Auroratek demonstration from Bill Alek at TeslaTech conference

[TheCell]

<However, if you watch the video carefully you will (perhaps) note that part of one of his arguments has to do with how much of a sinusOIDAL trace is above and how much is below the zero reference line. Since the use of AC-coupled input brings the _average_ of any reading down, or up, to the channel's zero reference line... the data from an AC coupled scope channel cannot be used to support such arguments. >

He means the sinusodial of the POWER curve which is a product of volts and amps. And this Power-curve can be above zero or below zero whether real power is being consumed or produced.
You can have an induction motor used as generator, and coupled to the mains with using a capacitor in series.
Driving the motor over its rated rpm will serve power into the net, while loading it with a resulting rpm lower than the rated rpm will consume power. During the whole process the amps average is allways zero. Both parts voltage and amps can be DC avg=0 while the product has a value !=0
And you will only start lying to yourself if the operating frequncy is lower than the cut off frequency of your scope.So his measure method may not be used by professionals, but to judge the facts it is sufficient.
While it is not beneficial to put any dc-offset voltage into his device (I think he knows that)
If you measure the voltage trace of the coupling capacitor in your scope (perhaps with another scope) and observer only pure DC there , than your measurement method is ok and the measures values are right.
Or there can be a small ac ripple on this cap. Now you can calculate the error in the power measurement produced by this effect. But this is only the case when going below the cut off frequency of the scope.
The DC-part is get lost when the signal of the amplifier passes through is internal cap (built in the amp).
When a DC offset were there the amps would shift to a value I_DC_MAX = U_OFFSET / R_PRIMARY within a small time . Flux would go to the limit, the transformer unable to operate .

[MarkE]

The power is not a sinusoid even though he describes it as such.  If there is zero phase shift then the power waveform is a sine squared wave at twice the operating frequency.  At exactly +/-90 degrees phase shift the energy absorbed in one quarter cycle exactly matches the energy released in the complementary quarter cycle. 

An oscilloscope capable of multiplication can show the power waveform.  Some more advanced scopes  can do all the math needed to display the instantaneous and cycle by cycle power.  Otherwise a scope that outputs a cvs file can be used with spreadsheet software to do the same thing.

[TheCell]

It does not matter whether he uses a sinusodial waveform or a distorted or other.
He can use AC-coupling as long he does not provide a DC path for the signal to his transformer, for whatever reason. And his operating freq. must be above the cut off frequency of the scope.

[MarkE]

It does not matter whether he uses a sinusodial waveform or a distorted or other.
He can use AC-coupling as long he does not provide a DC path for the signal to his transformer, for whatever reason. And his operating freq. must be above the cut off frequency of the scope.

That is almost correct.  The pass band of the scope and probe combined needs to include all the signal content, and because he is measuring phase: with low phase distortion. 

[TheCell]

<and because he is measuring phase: with low phase distortion. >
I don't get it. ' with low phase distortion '
Is it the measurement method of this fluke 99 scopemeter?
Has distortion any impact on the accuracy of phase measurement?
(Well I guess it has , when it's heavily distorted, but I don't see any there . Must admit I am not trained.)