This is the most recommended way to do it. I would suggest it too. The
low pass filter is the most effective way to detect changes in the
signal without worrying about calibration or presets. 

The only issue is that the PIC should wait some time after power on to
then consider the output of the TLC271 valid, in some way, you can see
this delay (60 seconds) in several motion detectors, including the
expensive ones.

A resistor of 4M7 connecting both inputs of the TLC, with a 4.7µF from
pin + to ground, will give you a delay of aprox 22 seconds to load,
plenty for fast triggering at the - input.

Wagner.

Lance Allen wrote:
> 
> Jinx wrote...
>  >The PIR circuit is stable, no problems in the 6 weeks it's been on so far,
> > but if I could make it self-settling that would be a big advantage when it comes
> > to installing a number of them. ie set and forget, no need to worry
> > about any drift, if it happens. Putting a smaller preset between two
> > resistors does make it easier to set, but it does still need to be set
> > manually.
> >
> 
> At a quick glance I would say the answer would be to remove the
> 10M pot and feed the non inverting input through a low pass of the
> inverting input. The low pass being of a value such as to enable
> detection over drift.
> If the input rises fast enough.... the inverting terminal voltage rises
> but the non inverting isnt there yet.... trigger!
> 
> I remember doing this sometime in my distant past.
> _____________________________
> 
> Lance Allen
> Technical Officer
> Uni of Auckland
> Psych Dept
> New Zealand
> 
> www.psych.auckland.ac.nz
> 
> _____________________________