Bars: inverter-hours in comm loss per station (Unity Communication Error state). Lines: minutes
in which the PCC reactive power stepped more than kVAr per second on average, and minutes in
which a logger's Q setpoint was oscillating. Symbols on top mark days with an intervention indicator (see below).
Click a day to list its outages.
Outages by RS485 bus
One row per inverter, grouped by the RS485 bus it sits on and listed in chain order from the EMU outward
(). β marks the last inverter on each bus, the one the drawing shows with the termination
resistor. Bars are the device's state to the second; a blank row means it was running and answering. The strip on top is the
PCC reactive power (minβmax envelope; zoom in below h and the actual 1 s trace, or 1-minute means
where no 1 s data is stored, is drawn over it). Scroll to zoom (down to seconds), drag to pan, or use the buttons; hover a bar
for exact times, click it to open the signals around it in the Zoom panel below.
Zoom:
Comm-loss episodes
Outages starting within s of each other are one episode (a cascade along one
COM port shows as one row). Spread is the time from the first to the last inverter dropping; spacing is
the median time between consecutive drops. The Q-step columns are the PCC's mean |1 s step| (kVAr/s) in the
10 minutes before and the worst minute during (both minute averages); Q swing is the largest max β min of PCC Q within one minute (kVAr). Pattern: cascade = inverters dropping one after another
about 10β20 s apart (the logger timing out each one in turn on a polled bus); simultaneous = all within 3 s
(the logger, its uplink or the whole station going away). Click a row to zoom in.
Zoom
Outage records in this window
One row per device and state record, to the second. Offset is from the first drop in the window;
gap is the time since the previous device dropped. No reading rows are periods in which the logger polled the
inverter and got no answer (work state 0xFFFE). They often follow a Communication Error while the link recovers, and
sometimes appear on their own as short timeouts that never became one.
Firmware and intervention indicators
Key stops (someone stopped the inverter)
Daytime restarts, with what preceded them
Many inverters changing state in the same second (other than comm loss / recovery)
Serial number / software version changes
Share of daylight running time in dispatch run (following a power setpoint), per inverter and week
With the plant curtailed, inverters that receive and apply setpoints spend part of the day in
dispatch run. A station whose share collapses while the others don't has stopped following setpoints (or reports
its state differently). Blank: no daylight readings that week.