The blue things with markings "WKO 440-X1 300-Y2 1nM" that look like capacitors, are capacitors - made by Vishay and rated for 300 volts or more.
There are 6 big electrolytic capacitors. Four are 160 volt, 3300 microfarad capacitors. These are in parallel across the PV array. Two are 80 volt, 6800 microfarad capacitors. These are in parallel across the battery.
Note that all six capacitors share a common negative, and the PV array negative and battery negative are connected to this common negative. This is NOT the case ground however.
There is a row of 8 TO-220 package devices on a heat sink at the end of the power PCB. Four of these are STPS40170CT devices, which are dual Schottky diodes in a package. If we number the To-220 pins 1,2,3 from left to right then 1 is the tail of one diode, 3 is the head of the other diode. The two diodes are joined head to tail and this is brought out on pin 2, as well as on the tab of the device.
The other TO-220 devices are FDP2532 N-channel power mosfet devices. Rated at 150 volts and 79 amps! Pins 1,2,3 as above and G,D,S with D also connected to the tab. The four of these are in parallel, with their drains all connected to the input power bus and the Sources all connected to the ouput power bus. The center of all the diode arrays are connected to the output power bus. The other ends of the diodes are mostly connected to the input bus, but other interesting things seem to be done with them that I never sorted out.
Clearly the four FET's are the heart of the unit and do a PWM chopping thing on the input voltage, feeding it to a filter. The first thing the chopped output goes through is a big inductor, then it goes to a relay (that turns everything off at night so the batteries do not power the unit in reverse). After the relay we get the two output filter capacitors and then away it goes to the batteries.
This is the story for the power handling, everything else is control and monitoring.
The ground fault fuse is between this "internal negative" bus and the case ground when JP1 is installed "to the right". On our unit, the trace from the fuse (case ground) to the center of JP1 melted during some bad situation. (I suspect someone installed a fuse with size larger than the specified 1 amp).
So, what is happening here is that with the JP1 jumper installed "to the right", i.e. toward the while square on the sikscreen, i.e. toward the bottom of the unit when mounted on the wall; the PV array negative (which is also battery negative) is connected to earth ground through the JP1 jumper and the GF fuse.
With the jumper in the other position, this ground connected is broken and the fuse can never blow. This should be selected when an external negative to ground bond is required (and exists) or where the PV array should not be grounded in any way. See page 2-9 in the manual.
Tom's notes / firstname.lastname@example.org