| View previous topic :: View next topic |
| Author |
Message |
2xhp
Joined: 14 Jan 2019
Posts: 39
|
| WDT times correct for only half the settings |
 Posted: Thu Jan 21, 2021 11:35 am |
 |
|
Hi,
I'm using a dsPIC33EP64GS804 with compiler 5.100. I get the correct WDT response for half the settings tried. Looks like the ones with a 1 in the LSB are using the next higher time. Results:
WDT_16MS: 32 ms reset
WDT_32MS: 32 ms reset
WDT_64MS: 128 ms reset
WDT_128MS: 128 ms reset
WDT_256MS: 512 ms reset
WDT_512MS: 512 ms reset
WDT_1S: 2000 ms reset
WDT_2S: 2000 ms reset
WDT_4S: 8000 ms reset
WDT_8S: 8000 ms reset
Wondering if this might be a compiler bug or if I'm missing something in the WDT setup somehow.
Thanks in advance. |
|
 |
temtronic
Joined: 01 Jul 2010
Posts: 9247
Location: Greensville,Ontario
|
|
| Posted: Thu Jan 21, 2021 3:50 pm |
|
|
You should post compiler version as well as the list of fuses ( btm of the listing)...
Those that use that PIC can easily figure it out.... |
|
|
2xhp
Joined: 14 Jan 2019
Posts: 39
|
|
| Posted: Thu Jan 21, 2021 5:59 pm |
|
|
| temtronic wrote: | You should post compiler version as well as the list of fuses ( btm of the listing)...
Those that use that PIC can easily figure it out.... |
Configuration Fuses:
Word 1L: FFFF NOWRTB NOBSS NOBS NOWRT NOPROTECT NOWPCFG NOPROTECT_CFG NOAIVT
H: 0000
Word 2L: FFFF
H: 0000
Word 3L: FFFF
H: 0000
Word 4L: FFFF
H: 0000
Word 5L: FFFF
H: 0000
Word 6L: 7FFF
H: 0000
Word 7L: FFF8 FRC IESO
H: 0000
Word 8L: FF7A HS OSCIO IOL1WAY CKSNOFSM PLLWAIT
H: 0000
Word 9L: FFC9 WPOSTS10 WPRES32 WDT_SW WINDIS WDTWIN_25%
H: 0000
Word 10L: FFFF
H: 0000
Word 11L: FFDE ICSP2 NOJTAG NODEBUG NOBTSWP
H: 0000
Word 12L: FFF7 PWMLOCK ALTI2C1 NOALTI2C2 NODBCC
H: 0000
Word 13L: FFFF ALTWREG1=NO ALTWREG2=NO
H: 0000
Compiler version is 5.100. |
|
|
temtronic
Joined: 01 Jul 2010
Posts: 9247
Location: Greensville,Ontario
|
|
| Posted: Thu Jan 21, 2021 6:16 pm |
|
|
Looks like Word #9 has the watchdog settings ?
Though I don't use that PIC, it looks like the watchdog has pre and post scalers as well as a 'window' range ?? Also can be enabled by software ?
I don't see any obvious 'wdt-value-ms' settings, so does the compiler take your xxxms and convert into pre/post values ?
If so, that may be how the times get messed up.
Again I don't know, just thinking of possible things to look at.
dsPICs are mighty powerful, beyond me...18's are challenging enough for this dinosaur ! |
|
|
2xhp
Joined: 14 Jan 2019
Posts: 39
|
|
| Posted: Thu Jan 21, 2021 6:22 pm |
|
|
| temtronic wrote: | Looks like Word #9 has the watchdog settings ?
Though I don't use that PIC, it looks like the watchdog has pre and post scalers as well as a 'window' range ?? Also can be enabled by software ?
I don't see any obvious 'wdt-value-ms' settings, so does the compiler take your xxxms and convert into pre/post values ?
If so, that may be how the times get messed up.
Again I don't know, just thinking of possible things to look at.
dsPICs are mighty powerful, beyond me...18's are challenging enough for this dinosaur ! |
I thought maybe it was a pre or postscaler. I did try various fuse options for that and no change. But the fact that half the times are correct and the other aren't (and they are adjacent times that are identical) leads me to think it's something outside of my control (i.e. compiler related). But I'm new to the dsPICs, so could very well just be missing something.
Thanks for the thoughts. |
|
|
Ttelmah
Joined: 11 Mar 2010
Posts: 19553
|
|
| Posted: Fri Jan 22, 2021 1:23 am |
|
|
On all to these chips, the watchdog 'timing' specified in setup_wdt, has to
be translated into fuses. On most of the chips, MicroChip give a table of
the equivalent timings for the prescaler/postscaler combinations. On this
one they don't.
If you look, they use bit 0 for the WDT ON/OFF flag, then have the count
values. They correctly start these at 2. What you have to do is not use the
WDT_ON flag.
Basically:
setup_wdt(WDT_1S);
Without using the WDT_ON flag, does correctly configure one second.
You only use the WDT_ON or WDT_OFF flag "on it's own", to enable or
disable the watchdog. Don't 'OR' this into the value being used.
A quick check, shows the above line gives:
| Quote: |
Word 9L: FFCA WPOSTS11 WPRES32 WDT_SW WINDIS WDTWIN_25%
|
Which gives (32000/32)/(2^10) = almost exactly 1 second
Using:
setup_wdt(WDT_2S);
gives
| Quote: |
Word 9L: FFCB WPOSTS12 WPRES32 WDT_SW WINDIS WDTWIN_25%
|
The correct 2 second setting.
Generally where flags are to be or'ed in, they do tell you this in the
header file. This one does not say to do this.... |
|
|
|
FAQ 
Search
Register
Profile 
Log in to check your private messages 
Log in
