On Sunday 15 June 2003 07:29, Olin Lathrop scribbled:
> Don't look to the app notes for examples of well written code.  I've se=
en
> too much bad coding to trust any of them.  Personally, I think they are=
 a
> waste of time at best, downright misleading at worst.

I'll actually have to agree with this, as I've found other (unrelated) ap=
p=20
notes w/code that was obviously not tested, and could not even compile.


> That won't prove anything.  The problem is subtle and small, so it may =
not
> occur, or you may not notice if it does occur a few times.  Eventually =
it
> may cause long term time drift.  Trust the data sheet, ignore the app
> notes.
> ...
> It *might* be OK to write the high byte when you know a carry from the =
low
> byte can't happen.  However, the data sheet is unclear on this, and I
> wouldn't count on it.

Unfortunately, the datasheets have had their share of errors as well.  I =
think=20
the key here is to find out *why exactly* Microchip indicated that writin=
g to=20
TMR1H should be done carefully.  I may have to give them a call this week=
=2E =20
So far, in about 14 hours of testing, it's keeping excellent time.  But t=
his=20
technique aside, I do like, and can agree with being more trusting of the=
=20
technique that monitors the upper bit(s) on TMR1H.  Should be easy to=20
setup/test, as I was doing that already (a separate app) with my own coun=
ter=20
register, and just need to strip out a bunch of code and read TMR1H inste=
ad.

Cheers,
-Neil.

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu