Tested the 100/20 with a lab power supply on the PV input for convenience of control. A single 12V 18AH AGM battery with a lightly loaded 1000 watt inverter (approx 40 watts) was used to drag the battery down. The battery was connected to the device to allow it to detect and store the voltage before tests were started.
With no battery connected to the controller at 'sun up', and a meter on the load output terminals and PV input being slowly raised to simulate normal conditions the load output stayed low until PV exceeded the +5v requirement for the Victron to start outputting current and then it went to regulated float, regulated 13.8
With a battery connected, and PV at a low level the battery was disconnected with almost no current flowing to the battery and the voltage rose up to 17.9 before dropping back down as capacitors discharged to 13.8. This only happened once, I could not repeat the result, even with various voltage and current limits from the lab power supply to the Victron. Even going to 5 amps into the battery and disconnecting it, the output resolutely stayed within limits, never exceeding absorption stage voltage. It may have been the cheapie lab power supply having its own issues and briefly yanking the PV input quite high, with the Victron doing its best to contain it.