Detail for my Mk2_3phase_RFdatalog_4.ino sketch

/* Mk2_3phase_RFdatalog_4.ino
 *
 * Issue 1 was released in January 2015.
 *
 * This sketch provides continuous monitoring of real power on three phases. 
 * Surplus power is diverted to multiple loads in sequential order.  A suitable
 * output-stage is required for each load; this can be either triac-based, or a 
 * Solid State Relay.  
 *
 * Datalogging of real power and Vrms is provided for each phase.
 * The presence or absence of the RFM12B needs to be set at compile time
 *
 * January 2016, renamed as Mk2_3phase_RFdatalog_2 with these changes:
 * - Improved control of multiple loads has been imported from the 
 *     equivalent 1-phase sketch, Mk2_multiLoad_wired_6.ino
 * - the ISR has been upgraded to fix a possible timing anomaly
 * - variables to store ADC samples are now declared as "volatile"
 * - for RF69 RF module is now supported
 * - a performance check has been added with the result being sent to the Serial port 
 * - control signals for loads are now active-high to suit the latest 3-phase PCB
 *
 * February 2016, renamed as Mk2_3phase_RFdatalog_3 with these changes:
 * - improvements to the start-up logic.  The start of normal operation is now 
 *    synchronised with the start of a new mains cycle.
 * - reduce the amount of feedback in the Low Pass Filter for removing the DC content
 *     from the Vsample stream. This resolves an anomaly which has been present since 
 *     the start of this project.  Although the amount of feedback has previously been 
 *     excessive, this anomaly has had minimal effect on the system's overall behaviour.
 * - The reported power at each of the phases has been inverted. These values are now in 
 *     line with the Open Energy Monitor convention, whereby import is positive and 
 *     export is negative.
 *   
 * February 2020: updated to Mk2_3phase_RFdatalog_3a with these changes:
 * - removal of some redundant code in the logic for determining the next load state.
 * 
 * July  2022: updated to Mk2_3phase_RFdatalog_4, with this change:
 * - the datalogging accumulator for Vsquared has been rescaled to 1/16 of its previous value 
 * to avoid the risk of overflowing during a 20-second datalogging period.   
 *
 *      Robin Emley
 *      www.Mk2PVrouter.co.uk
 */