. For by far the 
most PICs the derived setting matches the datasheet specifications, randomly 
checked with several datasheets. Therefore the old specification in 
devicespecific.json has been removed, except for a number of PICs for which the 
derived specification from MPLAB seems not completely correct.

Original comment by robhamerling on 2 Apr 2012 at 12:16

[GoogleCodeExporter]

More checks showed that the default configuration bits settings as of Revision 
3019 (derived from the MPLAB .dev files, partly overridden by an alternative in 
devicespecific.json)) are OK. 
When faulty or inconvenient settings reported devicespecific.json will be 
modified accordingly. No need to open a new issue or reopen this issue.

Original comment by robhamerling on 6 Apr 2012 at 10:10

• Changed state: Fixed