5.080 PIC24 fuse issues

Ttelmah · Joined: 11 Mar 2010 Posts: 19552

Hi,
Has anyone else had issues with 5.080, and the fuses on PIC24's?.
May be specific to just one chip, or may be more generic...

I have code that compiled with 5.078 loads and runs fine. Compile with 5.080 and it won't work.
PIC24FJ256GA702

They seem to have 'tidied' the fuse handling, with half the number of lines of Intel hex now being generated for the fuses, but a couple not don't seem to be working right.
When you load the file to the programmer, some fuses are setting differently, and a couple don't seem to allow you to set them.
Have had to manually set the fuses in the programmer to get the code to work.

So has anyone else seen fuse issues with the current compiler?.

gaugeguy · Joined: 05 Apr 2011 Posts: 303

I have code for a 24FJ256GA704 where I moved up from 5.075 to 5.080. With 5.080 I have a new dll from CCS for a const problem I was having.
There are differences in the same code compiled between the two but the config words in the hex file are the same between them.

Ttelmah · Joined: 11 Mar 2010 Posts: 19552

That's the .LST, not the .HEX. Very Happy

That is what is 'odd'. The .LST is the same between both versions. The .HEX changes. When I read the fuses in with the programmer, there are things not being set. I'm talking to CCS (having huge hassles with something else as well...), but unfortunately Darren is away for a few days. Sad

The .LST doesn't display the values right anyway. If you look your's shows byte 3 and 4 in each word being 0. However byte 3, has to be 0xFF, since this always reads as 0xFF from the chip. If though you look at the .HEX, you have:
:105E0000FFFF00FFFFFF00FFFFFF00FFFFFF00FF9E

Which decodes as
0xFFFF
0x00FF
0xFFFF
0x00FF
0xFFFF
0x00FF
0xFFFF
0x00FF
Only one byte in four as 0x00.

It correctly has the third byte as 0xFF.

So the .LST and the .HEX are not actually agreeing... Sad

The .LST gives:

gaugeguy · Joined: 05 Apr 2011 Posts: 303

I did compare the hex files, just pasted the listing in to get human readable version to see.

gaugeguy · Joined: 05 Apr 2011 Posts: 303

This is the build version of 5.080 I have with the new dll from CCS

Ttelmah · Joined: 11 Mar 2010 Posts: 19552

Thanks.
As you say looks the same.
It may well be chip specific. I have another project on a different chip, and this seems OK.
For now I'll run with .078, and wait to see what CCS say. The other issues I'm having are more 'major' in some ways. May suggest something they have changed is 'uncomfortable' with my code...

Ttelmah · Joined: 11 Mar 2010 Posts: 19552

Just an update to this.

This fault, and another that I had found with large const arrays in PCD, have both been fixed in 5.081.