Dear John.
From your earlier statements I assumed, that you do not need a signal for the 3 seconds,
where the peaks occur. For this scenario a value of 0 at parameter TG is freezing the last
input value.
Now I understand, that you need the input for a high alarm observation so forget the earlier
recommendation with value 0 or with function block VC. This will not work.
You still may use the smoothing filter but you have to "play" a little with the parameter TG.
The impact of this parameter is not directly predictable. A higher value results in a smoother
and more delayed output, a lower value results in a more rippled and shorter delayed output.
If you have an input signal which jumps from 0% to 100% a TG value of 30 means, that
the output will reach 63% after 3 seconds. This is predictable but you will not have such
inputs in reality.
A second option to do your high alarm observation is to use an on-delayed timing relay.
If the input signal is higher than the set point for more than 3 seconds it will cause an alarm.
Regards
Maikky