Same here... I have a product using a 12C509, designed in 1999, still in pr=
oduction=20
now. Likewise, it uses a single input to set a mode of operation (would you=
 believe=20
I'm also using GP3?), and in my case totally relying on internal pull up. P=
CB track to=20
ground is cut or un-cut at assembly stage, covered with conformal coating t=
o reduce=20
problems if (when) moisture is present. Program reads the input only at pow=
er up to=20
decide what mode it's going to operate in. Nothing fancy... I only enforce =
a 1ms=20
delay after power up to make sure the pull-up has had more than enough time=
 to=20
overcome capacitance on the input that might for example be exagerated by=20
moisture... could have done a proper debounce but in this case didn't bothe=
r.

Despite my example above, good debounce is essential where it matters and g=
ood=20
practice everywhere else. Additonally I've sometimes used a schmitt-trigger=
-like=20
software implementation: sample the input (or result of debounce routine) a=
 number=20
of times and keep a rolling tally of the number of 1's read. If 1's count >=
 80% then=20
read as 1, less than 20% then read as 0. Any count in betweeen means no cha=
nge,=20
and this dead-band gives a considerable amount of resilience to extreme=20
interference. Combined with appropriate pull-up resistor and hardware filte=
ring of=20
course.

AFAIR an internal pull up is somewhat im-precise but usually equates to 50k=
 or=20
more. So not suprising to get a 1 in a million false reading... depending o=
n electrical=20
noise in the environment, the odd charged particle passing through, phase o=
f the=20
moon, presence of dead fish being waved. I'm not sure if it I'd expect the =
resistance=20
to change with age, though if the input was exposed to static then perhaps=
=20
progressive damage might show up like this?

On 25 Aug 2016 at 21:54, James Cameron wrote:

> Hit an amusing one today; an aging 12C509 still in production, with
> code from 2001, which reads GP3 to set reporting units, which had GP3
> floating by my soldering error, had begun reading a zero occasionally;
> about one in a million reads.
>=20
> So I added the missing strap.  Let's see how it goes.
>=20
> Hmm.  But pull-ups are enabled.  Either pull-ups are damaged, or the
> circuit was picking up stray RF.  The GP3 trace is about 2cm long.
>=20
> Relevant code selective copy and paste;
>=20
> 	;; set option register to enable pull-ups and enable GP2
> 	MOVLW	B'10011111'	; NOT_GPPU, T0CS
> 	OPTION
>=20
> ...
>=20
> 	MOVLW	B'00111011'
> 	MOVWT
>=20
> ...
>=20
> 	;; test for fahrenheit mode, gp3 low
> 	;; (if left open, it is pulled up to high)
> 	BTFSS	GPIO,3
>=20
> ...
>=20
> --=20
> James Cameron
> http://quozl.netrek.org/
> --=20
> http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
> View/change your membership options at
> http://mailman.mit.edu/mailman/listinfo/piclist
>=20



--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.