about the sample and hold, I started reading their principle of working and I was thinking about using a LF198 at the output of the first circuit... I'll study it more to grasp the concept (I'm not an electronics professional, it's just a hobby, my knowledge is very lacking...)

"...a bit of code that say "If the voltage dropped from nominal to <? in n readings, change the last n readings to zero". This is a bit more code, but much simpler hardware."

I'm doing this already, but I think is prone to misleading records... I'm doing 32000 readings, averaging, then recording. with a 2k2 resistor and a 220uF capacitor I have a steady reading and a "fast" drop when a power outage happens. but murphy states that the power outage will happen in a given moment such it a)will not drop fast enough, b)will mess my recording :-)

by the way... murphy won't say that my peak detector will be discharging when a power failure occurs, fooling the tiny capacitor in the sample and hold and messing my record too? the same problem in a more sophisticated version?