>      Anybody got a good averaging algorithm?

>      Snip------------
>
>      (newbee protection on)
>      It seems like you could take readings pretty fast, and then
>      save the measurement only when 'N' consecutive readings are the
>      same.  This would ignore spikes, and still track changes.
>      (newbee protection off)

I seem to remember trying this once, and being amazed that I could
get ten spikes in a row.  Of course, I have a test I run that
specifically challenges things with hash and noise.  It consists of a
Rube Goldberg piece of plywood, on which is mounted two chattering
relays, a couple of universal motors, a spark igniter and a lamp
dimmer.  I figure you might see all these things in an average home.
Best Regards,

Lawrence Lile