A better experiment is a double blind trial. Fill one hydrogen tank with standard hydrogen and another tank with deuterium. Code both of them with a scramble. The e-cat should fail with deuterium and should work with hydrogen.
This does not require any access to internals, and could even be done on the sly without alerting rossi.
This is a typical thing with "free energy" scams. The same thing was tried back in the "piezos" era, when putting resonant piezoelectric devices in an oscillating tank circuit was supposed to produce more power than went in. (But it did not, of course, run self-sustaining, without outside power).
Measuring wattage of unusual waveforms is hard. W = E*I, or watts equals current times voltage, but you have to integrate the instantaneous product over time. Many wattmeters don't do a full integration; they're mostly for sine waves, and may have frequency limitations. The E-Cat guy won't let observers use a scope to see that "proprietary waveform". All it's doing is driving a resistance heater, so the waveform should not matter to the heater.
A simple RF wattmeter, which has a resistor whose heat output is measured, would clear this up.