Bifilar pancake coil overunity experiment

[F6FLT]

...
So removing the 1 Ohm csr and using the current probe in its place we get (see screenshots 2 and 3):

CH1 yellow = 1.92V
CH3 purple = 855.7mV
CH4 green  = 43.76mA
phase CH1-CH4 = -55.47°
Math  CH1xCH4 = 48.59mW


So, Scope Math (red) calulates the input to be 48.59mW   (using uncorrected phase difference)
Manual calculation for input shows Vrms x Irms  x Cos (phi)   = P ave
                                                    1.92  x 0.04376 x Cos (-55.47 + 10) = 34.88mW          (here we corrected the phases from -55.47 plus 10 to -65.47°)

   
Pout = CH3²/R1
        = 0.8557²/20
        = 36.61mW

COP = 36.61 / 34.88 = 1.049

Regards Itsu

Hi Itsu

You result is still intriguing. You said you have two bifilar coils. One is 133.8uH/131.2uH, the other 125.5uH/122.4uH, with a Q=50 @ 100Khz.

Which coil did you use, and what are their capacities and resistances? May be you already gave them and I missed it. With these values I could make a LTspice model and compare to your traces to see if there is something different.

Coils resistance provides supplementary losses so that your COP could be higher.

[Void]

Coils resistance provides supplementary losses so that your COP could be higher.

Or the COP could potentially be lower than measured as well, due to measurement error.
IME, scopes are not super accurate at lower power level measurements like this, and that is to be expected.
Another possible way to double check the measurements on low power level circuits like this
is to scale the input drive up to a higher power level of say at least about 1 Watt or higher, and measure again
in the same way with the scope. That way potential measurement error should become less significant, assuming of course
that the measurement approach is correct and one is not overlooking one or more things. 
If you are still seeing interesting results, then you may really be onto something.

I have tested with some low power circuit configurations in the past which seemed to show some
potentially interesting results, but I have put them on the back burner until I can get the chance to
test with them again at higher power levels. I have found it can be just too hard to try to draw any sort
of definite and reasonable conclusions when powering at low power levels. If powering at higher power levels,
measurement error should become a lot less of a factor (although over sights are always still potentially there),
and it also should make it somewhat easier to try to self-loop a circuit as well.

[partzman]

My previous post #265 utilized a pcb bifilar coil arrangement in which each coil was separated by the .062" thickness of the FR4 G10 glass substrate.  This substrate also served as the dielectric between the two coils which overall does not provide the most efficient results due to the lower distributed capacitance.  This post is of the same basic circuitry but uses two of the pcbs sandwiched together with two layers of soldermask providing the dielectric resulting in a higher distributed capacitance.

Two pix are again shown with the first showing the waveforms and measurements and the second showing the data for CH1(yel) or the input signal as 104.9v rms.

The following calculations from the measurements will take into account the separate power levels in the 1 ohm CSR and the 20 ohm load.  Again statistics for the measurements over time are shown plus the horizontal resolution is now increased to 5 million points.  The vertical resolution is 8 bits and the level of each measurement has been adjusted for near full vertical deflection so we can assume the vertical error to be ~1/256 = .4%.

So, for Pin we see the Math channel measures a min = 98.98mw avg and max = 129.5mw avg. 

We see the current across the CSR = 315.4ma rms and the voltage across both the CSR + Rload = 6.5v rms.  Since the output current and voltage are in phase, we can calculate the power across the CSR as .3154^2*1 = 99.5mw rms.  The voltage drop across the CSR = .3154*1 = 315.4mv rms. 

We can now calculate the power across the load resistor as (6.5-.3154)^2/20 = 1.91w rms resulting a total Pout = 1.91+.0995 = 2.01w rms.

Using the Math channel Pin results for the range of COP we have COPmin = 2.01/.1295 = 15.52 and COPmax = 2.01/.099 = 20.3.

Using the input VAR and phase angle measurements we have a PINmin = 109.4*.3154*cos(-89.22) = 469.7mw rms and PINmax = 109.4*.3154*cos(-88.88) = 674.4mw rms.  This results in a COPmin = 2.01/.6744 = 2.98 and COPmax = 2.01/.4697 = 4.28. 

The VAR = 34.5va so the reactive to real ratio is 34.5/2.01 = 17.16:1.

Regards,
Pm 

[itsu]

Hi Itsu

You result is still intriguing. You said you have two bifilar coils. One is 133.8uH/131.2uH, the other 125.5uH/122.4uH, with a Q=50 @ 100Khz.

Which coil did you use, and what are their capacities and resistances? May be you already gave them and I missed it. With these values I could make a LTspice model and compare to your traces to see if there is something different.

Coils resistance provides supplementary losses so that your COP could be higher.

F6FLT,

still using the 2th coil which measured:

125.5uH/122.4uH, with a Q=50 @ 100Khz
both 1.5 Ohm and 2.5nF capacitance between them.

Itsu

[Void]

Hi Partzman. I don't know exactly what your circuit arrangement was when doing your measurements,
but, if I understand you correctly, measuring a COP of about 20+ one way and measure a COP
of about 2+ another way seems to indicate that something is quite wrong there somewhere.