Since this is my circuit being discussed I will answer the question regarding measurement error with this specific configuration by saying  "is there any error"?  TK may have been tongue-in-cheek but was he?

The potential measurement error with this type of circuit (a reactive to real power converter) comes from two sources.  First, the high ratio of reactive to real power and second, the requirement of the scope used to discern "minutes" of phase angle measurements.  This requires extreme care when deskewing the probes used (particularly the current measurement) plus probe connections and layout.

Are these converters capable of producing more energy out than in, yes they are but with a caveat in determining if they are or not.  The input sinewave is the point of reference for the input power.  However, in the test TK shows and in nearly all others I've done, this sinewave was generated by some outside means such as a signal generator for low levels or a pulse generator driving a resonant L/C circuit.  The former is usually produced by Direct Digital Synthesis which may or may not require more energy than is produced at the output, but the later always requires more energy.

Herein lies the problem to be solved IMO!

Pm

I made a quick replication of Partzman his bifilar transformer (PBT) as being presented by Tinselkoala in his video
here:  https://www.youtube.com/watch?v=MNzbc-N-e9c

Using almost the same setup and the same measurement techniques according to the below diagram.

My values are slightly different, but the measurement points are the same:

R1= 2x10 Ohm 1% inductionfree in series = 20 Ohm
R2= 1 Ohm 1% inductionfree
Frequency= 1.44Mhz sine wave
FG isolated via isolation transformer plus running from a battery
CH1 (yellow) 1.9V Cycrms
CH2 (Blue) 47mA Cycrms
CH3 (purple) 969mV Cycrms
Math (red) CH1 x CH2 = 43mW
Phase difference CH1 => CH2 = 61°  (blue = current = leading)

As my scope only can present 4 measurements at a time, see the below 2 screenshots for the 5 important measurements.

Scope Math (red) calulates the input to be 43mW
Manual calculation for input shows Vrms x Irms  x Cos (phi)   = P ave
                                                     1.9  x 0.047 x Cos (-61) = 43mW       (same as scope instantaneous calc's)

Pout = U²/R
        = 0.969²/21
        = 44mW

So my setup and calculations comes to a COP=1

Video here: https://www.youtube.com/watch?v=wuGGHpAmdK4

Regards Itsu

...
So my setup and calculations comes to a COP=1

Video here: https://www.youtube.com/watch?v=wuGGHpAmdK4

Regards Itsu

Hi Itsu,

Thank you for presenting us your experiment, which was obviously carried out with care. I'm not really surprised about the COP.
So now the COP difference is due either to a measurement error on the TK side or perhaps, last chance, to a particular difference between your setups, to be identified.

Ah you calculated ch3 ^ 2 / (R1 + R2), yes that should be right.

COP exactly 1 is very weird though, that a circuit is that efficient. That looks like an anomaly, worth to research more. In COP 2.9 i don't much believe though, i think if overunity would be found, it would be small, the capacitance is small, etc.

So 2.9 times measurement error, that's unimaginable. My result also went closer to COP 1, though always remained slightly higher than 1, but i think it's highly possible that when making it a bit more precise, it goes below 1, so it's not a significant result, a possible measurement error was finally higher than the possible overunity. But i used Hitachi analog oscilloscope and only started to develop my method of digitizing trace data using Inkscape. But they used top notch digital Tektronix, with all the capabilities.

Itsu,

I have several comments regarding your TK replication.  First, the phase angle is quite low which would mean the Q of your coil is low and/or you might carefully check the dot polarity of your coil connections.

Second, let's look again at your measurements and calculation.  You measure Vin = 1.9v crms, Iin = 47ma crms, Vout = 969mv crms, and the input phase angle as -61.85 degrees.  Therefore the Pin is 1.9*47e-3*cos(-61.85) = 42.13mw and Pout is .969^2/21 = 44.71mw for a COP = 44.71/42.13 = 1.061. 

Since these voltage and current measurements were taken on a cycle basis, I would be somewhat confident in the result.

However, the Math channel calculation may possibly be inaccurate depending on whether it was measured between the cursors or full screen.  You might want to check what cursor setting you had during these measurements.

Regards,
Pm

Itsu,

OK, thanks for re-running your test and I would conclude that for some reason your bifilar coil has a somewhat reduced Q or possibly some other problem that is affecting the overall performance.  For comparison I have attached  the test results of a pcb based pancake coil.  The schematic is identical to that posted by TK and yourself with  one exception, a mosfet driver is providing a symmetrical square wave pulse with a 20v peak to a 35uH inductor which resonates with the distributed capacitance between the windings.  The resulting 1.7 MHz sine wave seen on CH1(yel) is the reference measurement input waveform.

The load is 20 ohms and the CSR is 1 ohm both non-inductive.  The probe used for current sensing is attached directly to the CSR with an RF spring and probe tip.

In  the first scope pix below, CH1(yel) is the input waveform, CH2(blu) is the voltage across the total load of 21 ohms and CH4(grn) is the current through the 1ohm CSR.  The Math(red) channel is the calculated mean input power as measured between the vertical A&B cursors.  Statistics of the measurements are shown and continuously averaged over 16 cycles with a 1Ms sample rate for the horizontal window.

As seen, the mean input power in the Math channel ranges from 411.9mw min to 450.8 max.  The statistical average output voltage across the load is 4.94v rms resulting in a Pout = 4.94^2/21 = 1.162 resulting in an apparent COPmin = 1.162/.4508 = 2.58 and a COPmax = 2.82.

The next scope pix shows the data for CH4 to allow comparison between the Math channel and measurement results using the input voltage/current phase angle.

Here we see the CSR current is 234.5ma rms.  Therefore, the Pinmin = 98.01*.2345*cos(-88.91) = 437.2mw and Pinmax = 98.01*.2345*cos(-88.28) = 689.8mw.  This should illustrate the demanding requirement for phase angle accuracy and how it affects the input calculations.  So, we see an apparent COPmin = 1.162/.6898 = 1.68 and a COPmax = 1.162/.4372 = 2.66.

A pix is also attached of the pcb based coil assembly.

Regards,
Pm

PM,

thanks for the detailed info, allthough you use 4.94V in your Pout calculations (Pout = 4.94^2/21) while the
screenshot shows a mean of 4.694 in the blue data, the COP still is high and similar as TK reports.

But indeed, a slight deviation of the phase has a big influence on the end result.


I have a second identical bifilar pancake coil which i will use to double check my COP = 1 results.
I cannot think of what the cause is for the low (61°) phase angle in my present coil.

Itsu 

You're right, ayeaye. The measurement method is wrong, not only Pout but also Pin.

The output power is not equal to CH3²/(R1+R2) but is equal to Pout = (CH3-CH2)²/R1 + CH2²/R2 as you stated.
And the input current is not given by CH2 because current from R1 is added so Pin is wrong too.

(As I may be wrong too, please guys explain your method with Pout=CH3²/(R1+R2) while it's not the same current in R1 and R2).

For me, if you want use the voltage at R2 terminals to know the input current and have Pout=CH3²/R1 + CH2²/R2 and Pin=CH1*(CH2/R2) then you have to connect the probes this way:

Power calculations of the circuit attached to your post cannot be calculated using a scope with common grounds,unless the P/in and P/out are calculated separately.

The only way to measure P/in and P/out at the same time,using a scope,is to have a scope with isolated grounds.

P/in should be calculated by voltage across the primary of the isolation transformer,and current through the primary of the isolation transformer,as it is part of the circuit,and where the FG is the source.