In new firmware v5.39 we introduce a new method for filtering the data coming from the analog / sensor inputs (temperature, humidity, alarm, etc.) - "Moving Average" filter.
MA Filtering   This is one of the most optimal methods for removing noise in measurements, which uses averaging of recent measurements (via a floating window). The length of the "window" is selected in the interval 2 ^ N, where N = 0 (no filtering) ... 8 (maximum filtering). Averaging more measurements results in better smoothing of data, but reduces response time (very short changes will not be recorded).
   The other significant change is that the data from this filter is now used for absolutely all services that extract data from the inputs: Automation, Web, MQTT, SPC. In the previous filtering method (which is not a real filter, but rather a delay in the reaction) only the Automation block used it, and all other services used direct data from the inputs, without any filtering.
   In SNMP, the "ioValue" object is preserved in its current form and uses direct data (in order to maintain backward compatibility with already built systems that rely on speed), but a new "ioValueFiltered" object is added, which uses the data from the MA filter.