Brown-out fuses on PIC18F8722

[Gavin_Minion]

Good Morning,

I have encountered an issue when compiling code for the PIC18F8722 with the fuse BORV42 set.

The version 4.122 compiler gives an error "Unknown Keyword" for the fuse. The code was previously compiled correctly under version 4.101. The code also compiles correctly if the fuse BORV45 is used.

the header for the device lists the following fuses as supported (amongst others):
//////// Fuses: WDT128,WDT,NOWDT,TO,BORV27,BORV42,BORV45,PUT,NOPUT,CPD,NOCPD

Whilst investigating this issue, I have discovered that there is an issue with using the brown-out detection specified by BORV42 on PIC18F7822 processors running at full speed (40MHz). This may be related, but does not explain why the fuse is not valid - it should be.

Does anyone have an explanation for this error?


I am using CCS compiler version 4.122 and

[Ttelmah]

Do you have the IDE?.
They appear to have left out a few fuses at various times on various chips. This one is missing on the 4550 as well. With the IDE, you can simply add them using the device editor. I suspect the .h file list was last rebuilt on an earlier compiler version that had the fuse....
If you don't have the IDE, moan to CCS, they should give you a fixed .dat file in only a few hours. Tell them anyway, since then it'll get fixed on future versions.
I have had significant brownout problems on a number of recent chips. Had to return over 5000 18F4550's, where BORV45, was having chips reset at 4.9v!. Have seen similar extreme values on two other chips as well. Microchip seem to be totally 'aware' of this, so your BORV42 problem may be something similar.....

Best Wishes

[jeremiah]

It's not a valid fuse for that particular chip. I just checked the data sheet and you have four choices:

From, page 380 of the datasheet, here are the typical brownout voltages for each fuse setting.

[Gavin_Minion]

Good Morning Jeremiah,
You are quite correct that it should not be a valid fuse. However, I did try BORV43 and it gave the same error.
(I also tried BORV40,BORV41 and BORV44 too!)

Thanks Ttelmah for the information, I have checked and found that the only fuses present in the IDE are BORV20, BORV27 and BORV45.

Now I just have to decide whether to implement BORV43 (and risk the device becoming unstable before a brown_out reset if the voltage drops to 4.11v) or use BORV45 (and risk the device resetting at 4.82v, which is quite possible with the regulators specified for the board)

Decisions, decisions...

[Ttelmah]

Remember you can also change the reset voltage, by just adding a #ROM statement to set this one fuse word. This then is a one second removal job when CCS get it right....
Just 'FYI', this particular voltage is listed as '4.2v', in MPLAB, not 4.3v. The values are very nominal, and I presume the 4.2v 'name' was selected for compatibility...
So just add (for example):

[jeremiah]

Gavin,

The CCS IDE has a tool that tells you all the valid fuses it knows for a particular part (I believe it is called "Valid Fuses"). Did you check that to see what it thinks the fuses should be named?

[Ttelmah]

[Gavin_Minion]

Thanks again Ttelmah,

I decicded to add the fuse BORV42 to the IDE to retain compatibility with the stored header and with earlier versions of the code (better than launching a change procedure for the code just to get the fuse changed!)

On balance, I think it is safer to retain the BORV42 setting and hope that the brown_out reset occurs before the processor becomes unstable (which is an unlikely situation) rather than going to BORV45 and risking the brown_out occuring at 4.8v (which is much more likely).

In the long run, I will solve the problem by reducing the processor speed in the hardware design. At 30MHz, the processor will be stable at all (running) voltages with the BORV42 setting.

Many thanks everyone for your assistance.