Obviously, there is a path to Vdd through the PIC.  You're probably
damaging the PIC, as you are passing current through it in an
unconventional manner.  I recommend keeping Vdd on, any time there is power
to the board :)


----------
> From: Sebastian Dols <sdols@ESEGI.ES>
> To: PICLIST@MITVMA.MIT.EDU
> Subject: strange behaviour
> Date: Wednesday, April 07, 1999 9:17 AM
>
> Ok, I have detected a strange behaviour in my 16f84 last nigth, and I
want
> to share with you.
>
> Imagine a 16f84, ceramic resonator 1Mhz, _MCLR to Vdd, five leds in
rba0:4,
> with no current limiter resistor in the led paths. portb is all input,
> interrupt in rb4:7 change (nothing strange, I think). Program coded to
shift
> led blinking, period 1 sec, change of direction on portb signal. No
problem
> to compile, no problem to run, all OK... But if NO 5V to  the PIC vdd
pin,
> if 5V (signal) is applied to any of portb pins, all the leds in porta are
> emitting light at the same time...
>
> Someone knows what happens? I suspect about my protoboard, but there is
no
> way between 5V to the leds to GND if i remove the PIC (the leds do not
emit
> in that case, of course)
>
>
> Thanks in advance, keep on PICing.