STEORN DEMO LIVE & STREAM in Dublin, December 15th, 10 AM

[Omega_0]

About lumen's idea that some energy is getting through the input without being measured, I think that a good way to verify this will be to use a big cap as power source. Suppose its charged to Vc volts then it has 0.5*C*Vc^2 J in it. Let it drive the transformer in "OU mode" for a while and measure the output as usual. But this time, ignore the Iin*Vin, instead disconnect the cap and measure its voltage again, which should give the remaining energy and the difference with initial value would be your input energy.

Note that we are working with energy here, not power. So time should be measured accurately. Let me know what you think about this method.

[gyulasun]

...
Further, capacitive coupling between the windings of the transformer seems far fetched to me. There may be such a thing but I've never heard of it. A reference or link would help.
...

Most of the transformer equivalent circuits do not include any capacitances of a real transformer , mainly because the low frequency equivalent schematic is considered and at low frequencies the effects of capacitances are small and negligiable.  So you have to search for wideband equivalent transformer circuits like for instance audio transformers.
See Figure 13 and Chapter 1.2.3 in Page 8 of this PDF file:
http://www.jensen-transformers.com/an/Audio%20Transformers%20Chapter.pdf
Then go to Figure 21, Page 12,  there is a 'bump' in the frequency response depending on the residual damping resistance.  I think lumen may have thought of such 'peculiarities' in amplitude response in the funtion of frequency.

Here is another link, the chapter deals with pulse transformers and the coils self capacitances (the step-down transformer is also dealt with).
http://www.vias.org/eltransformers/lee_electronic_transformers_10_02.html
and also for audio transformers:
http://www.vias.org/eltransformers/lee_electronic_transformers_06_09.html

Also, if you look at Fig. 107 here:
http://www.vias.org/eltransformers/lee_electronic_transformers_06_07.html
then Figure (d) shows both a mutually and an inductive bottom coupled double LC equivalent circuit for the high frequencies, and as such it must have a "selective" frequency response.

I do not mean to suggest that the phenomena you have found is a simple resonant effect for your transformer (like the 'bump' referred to above). It needs further testing of course.

rgds, Gyula

[Omnibus]

Omega_0,

Can't agree more that more heads are needed and, as I said, I'll do my best to have this reproduced independently. Otherwise it has no value as almost anything else we're seeing that is kept secret which prevents third party replication. The idea to have Naudin reproduce it is very good but I'm not sure he has the equipment to do so. I've never seen him carry out quantitative energy balance.

As for the heat, I agree that a direct method of determination is needed which will also account for the core losses but the input energy is hard to measure (more so than even measuring the heat) because the period one full curve is registered by the scope is too short for a meaningful calorimetry. It has to be short in order to be able to integrate it properly with all the details. On the other hand, the time to drive the device should be long so that enough energy is put in to have the temperature raise sufficiently for an accurate calorimetry. I was thinking, maybe, once the exact joules are known for the short period of the curve registration then one can use an external timer. Once the extended time period, necessary for good calorimetry, is accurately measured one can divide into it the shot time interval from the scope. The result from this division times the joules for each short time  period (the registration time period of the scope) will give the overall input energy spent to do the calorimetry. Will have to think more about it.

As for your cap idea, I don't see how it can be applied straightforwardly because the experiment is done with AC. Unless some converter is added but I have no idea how to do that let alone that there will be losses in the converter which will be hard to account for. The solution to @lumen's idea would be to avoid the current common ground approach and measure the voltages with differential probes and the currents with Hall effect current probes. This amounts to spending no less than ten grand, if you are to use really good probes. That has been a problem from day one, as you know.

[Omnibus]

@gyulasun,

Thanks a lot. These are great links and I'l have to study them more thoroughly. The fact remains, however, that once the measured input and output currents and voltages are the real ones and the calculations indicate OU then circuit mechanisms of energy exchange are immaterial regarding the reality of OU. These mechanisms can only aid in explaining where the OU comes from but will not affect its reality. So, the struggle here is to ascertain that what we're measuring are the true I's and V's.

Therefore, @lumen's argument has to be understood better and addressed if it can be a mechanism of unaccounted for error in measuring current, as can the parasitic inductance be. You were the first to ask for adding capacitances to chase away eventual resonance from the 47kHz point. I did place capacitance in the 0.001uF range in parallel with the input but that 47kHz phase shift point  didn't go away.

[Omnibus]

What really bothers me is the lagging of voltage behind the current for frequencies below 47kHz shown in the diagram I already published. It's true that the current below 500Hz is so messed up that it isn't even a sine wave but at higher frequencies it has a sine form just as the voltage trace does. And yet, the voltage is lagging behind the current while it should be otherwise. The capacitance that @gyulasun gave references for seems to kick in at way higher frequencies. So, is this some kind of measurement error or it's the effect itself. Now, these data are taken with a load on the secondary coil. I can't wait to see what would be without a load. Will it return to "norma", that is, will we again have current lagging behind the voltage as inductive coils behave? That I'll know on Sturday whe I return. Then I'll also do the experiment which @lumen suggested.