Explanation for mimium sampling frequency for MCP AC provider

Author: ArneTR

content/en/docs/measuring/metric-providers/psu-energy-ac-mcp-machine.md

@@ -18,6 +18,7 @@ We use the board implementation here: [AMD00706](https://www.microchip.com/en-us
 **Spec highlights:**
 - 0,5% Accuracy on active power measurements in a 1:4000 dynamic range (~ 3.75 mW minimum resolution)
 - 8 MHz internal processing clock (theoretical maximum of 12.5 us resolution)
+    - Power is averaged over multiples of 20 ms (Selectable from N={1,2,3,4})
 - Internal accumulation of energy in register (currently not active in our setup. We use active power only atm.)
 
 
@@ -61,3 +62,10 @@ Where:
 
 Any errors are printed to stderr.
 
+### Remarks on the minimum average resolution on 20 ms
+
+Although the chip has a very high internal sampling frequency it far higher than 20 ms it does not output measurement data from smaller time frames.
+
+The effective retrieved data is always averaged over a minimum of one full grid power frequency cycle (typically 50 Hz / 20 ms). The reason for that is that the *Active Power* is defined as the integral over one full cycle
+
+Technically you can also analyse only parts of a full-cycle. However it has no meaning for the *Active Power* as we want to analyze the consumption of the software and not small spikes in AC load that however might be negated with inductive or capactive induced phase-shifted surges later.