It seemed so simple.  I wanted a yellow LED in my PIC16C84 design, so I
connected it via a 3K3 resistor between RA4 and Vdd.  So far so good.

Then I noticed something odd.  With the PIC spending most of its time in
sleep mode, its quiescent supply current (with the LED off) varied from
30uA to 90uA in sympathy with the ambient light level - disconnection of
the LED removed this variation.  A voltmeter (>1M impedance) showed RA4
to sit at +3.6V relative to Vss when its open-drain output was inactive.

It seems as if the photocurrent injected by the LED is entering RA4 and
affecting the operation of the PIC (maybe via the Schmitt input buffer).

Several questions spring to mind:

1.  What is causing the supply current variations?
2.  Is there any risk of misoperation of the PIC in this configuration?
3.  Would a non open-drain port pin be a better choice to drive an LED?

Any thoughts?

- Ian Chapman