Storing Cookies (See : http://ec.europa.eu/ipg/basics/legal/cookies/index_en.htm ) help us to bring you our services at overunity.com . If you use this website and our services you declare yourself okay with using cookies .More Infos here:
https://overunity.com/5553/privacy-policy/
If you do not agree with storing cookies, please LEAVE this website now. From the 25th of May 2018, every existing user has to accept the GDPR agreement at first login. If a user is unwilling to accept the GDPR, he should email us and request to erase his account. Many thanks for your understanding

User Menu

Custom Search

Author Topic: Bifilar pancake coil overunity experiment  (Read 93418 times)

ayeaye

  • Hero Member
  • *****
  • Posts: 866
Re: Bifilar pancake coil overunity experiment
« Reply #150 on: October 18, 2018, 10:37:07 AM »
I don't know how an oscilloscope saves a csv data to the usb stick, so i have to guess. They mostly open the csv file in excel, so no one even knows how its written, no one opens it with a text editor, though it's a text file. The thing i may be missing, are the numbers there in quotation marks, or not. Of course it can be opened in excel, then a column copied to Trinket, though i feel it's like a kind of banal way. In any case the first two rows are likely header, and must just be omitted.

As i have seen, it's time in seconds and voltage in volts, both float numbers in the exponent form. I don't know why so, i think the time column can just be ignored, as the samples are all after regular intervals, check it, but it's almost certain. All that is necessary for the list, is the second column.

There should be only two ways how csv can be written, without quotation marks and with quotation marks. The following are the trinkets to extract the second column in both ways.

https://trinket.io/python/fbf69fdd0e

https://trinket.io/python/85ad4a832f

Then when the voltage is already in volts, there is no need to divide by 50, or whatever number of units in a scale. Because the unit already is volts. And write the y scale just 1. The x scale, when the time column is omitted, should then be the time interval between samples in us. The samples there are float as well, otherwise scale is float, and only multiplying by float makes it float.

Yes all the calculations can be done in excel the same way as in Python, you see how, and there may be a precision problem as i said. Oh well, the lame excel, but do it if you want it so. I still consider the standard to explain things, is Python. One can make a mistake in Python, where all variables are named, it is difficult to make it faultless in python, but then what about excel with its totally not intuitive a3, b4 stuff.

Tell me excel, or maybe tell me matlab. And then you effectively do it so that i cannot say anything at all. Because in what should i talk then, excel or matlab or python? I don't understand why in hell they recommend to open these csv files in excel, people for whom precision should be so important. Excel is not proper for that purpose at all.

Paste your csv file somewhere? It's a text file as i said. Or paste it into a Trinket input.txt, and post the trinket here :) Then i can write a script to extract data from it in an instant.

And how you likely did guess, i should post these two trinkets here, yet again. Why in hell i had to convert the input to int, there was no need to, all works the same when it's converted to float.

Trinket for the input part.

https://trinket.io/python/489cd08ab6

Trinket for the output part.

https://trinket.io/python/7a74b9bc3f

« Last Edit: October 18, 2018, 06:07:44 PM by ayeaye »

ayeaye

  • Hero Member
  • *****
  • Posts: 866
Re: Bifilar pancake coil overunity experiment
« Reply #151 on: October 19, 2018, 07:42:59 AM »
It is easy to convert CSV, if that really is a problem. No need for spreadsheet, no need for excel. I did write a Python script as you may guess, but if you don't feel like doing it. Like the following web page, choose output options, then choose only one column, and you got it. This is only one of many such online converters. Not to talk about other tools, there are a great number that can do it, CSV is so common format. It's also so terribly simple. I don't understand finding ways to go around, instead of straightforward taking CSV out of oscilloscope and directly doing calculations. It is there and it's a great thing, so why not use it.

http://www.convertcsv.com/csv-to-csv.htm

Otherwise, what do you prefer to use for calculations? Like spreadsheets, they are not precise enough as i did show earlier, excel is also not really free. Only spreadsheet calculator i found was precise enough, but i cannot provide it online anywhere, may be too minimal for you as well.

How are you going to do calculations from the oscilloscope traces?

What else can i say. In Python something is float when it is assigned a float value, that is like 0.0 instead of 0. Any expression where at least one variable or value is float, is also float. while (True): is followed by colon, it just has to by syntax. Otherwise all the statements that follow, have tab before them (indent). It is a tab that is converted to two spaces when you copy code from there, any number of spaces work instead, but there always has to be the same number. The statements that are in such way indented, are in the loop, loop ends with a statement that is less indented, and that statement is already out of the loop. Going through the loop one time, is called iteration.

You don't need to know all Python, only a small subset of it, for almost all electronics calculations that you may need to do. This small subset is almost all that one needs to know to understand these small trinkets above. Is it really so absolutely impossible to learn.

The latest scripts for input and output parts posted here again, for eternity. These are exactly the same as in the trinkets above, just posted here in case if trinkets disappear in some mysterious ways, or such.

Thank you for all your attention.

Quote
#Time scale in us
XU = 0.02
#Voltage scales of ch1 and ch2 in V
YU1 = 5.0
YU2 = 0.2
#Resistor resistance in ohms
R = 47.0
#Frequency in Hz
F = 1000.0

#In calculations XU is in ns for unit
#Voltages are in mV, scale is 50 units
XU *= 1000 / 50
YU1 *= 1000 / 50
YU2 *= 1000 / 50
e = 0.0
f1 = open("input1.txt")
f2 = open("input2.txt")
while (True):
  s1 = f1.readline()
  s2 = f2.readline()
  if (len(s1) < 2 or len(s2) < 2): break
  vs = float(s1) * YU1
  vr = float(s2) * YU2
  vl = vs - vr
  pl = vl * vr / R / 1000
  e += pl
f1.close()
f2.close()
e *= XU / 1000000
print("Input power was %.4f uW" % (F * e))

Quote
#Time scale in us
XU = 0.02
#Voltage scale in V
YU = 0.2
#Resistor resistance in ohms
R = 47.0
#Frequency in Hz
F = 1000.0

#In calculations XU is in ns for unit
#Voltages are in mV, scale is 50 units
XU *= 1000 / 50
YU *= 1000 / 50
e = 0.0
f = open("input.txt")
while (True):
  s = f.readline()
  if (len(s) < 2): break
  vr = float(s) * YU
  pr = vr * vr / R / 1000
  e += pr
f.close()
e *= XU / 1000000
print("Output power was %.4f uW" % (F * e))


ayeaye

  • Hero Member
  • *****
  • Posts: 866
Re: Bifilar pancake coil overunity experiment
« Reply #152 on: October 19, 2018, 09:33:59 AM »
Now say you get CSV from the oscilloscope, where samples (points) are already in volts, the scripts for input and output parts would be like this.

Quote
#Time between samples in ns
XU = 0.4
#Voltages are in mV
YU1 = YU2 = 1000.0
#Resistor resistance in ohms
R = 47.0
#Frequency in Hz
F = 1000.0

e = 0.0
f1 = open("input1.txt")
f2 = open("input2.txt")
while (True):
  s1 = f1.readline()
  s2 = f2.readline()
  if (len(s1) < 2 or len(s2) < 2): break
  vs = float(s1) * YU1
  vr = float(s2) * YU2
  vl = vs - vr
  pl = vl * vr / R / 1000
  e += pl
f1.close()
f2.close()
e *= XU / 1000000
print("Input power was %.4f uW" % (F * e))

Quote
#Time between samples in ns
XU = 0.4
#Voltages are in mV
YU = 1000.0
#Resistor resistance in ohms
R = 47.0
#Frequency in Hz
F = 1000.0

e = 0.0
f = open("input.txt")
while (True):
  s = f.readline()
  if (len(s) < 2): break
  vr = float(s) * YU
  pr = vr * vr / R / 1000
  e += pr
f.close()
e *= XU / 1000000
print("Output power was %.4f uW" % (F * e))

Trinkets with these scripts. Input part.

https://trinket.io/python/d99dee2247

And output part.

https://trinket.io/python/fdd30eb52e

I don't know what about this 10x attenuation by probes, i assume in the samples of digital oscilloscope this is already considered.

« Last Edit: October 19, 2018, 02:18:23 PM by ayeaye »

F6FLT

  • Sr. Member
  • ****
  • Posts: 394
Re: Bifilar pancake coil overunity experiment
« Reply #153 on: October 19, 2018, 05:59:20 PM »
1)
When my bifilar coil is not connected to anything but is weakly coupled to the generator by a 2-turn loop, the resonance is around 2400 KHz.

This resonance is easily detectable, either by moving the scope probe towards the center of the coil (electric field), or by connecting the end of the probe to its ground wire and coupling this small loop to the coil (magnetic field).

This case also corresponds to the behavior of my monofilar coil.

2)
But when connecting the generator to one of the conductors at one end, and the ground to the other conductor at the other end (see diagram), the resonance drops dramatically, in my case to 28 Khz.

Surprisingly in this case of resonance, in the vicinity of the coil the probe detects the same weak electric or magnetic field as if there was no resonance.
Resonance is detected only by the increase in current supplied by the generator and by the significant voltage drop due to the overcurrent of the inrush current in the coil. We fall from 20v pp to a few hundred mV.

I confirmed this effect with a "bifilar" coil of a very different nature: an RG214 coaxial cable. The two conductors are the inner conductor and the shield, they are coupled by the 106 pF/m capacity of this type of cable. My coaxial loop is a single turn of 6m in circumference.
At the resonance frequency which is now 2550 Khz instead of 28 Khz with the other coil, the voltage drops from 20 V to 600 mV (sign that the loop has a much lower impedance than the 50 ohm of the generator), but the probe near the conductors does not detect a signal higher than what it is outside resonance.

In both cases, that of the flat coil or that of the coxial, as there is an overcurrent at resonance, at least the magnetic field should increase. But we don't detect an increase. I don't have a clear explanation for that. Does anyone have an idea?

ayeaye

  • Hero Member
  • *****
  • Posts: 866
Re: Bifilar pancake coil overunity experiment
« Reply #154 on: October 19, 2018, 06:22:55 PM »
In both cases, that of the flat coil or that of the coxial, as there is an overcurrent at resonance, at least the magnetic field should increase. But we don't detect an increase. I don't have a clear explanation for that. Does anyone have an idea?

Well, when the current increases, and the voltage on the coil decreases, then the coil consumes less energy. All current just goes through it, it is not consumed. It would be weird to expect the magnetic field to increase in that case, it would be weird if it doesn't *decrease*.

It would be interesting to see if you do input and output power calculations at resonance. Does the input power decrease at resonance? Ah yes then, but what means resonance when the input is pulse. Pulse length when the voltage on the coil is minimal?


gyulasun

  • Hero Member
  • *****
  • Posts: 4117
Re: Bifilar pancake coil overunity experiment
« Reply #155 on: October 19, 2018, 06:57:28 PM »
Hi F6FLT,

I think the 'poles' of the fields are so close to each other, both for the case of the bifilar at 28 kHz or for the one turn coaxial loop, that they (i.e. the poles) are able to close themselves magnetically so that little stray fields remain to the outside world. The case is similar to closed magnetic circuits: if you have two identically sized horse shoe magnets and you let their facing unlike poles join to each other (to form a kind of horse race track), then the strong fields near the earlier unclosed ends reduce to much weaker fields.
The electric field for the case of the coaxial cable loop should behave similarly: the shield is connected to the generator BNC body output i.e. to a very low voltage potential and the E field is mainly confined inside the coax dielectric. Very likely the magnetic field is also able to close itself mainly inside the coax.
The E field for the bifilar coil also remains inside its structure, I think.

If you feed a DC current into the bifilar coil, where do you think the static poles would appear?

Would like to ask whether you already had time to read my reply to you on the measurements of coils self capacitance? (reply #145, previous page)

I partially agree with ayeaye in that the power that remains to feed the coil or the coax at resonance is small but still I maintain what I have written above.
Ayeaye wrote this: "All current just goes through it, it is not consumed."
I would say: Any time current is able to flow, it creates losses (in the form of heat and radiation), so it is consumed to a certain degree (worst case is 100%).

Thanks,
Gyula

ayeaye

  • Hero Member
  • *****
  • Posts: 866
Re: Bifilar pancake coil overunity experiment
« Reply #156 on: October 19, 2018, 09:10:34 PM »
An ignorant thinking perhaps, but is it resonance, or is it frequency above which it just cannot do induction any more? Resonance i guess should be when the input power is the greatest, then i think also the output power is the greatest. Like the length of pulse when it's pulsed, this i think is the only thing that matters when it's pulsed.


F6FLT

  • Sr. Member
  • ****
  • Posts: 394
Re: Bifilar pancake coil overunity experiment
« Reply #157 on: October 19, 2018, 10:11:40 PM »
Hi F6FLT,

I think the 'poles' of the fields are so close to each other, both for the case of the bifilar at 28 kHz or for the one turn coaxial loop, that they (i.e. the poles) are able to close themselves magnetically so that little stray fields remain to the outside world. The case is similar to closed magnetic circuits: if you have two identically sized horse shoe magnets and you let their facing unlike poles join to each other (to form a kind of horse race track), then the strong fields near the earlier unclosed ends reduce to much weaker fields.
The electric field for the case of the coaxial cable loop should behave similarly: the shield is connected to the generator BNC body output i.e. to a very low voltage potential and the E field is mainly confined inside the coax dielectric. Very likely the magnetic field is also able to close itself mainly inside the coax.
The E field for the bifilar coil also remains inside its structure, I think.

Hi Guyla,

I understand what you're saying about the fields that are confined. It's surely one reason, in particular for the electric field. But I just found an additional explanation that I completely missed.
Remember that the input signal drops at resonance from 20v pp to 600 mv (and the current increases proportionally) and nevertheless I observe the same voltage from the open test probe feeling electric fields or with the "looped probe" feeling magnetic fields. From this I conclude that each field, the electric field and magnetic field that the probe detect, don't depend neither on the input voltage alone nor on the input current alone but depend on both. How is another question.

It's funny to see the strong resonance without the least effect on the amplitude but with a strong effect on the phase.
At resonance the signal probe is 90° out of phase with the input signal from the generator. When we change the frequency by only 2.5%, the signal probe becomes in phase or in opposition with the input signal (depending on the frequency lower or higher than the resonant frequency).

Quote
If you feed a DC current into the bifilar coil, where do you think the static poles would appear?
I don't understand the question in the context, no current is possible here in DC.

Quote
Would like to ask whether you already had time to read my reply to you on the measurements of coils self capacitance? (reply #145, previous page)

I partially agree with ayeaye in that the power that remains to feed the coil or the coax at resonance is small but still I maintain what I have written above.
Ayeaye wrote this: "All current just goes through it, it is not consumed."
I would say: Any time current is able to flow, it creates losses (in the form of heat and radiation), so it is consumed to a certain degree (worst case is 100%).

Thanks,
Gyula

I had read your reply on the measurements but didn't yet take time to study enough the method.

I remain troubled by the question of the probed magnetic field that doesn't increase at resonance while the current is increasing. The current goes the same direction in both wires and the magnetic field is proportional to the current only. So we should have an increase. May be it's a question of phase difference between the wires...
...to be continued

F6FLT

  • Sr. Member
  • ****
  • Posts: 394
Re: Bifilar pancake coil overunity experiment
« Reply #158 on: October 19, 2018, 10:31:00 PM »
Well, when the current increases, and the voltage on the coil decreases, then the coil consumes less energy. All current just goes through it, it is not consumed. It would be weird to expect the magnetic field to increase in that case, it would be weird if it doesn't *decrease*.

It would be interesting to see if you do input and output power calculations at resonance. Does the input power decrease at resonance? Ah yes then, but what means resonance when the input is pulse. Pulse length when the voltage on the coil is minimal?

The "looped" probe detect the magnetic field. V=dΦ/dt and the magnetic flux Φ crossing the probe loop area is proportional to the flux of the bifilar inductance L, i.e. to -Ldi/dt. If i is increasing, V should increase proportionally. It's only a question of current in the bifilar coil, not a question of voltage. The current generating the magnetic field of the bifilar coil is the sum of the current in each wire. These currents go in the same direction. I will try to measure them and to look at their phase relation.

ayeaye

  • Hero Member
  • *****
  • Posts: 866
Re: Bifilar pancake coil overunity experiment
« Reply #159 on: October 19, 2018, 10:44:30 PM »
At resonance the signal probe is 90° out of phase with the input signal from the generator.

And what does that 90° there mean, as much as i understand it means that then the back-emf is almost equal to the forward emf, means that the net power is zero, though induction and power is likely at its greatest. Now when there is overunity, it should go over 90°, which means that then back-emf generates more power than the forward emf provides. One video shows a bifilar pancake coil charging a battery that provides the power, this is what should happen at overunity. Unfortunately nothing bout the battery voltage is measured there, so not enough evidence for overunity.

How comes, it's when it's sine, power does two cycles during one voltage cycle. It is that power can be both positive and negative. Negative is the power generated by back-emf. So in that case it's seen that the negative power is equal to the positive power. Means the average power is zero.


gyulasun

  • Hero Member
  • *****
  • Posts: 4117
Re: Bifilar pancake coil overunity experiment
« Reply #160 on: October 19, 2018, 11:51:11 PM »
...
Remember that the input signal drops at resonance from 20v pp to 600 mv (and the current increases proportionally) and nevertheless I observe the same voltage from the open test probe feeling electric fields or with the "looped probe" feeling magnetic fields. From this I conclude that each field, the electric field and magnetic field that the probe detect, don't depend neither on the input voltage alone nor on the input current alone but depend on both. How is another question. 
Hi F6FLT,

Yes, the EM field strength also surely depends on the power fed into the coil or coax loop at  or near resonance. With a 50 Ohm generator output impedance only a few Ohm (or even under 1 Ohm) resonant impedance is connected in parallel, so a hugh voltage divison takes place via the internal impedance of the generator towards the coil, hence the small 600 mVpp amplitude remaining across the generator output i.e. across the coil input or across the coax loop input.
And a 600 mVpp voltage can maintain only a certain amount of current, certainly much less than the 20 Vpp would.

If you had a generator with much less than say 1 Ohm output impedance, then there would be much less internal voltage drop across the internal impedance at resonance, so much more voltage would be available across the coil hence the power feeding the coil would be much higher.  I am sure you know these, sorry. But then the EM field would surely be stronger than now from the 50 Ohm generator.
Quote
It's funny to see the strong resonance without the least effect on the amplitude but with a strong effect on the phase.
At resonance the signal probe is 90° out of phase with the input signal from the generator. When we change the frequency by only 2.5%, the signal probe becomes in phase or in opposition with the input signal (depending on the frequency lower or higher than the resonant frequency). 
Yes the phase changes drastically around the resonance, it should because either inductive or capacitive reactance starts dominating the moment you detune the generator from the resonant requency. The voltage amplitude can remain quasi unchanged near the resonance because the figure of merit, Q is relatively low due to the 50 Ohm generator impedance.  This 50 Ohm simply 'defines' the impedance the coil 'sees' at and in the vicinity of the resonance. See my further comment at the bottom too.


Quote
"If you feed a DC current into the bifilar coil, where do you think the static poles would appear?"

I don't understand the question in the context, no current is possible here in DC. 

I simply meant a case when there is only a static DC current fed into your bifilar coil, then where the magnetic poles develop?  I asked this to reveal the closeness of the magnetic poles to each other so they can combine easily.   But let's put this question for a later time if you wish.

Quote
I had read your reply on the measurements but didn't yet take time to study enough the method. 

Okay, thanks.

Quote
  I remain troubled by the question of the probed magnetic field that doesn't increase at resonance while the current is increasing. The current goes the same direction in both wires and the magnetic field is proportional to the current only. So we should have an increase. May be it's a question of phase difference between the wires...
I think it is also a question of the phase difference between coil input current and input voltage, when they are in phase the real power should be at a maximum but there is the generator impedance which prevents high current both at resonance or at off resonance (as long as at off resonance the capacitive or inductive reactance will not be higher than 50 Ohm, beyond that range already the reactances would infuence hence define coil current).

Gyula

ayeaye

  • Hero Member
  • *****
  • Posts: 866
Re: Bifilar pancake coil overunity experiment
« Reply #161 on: October 20, 2018, 12:08:09 AM »
There is no need for input and output part, and only one script is necessary.

Because when the power is positive, it's input, and when the power is negative, it's output. So the script becomes the following.

Quote
#Time between samples in ns
XU = 0.4
#Voltages are in mV
YU1 = YU2 = 1000.0
#Resistor resistance in ohms
R = 47.0
#Frequency in Hz
F = 1000.0

ei = eo = 0.0
f1 = open("input1.txt")
f2 = open("input2.txt")
while (True):
  s1 = f1.readline()
  s2 = f2.readline()
  if (len(s1) < 2 or len(s2) < 2): break
  vs = float(s1) * YU1
  vr = float(s2) * YU2
  vl = vs - vr
  pl = vl * vr / R / 1000
  if (pl >= 0): ei += pl
  if (pl < 0): eo += pl
f1.close()
f2.close()
ei *= XU / 1000000
print("Input power was %.4f uW" % (F * ei))
eo *= -1 * XU / 1000000
print("Output power was %.4f uW" % (F * eo))

The following is the trinket with that script. This script needs a full cycle of channel 1 in input1.txt and channel 2 in input2.txt, or maybe less if after some time both are 0. You can see that the results are exactly the same as when doing it separately for input part and output part.

https://trinket.io/python/a90a61d84e

This script can be used to calculate input and output power of the coil also when the signal is sine, or when the signal is whatever, it's universal. No need to separate any parts.

Do you see :)

I took the figure below from wikimadia commons, By 121a0012 - self-made, after w:Image:ACPower03CJC.png by User:C J Cowie, Public Domain, https://commons.wikimedia.org/w/index.php?curid=2866717 .

« Last Edit: October 20, 2018, 10:02:59 AM by ayeaye »

ayeaye

  • Hero Member
  • *****
  • Posts: 866
Re: Bifilar pancake coil overunity experiment
« Reply #162 on: October 20, 2018, 12:52:13 PM »
The following is the Inkscape script trinket with the channel 2 data above. Assuming that the output of the oscilloscope is voltages, so that the output of that script can be directly used for the input of the script for digital oscilloscope. That is, it supposed to provide the same sample list as in the csv file taken from the digital oscilloscope. So these who have analog oscilloscopes can do the calculations the same as if they had digital oscilloscope. All i know the people here use, are digital desktop oscilloscopes and old analog oscilloscopes, i have not seen anyone here using anything else.

https://trinket.io/python/a8d33f8989

This is what i offer, two Python scripts to calculate input and output power of a coil.

Seeing phi is a way to estimate the power efficiency when the signal is sine. With other waveforms, with digital oscilloscope, one can measure voltage on the coil and voltage on the resistor, then multiply these channels and show them as the math channel. This enables to quickly visually estimate the power efficiency, comparing the positive and negative areas of the waveform on the math channel.

It were easy to make power calculations if the oscilloscope did calculate positive and negative areas of the math channel. Unfortunately cheaper scopes cannot do calculations on the math channel at all, and cannot even output the math channel as csv. The accuracy of any calculations done by the scope may also be questionable on some scopes. So in that case there is no other way than to get the waveform data of both channels as csv, and do the calculations from that.

The more expensive scopes most likely provide more, and some may be able to accurately calculate positive and negative areas of the math channel. For that it were enough if they were like capable of calculating the area of the math channel waveform in a range. But that written here, should be so that everyone can do that, so that the results can be replicated by other people, and i think most here use rather cheap digital scopes with not many capabilities. Some use even analog scopes, like me, so i provided a method for them as well.


F6FLT

  • Sr. Member
  • ****
  • Posts: 394
Re: Bifilar pancake coil overunity experiment
« Reply #163 on: October 21, 2018, 10:00:18 PM »
The resonance frequency of my bifilar coil when working as a normal monofilar is 2330 KHz. Its capacity is 55 nF thus its inductance is 84.8 nH ( L = 1/(C*ω²)  ).

But with the bifilar coil connected as specified in reply #153, the resonance drops to 28Khz.
I have tried many LTspice simulations to get an equivalent circuit. The only LTspice simulation that fits what I observe needs the inductance (not the capacitance) to be multiplied by 5000.

It's indeed impossible to reach a so low resonance with values as small as 55nf and 84.8nH. The schematic is very simple but problematic. Did I miss something? (Otherwise imho it's useless to play with bifilar coils until we obtain an explanation for this elementary result).

ayeaye

  • Hero Member
  • *****
  • Posts: 866
Re: Bifilar pancake coil overunity experiment
« Reply #164 on: October 21, 2018, 11:20:31 PM »
It's indeed impossible to reach a so low resonance with values as small as 55nf and 84.8nH. The schematic is very simple but problematic. Did I miss something? (Otherwise imho it's useless to play with bifilar coils until we obtain an explanation for this elementary result).

My guess is that because of the capacitance, some of the induction happens Lenz free, and this enables very high inductance. This conjecture is just intuitive, not really thought through at all. But the matter is that there seems to be by now no better explanation.

Please measure the powers at resonance. Either the voltage on the coil and on the resistor, and phi between them, multiplication of them, or such. Not that there should be overunity, but this is an anomalous phenomenon, so worth to study. It is not excluded that in certain conditions there may be overunity, with a different resistance of the resistor, or whatever else.

Yes we don't know, and it's worth to research something, when we don't know.