No need for complicated energy measurements. Just connect the output to the input and then it should run on its own, maybe driving a motor for some days, without any battery or other external power supply. No one built such a device so far (with independent prove that it works), so all such "free energy" devices are bullshit.
I was about to post something similar. Doesn't even need to be a "power-hungry" motor. A single LED would be sufficient. To decouple (electrically) the output that is fed back to the input, simply use an isolated DC/DC converter. It's quiescent current is very likely to be lower than what his 20+ LED's use. So all that the circuit has to power is the isolated converter and a single LED. Add a buffer cap at the circuits input. Once it is running, flip a switch that toggles the input between the battery and the output of the converter.
If, as that guy claims, the circuit is producing more energy out than in, the LED should stay lit indefinitely.
Also, another good hint that "more energy out than in" is not real: the vast majority of our electricity is produced in fuel burning power plants. If any of that overunity stuff were real, those companies would be all over it. Imagine, they could charge you the same price for electricity, while not having to spend a penny on fuel. Which means way more profit for them. And they are good at scaling things up as well. Doesn't matter if all a single, simple circuit can drive is a few mA at a few volts. They would simply scale it up massively, be it by making the circuit itself beefier, or by simply massively paralleling millions of them.
That none of these two things ever happened is a very, very good indicator that this "overunity" stuff is bullshit. No small scale demo that runs for a year without any battery, nor any electricity companies increasing their profits that way.
Greetings,
Chris