OverUnity Research

I still am following member "Jagau" on "AboveUnity.com" forum doing a replication of the "Melnichenko Effect".

Jagau his thread can be found here:  https://www.aboveunity.com/thread/melnichenko-s-effect/
There is now also another replicator, Brian trying the replicate the effect here: https://www.aboveunity.com/thread/brian-s-melnichenko-s-effect-replication/

This thread of mine will try to follow the replicators in what they do and trying to replicate the effect myself.

Melnichenko his website can be found here: https://vk.com/id285085326 (use Chrome to auto translate) and his youtube channel here:
https://www.youtube.com/channel/UCEtqI2EhN32Mvq7Wp5G9Vpg/videos

This youtube channel shows many different circuits, but presently Jagau is referring to this video for his replication: https://youtu.be/f6baJ0tqBew

The "effect" suppose to be, as i understand it, that due to the coils setup (big primary air coil and smaller inside secondary ferrite coil) there is no influence from this secondary coil on the primary coil and thus the energy for the load of L2 is free.

Perhaps some Russian speaking member is able to confirm this effect as mentioned in the video by Melnichenko.


Jagau has started with a basic modified circuit without L2 and its load as can be seen here:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44860;image)

My present replication of this basic circuit looks like this:


(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44864;image)


With this as diagram:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44868;image)


Some data on the used circuit:

L1 primary air core 80 turns: 726uH
L2 secondary 2x 144 turns ferrite cores separated by 1.5mm gap: 4mH    (this L2 is not yet in use in the present basic circuit!!).

When inserting L2 into L1 in the next step, the L1 inductance raises to 850uH.
When shorting this inserted L2, the L1 induction drops to 710uH.
But again, this L2 is now not in use!

FG running 1.4KHz @ 9% duty cycle (adjust to get ~120V across Lamp1)
Lamp1 120V / 4W

Initial power measurements shows for:

input power:   4.57W, see screenshot 1    (yellow: voltage, blue: current through csr and red calculated power)
IGBT power:   4.63W, see screenshot 2    (yellow: voltage, green: current and red calculated power)
Lamp1 power: 4.66W, see screenshot 3    (yellow / blue voltage (differential) , green: current through lamp1 and red calculated power)

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44870;image)

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44872;image)

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44874;image)

As can be seen in the screenshots, the measured signals get more and more erratic and accurate power measurements gets harder to make.
I had to use the differential probing technique and advanced math function to measure the Lamp1 power due to its floating state.

It also shows that in this basic setup, almost all input power is being consumed by the Lamp1

Next step i guess would be to insert L2 into L1 and attach a load to L2 to see if this load (lamp2) can be lit for free (so without any influence on the input power and Lamp1 power).

Side note 1: AboveUnity.com is often offline for some periods, so you might get some time outs.
Side note 2: the input DDM's (and PS meters) show lightly less current (thus power when calculated) as compared to the scope probaly due to the current AC component in the signal.


Regards Itsu

Just some comments, this replication has started some months (march) ago already see here: https://www.overunityresearch.com/index.php?topic=3691.msg98251#msg98251        so it is not something i recently started with.

I opened this thread now to get it into the open as to overcome any doubts it was done behind closed doors.


Also the use of rms value's on the scope screenshots does not mean the scope or i am using them for power calculations.
Its just an indication of signals seen.

The math function of the scope for calculating the power is using many instantaneous samples of the involved signals to calculate power and average (mean) them.

Itsu

I added my secondary coil L2 inside the primary coil L1.

Connected to this L2, are 2 diodes, a switch and another 120V/4W lamp (lamp2) as load, see diagram and picture:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44881;image)


(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44887;image)


Data on this L2 coil and the influence it has on L1 as mentioned earlier:

L2 secondary 2x 144 turns ferrite cores separated by 1.5mm gap: 4mH   
L1 primary air core 80 turns: 726uH

When inserting L2 into L1, the L1 inductance raises from 726 to 850uH.
When shorting this inserted L2, the L1 induction drops to 710uH.

So the ferrite from L2 influences the inductance of L1 (increase).
Also the lowering of the L1 inductance when shorting L2 shows there is some interaction (mutual inductance) between both coils.

This influence is also seen with power on (L2 switch still OFF), as without changing any other parameters (input voltage, FG settings etc.) we see a decrease of input power measured across the IGBT (earlier screenshot 2) from 4.6W to 4.2W, and the voltage across the lamp1 / cap dropped from 123V to 110V.
This is probably due to the increased induction of L1.

Closing the switch in L2 shows a very dim Lamp2 being on, a further decrease of the input power across the IGBT (~0.1W) and a lower voltage across Lamp1 (now 107V).

Measuring the voltage / current across / through lamp2 shows it pulls about 127mW, see screenshot.

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44885;image)

So it seems that the power into Lamp2 load comes from the less power into Lamp1 (it shares the available power).

I will make some accurate measurements later.


Itsu

Itsu,

The L1 to L2 coupling in your above post is K=.406 using k=(1-Lpss/Lp)^1/2  .

Regards,
Pm

Hi PM,

thanks, so this can be considered as a "loose coupling" i guess.

Itsu

Doing some more accurate measurements on this new setup which includes L2 and its load see above post #2

Due to the increased inductance of L1 because of adding L2 inside, the voltage across lamp1 (120V/4W) has dropped from 123V to 103V.
So i raised the voltage across lamp1 back to 122V by increasing the FG duty cycle from 9% to 10%.

So with L2 in the circuit, but NOT loaded (L2 switch off), we have a similar situation as we had before without L2.

Data:

input voltage 36V
FG set to 1.4KHz @ 10% duty cycle square wave.
L1 850uH
L2 4mH
Voltage across Lamp1 121.8V (lamp2 off) 
"   "          "    Lamp1 119.4V (lamp2 on)
Lamp2 very very dimly on.


Power measurements WITHOUT L2 load (switch off):
Input power:   4.71W  see screenshot 1
IGBT power:   4.75W  see screenshot 2   
Lamp1 power: 3.93W  see screenshot 3
Lamp2 power: 0W

Power measurements WITH L2 load (switch on):
Input power:   4.71W  see again screenshot 1
IGBT power:    4.75W  see again screenshot 2   
Lamp1 power: 3.63W  see screenshot 4
Lamp2 power: 0.16W  see screenshot 5


So we see that the input power does not change when L2 load (lamp2) is activated.
But somehow even without L2 load (lamp2) activated we have a lower power into Lamp1 compared with No L2 inserted into L1, see post #1 above (extra losses into L1/L2/ferrite)?
When L2 load (lamp2) is activated, then power into lamp1 further drops (0.3W) but a part goes now to L2 load (lamp2) which consumes 0.16W.

So this present setup with L2 does not show any special effect, but there are several tuning options which can be explored.

Itsu

Experimenting with L2 right now.

I have used several:

# 60 turns ferrite core: 372uH, 
# 2x 144 turns in series ferrite cores with 1.5mm gap: 4mH
# a spool of thin dual magnet wire bifilar style, many turns with a bunch of copper coated welding rods as core: 376mH

But all show a barely lighted lamp2 (120V/4W) while keeping lamp1 (also 120V/4W) running on 120V.
It shows that Lamp2 is pulling 65 to 160mW of power with these L2's

A 12V AC 3W led lamp as lamp2 does show some decent light, but we all know how deceiving using leds can be as it only consumes 279mW.

Continuing the search.......

For Chris:   https://youtu.be/iKftVD0fAts?t=312


Itsu

Zooming in on the use of the 12V AC 3W led lamp as load for L2 (using the 2x144 turn L2 now).

This picture shows the lights of both lamp1 (120V/4W incandescent, left) and lamp2 (12V AC 3W, led):

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44993;image)


The screenshot shows the voltage across the led in yellow , the current through the led in green and the calculated power in red:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=44995;image)



The difference between the earlier reported 279mW and the present 381mW is the different FG setting.
Its now set at 1.2KHz @ 10% duty cycle to match 120V across Lamp1.

The earlier test was done with my initial FG setting of 1.4KHz @ 9% duty cycle to be equal across different lamps.

Regards Itsu

Quote from: Itsu on 2022.07.11, 08:23:24

...
A 12V AC 3W led lamp as lamp2 does show some decent light, but we all know how deceiving using leds can be as it only consumes 279mW.


For Chris:   https://youtu.be/iKftVD0fAts?t=312

...

Hi Itsu,

Thank you for sharing your test results with the Melnichenko setup under investigation elsewhere too.

Regarding your latest LED lamp test, it is very good you checked its real power consumption (279 mW and now 381 mW) because our eyes' light perception is responsive for the peak values hence we can easily be tricked by our eyes if we are to eyeball estimate the actual power of say a 3W LED bulb which gives a decent light or it is well lit as seen in your picture above.

I think you already did measurements on a LED lamp brightness versus input power at Stefan's forum some years ago.

I already also referred to a video at that forum in which an off the shelf 12 V AC/DC LED lamp teardown was shown, including detailed input current measurements in the function of DC input voltage, ranging from 6V to 30V.

What is interesting is that the 5W spec is actually 4.5W at and around 12V input and at 6V input and below there is no input current (confront this with an incandescent lamp).

Here is the video link https://www.youtube.com/watch?v=i-Roc5TLdnw (https://www.youtube.com/watch?v=i-Roc5TLdnw) and hopefully this helps experimenters not to eyeball judge the power consumption of LED bulbs by their brightness. I attached a snapshot from the end of the video on the results of input power measurements in the function of input voltage and current.

Gyula

Sorry to interrupt
I was wondering if the Temperature of bulb could also
Be used for reference ?.. member Ernie -(ION) had always championed
Temperature as good reference for experimenters to keep track of progress,
Yes incandescent bulbs were recommended
However it would be interesting if temp is linear with input on LED in question?

Please remove this post if inappropriate or ( for any reason)

Respectfully
Chet K

Hi Gyula,

yes, i have characterize incandescent lamps and leds before using a black box which is painstaking procedure and only valid for the lamp / led under test.
But it reveals much.

Thanks for the link to a similar test.





Chet,

i think it would be a valid procedure to measure the heat output of a lamp, i never did that, but wonder if this also works accurate on these low wattage (4 / 3W) lamps / leds.

I might give it a try though.






For Chris,

the beauty of power measurements with a scope is that the scope does not care what kind of signal it has, DC, AC, peak, square, sinuous etc.
It just takes millions  of samples of the signals (not rms, not mean, not ...) from its buffer and multiply (in this case) those instantaneously getting millions instantaneous power values which then get averaged (mean) presenting the average power across that buffer.

So nobody is using rms value's to calculate power, those rms value's you see on the voltage and current are JUST a representation of the signals measured by the scope, it does NOT use them for calculating power.

I don't pretend to be a guru in anything, so please anyone with extended knowledge on scope power measurement techniques step forward and either confirm, deny and/or improve on my above statement.


Concerning the dimensions of L2, i did use several different L2's, see above post #7, but all shows similar output value's (65 to 160mW), so i hope Jagau can shed some light on what dimensions he uses to get cop > 1.

Regards Itsu
   
Itsu

For Chris,


i made the same input measurement as in my post #1   screenshot 1    which is the input voltage (DC), current and calculated power.
Now instead of rms value's to show voltage and current i used Peak2Peak, but i could use any other indication like mean or apples if it would exist.

You see the calculated power in red (ch1 * CH2)  still shows the actual ~4W.

So the indication of the signals shown does NOT get used by the scope internal math function.

SonOfLuck is right about that, he just wants to help to get it straight.

Itsu

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45003;image)

Dear Chris,

I just did a similar screenshot of an earlier one, NOT an exact accurate measurement of the same situation as i have since the 1st screenshot changed several things like frequency, duty cycle, L2 etc.

It was just to show you that the indication of the signals (rms, mean whatever) is NOT being used by the math function which is what you asked:

QuoteIf Itsu can show in Scope Math Functions where he is setting Mean or Average Values for each Channel, I will accept his answer,

So here you have it, signals in mean, still the ~4W input calculated:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45005;image)


I am done with this now, continue to experiment.........


Itsu

I modified my black box to measure the dim 120V / 4W lamp1 using a little solar panel (put the solar panel closer) and calibrated this 120V / 4W lamp using a known variable DC voltage source.

The lamp1 when lit puts a certain amount of light on the solar panel which gets measured by a fluke DMM in mV's depending on the input on lamp1

The below Graph and data shows this relationship.

Instead of having to do elaborate measurements (differential probing, advanced math function on the scope) because this lamp1 is floating, i now can see on the DMM and the amount of mV's presented what the input wattage must be using the graph.

Also this input wattage is independent of the shape of the voltage signal of the lamp1.

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45011;image)

Hi Itsu,

This is a good idea and in this low 1-2 kHz frequency range the filament of the lamp may have a low enough self inductance value not to influence the DC resistance hence modify the loading impedance.

Thanks for your kind efforts on this setup.

Gyula

Thanks, the filament of these 120V / 4W night lights are straight, so not coiled which will help to keep the inductance down too.

Itsu

New L2 as advised by Jagau (1.41 x L1), so 113 turns as L1 is 80 turns see picture.

L2 measures 1247uH when inside L1 and 1258uH standalone.
L1 measures  951uH when L2 inside and  726uH standalone.
L1 measures  673uH when L2 shorted. 
Coupling factor K between L1 and L2 now 0.54     using k=(1-Lpss/Lp)^1/2   or   https://www.e-magnetica.pl/calculator/magnetic_coupling_coefficient

Using the 12V AC 3W led as lamp2 load.
Using the 120V /4W incandescent lamp in the blackbox.

Input voltage still 36V, input power 4.6W (L2 load on or off) see screenshot 1

FG 1.3KHz @ 10% duty cycle
120V across lamp1 (solar gives 1055mV which equates to 3.6W see graph) when L2 load switch OFF.
When L2 load switch ON, the voltage across Lamp1 goes down to 97V (solar gives 500mV which equates to 2.6W see graph)

Lamp2  (12V AC 3W led) when ON consumes 620mW see screenshot 2


So with the new coil (1.41 x turns of L1) we again have a steady input power with or without Lamp2 of 4.6W.
Without Lamp2 ON, the lamp1 consumes 3.6W.
With lamp2 on it consumes 620mW, but lamp1 drops then to 2.6W giving a total consumption of both lamps of 3.22W.

Overall efficiency with lamp2 OFF 3.6  / 4.6 = 78%
Overall efficiency with lamp2 ON  3.22 / 4.6 = 70%

I will do the same test using now another 120V / 4 incandescent lamp as lamp2.

Itsu

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45023;image)

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45017;image)

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45015;image)

Now using an incandescent lamp 120V / 4W as lamp2 (barely visible when on)

New L2 see above plus additional data.

Input voltage still 36V, input power 4.6W (L2 load on or off)

FG 1.3KHz @ 10% duty cycle
120V across lamp1 (solar gives 1055mV which equates to 3.6W see graph) when L2 load switch OFF.
When L2 load switch ON, the voltage across Lamp1 goes down to 118V (solar gives 1007mV which equates to 3.53W see graph) 

Lamp2 (120V / 4W incandescent lamp) when on consumes 76mW see screenshot 1     (see screenshot 2 for a zoomed in shot).

So with the new coil (1.41 x turns of L1) we again have a steady input power with or without Lamp2 of 4.6W.
Without Lamp2 ON, the lamp1 consumes 3.6W.
With lamp2 on it consumes 76mW, but lamp1 drops then to 3.53W giving a total consumption of both lamps of 3.606W.

Overall efficiency with lamp2 OFF 3.6  / 4.6 = 78%
Overall efficiency with lamp2 ON  3.606 / 4.6 = 78.3%

I agree that the resolution of the graph in this 1000mV area is very rough, but it still gives a fair idea of what the lamp1 is doing in terms of its input wattage.
It certainly will show any special effects when available.

Itsu

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45025;image)

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45027;image)

As there is a debate going on on AboveUnity.com concerning the use of rms versus average (mean) value's and/or the use of DDM's (true rms or not) for calculating power, the focus seems shifted off from the "Melnichenko effect" replication.

I always try to avoid using rms and/or average value's or the use of DDM's for calculating power and use the scope to do that for me using its math function to calculate many thousands instantaneous voltage and current samples (V x I) to arrive at the instantaneous power value's for the scope screen or buffer which then get averaged across that screen or buffer to get the average power.

Its the only reliable thing to do IMHO as the scope does not care if the presented signals are DC, AC, pulsed, spiky or whatever nor if its shown as rms, mean, p2p or whatever.


So while waiting for the members there to focus again on the "melnichenko effect" replication, i will try to explore the original melnichenko circuit (so not the present circuit which was slightly modified by Jagau) to see if any special effects can be noticed when using that one.

Up till now the modified melnichenko circuit did not show any special effect and the efficiency was shown to be in the max 78% range (cop = 0.78) despite using several different combinations of L1 and L2 coils, loads (lamp1 and lamp2) and measurement / calculation methods, something that other replicators also have mentioned.


Below the slightly modified circuit used up till now:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45051;image)





Here the original Melnichenko circuit (one of many) which was the base of the present thread on AboveUnity.com which i will explore for now:
(differences are mainly the positions of C2, D1 and Lamp.  The parts on the right of L1 (L2 circuit) was the same in both)

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45049;image)





Regards Itsu

The problem with this original melnichenko circuit is that now the BEMF pulse is not suppressed by the big C2 cap and even at only 5V input voltage we have a BEMF pulse of 177V.

No lamps are on with this input voltage so we need to increase this input voltage to make any sensible measurements, but therefor we need to suppress this BEMF pulse, like using a TVS diode (Transient-voltage-suppression diode) across the IGBT collector - emitter.

Itsu

By the way, i have made a LTspice simulation of the modified Melnichenko circuit used up till now.

It uses a MOSFET instead of the IGBT (not in the lib), and resistors for the lamp loads (3.3K for the 120V/4W lamp1, 28Ohm for 12V / 5W lamp2)

It takes a long time (minutes) for C2 cap to fill up in real time, so also in the sim, so i take a long for the calculations (10 min) and/or "start saving data" after 9min 50s or so to evaluate the traces.

But this makes the sim to run forever (hours) before any useful data is presented, so overall not very useful i think.

Below screenshot is from the sim stopped after 1 Hour running with only 0.03% (2s) finished showing the last 2 seconds or so of the run with the voltage across L1 in green, current through L1 in blue and the voltage across C2 / Lamp in red.

The voltage and current traces (green and blue) show the same shape as recorded with scope on the real circuit.
The red trace voltage across C2 / lamp should increase to 120V but is now only 50V but rising.

.asc file included if someone wants to give it a try.

EDIT:  it could the L2 circuit on the right needs to be grounded too as the simulation might crash. This is not something i want, but probably needed.



Itsu

Quote from: Itsu on 2022.07.21, 12:08:20
The problem with this original melnichenko circuit is that now the BEMF pulse is not suppressed by the big C2 cap and even at only 5V input voltage we have a BEMF pulse of 177V.

No lamps are on with this input voltage so we need to increase this input voltage to make any sensible measurements, but therefor we need to suppress this BEMF pulse, like using a TVS diode (Transient-voltage-suppression diode) across the IGBT collector - emitter.

Itsu

Itsu,

What appears to be happening when the BEMF pulse has such a large amplitude is that the secondary leads need to be reversed or D2 and D3 need to be reversed.  With a k=.4 and properly polarized secondary, the secondary pulse should be ~2.5 times the primary peak voltage or 1/.4 .

Regards,
Pm

Thanks PM,

but when i do that (swapping L2 leads), see screenshot, then the pulse gets worse and besides that, the L2 pulse (blue) is now induced in the magnetization phase of L1, while we want to utilize the demagnetization part (BEMF) of the L1 coil to induce power into the L2 coil.

Kees

Quote from: Itsu on 2022.07.21, 09:30:54

...
I always try to avoid using rms and/or average value's or the use of DDM's for calculating power and use the scope to do that for me using its math function to calculate many thousands instantaneous voltage and current samples (V x I) to arrive at the instantaneous power value's for the scope screen or buffer which then get averaged across that screen or buffer to get the average power.

Its the only reliable thing to do IMHO as the scope does not care if the presented signals are DC, AC, pulsed, spiky or whatever nor if its shown as rms, mean, p2p or whatever.

...

Hi Itsu,

This approach is the best to use but it needs to have a correct, two channel digital oscilloscope. 
There is another approach for those who have analog or other scope types with no Math function.

If a home experimenter, using his scope can measure the peak current his coil draws at the switch off moment and knows the inductance of the coil, then he can calculate the instantaneous energy (and power) the coil just possesses.

Here is the well known formula to calculate how much energy a coil receives from an input current:
   E = 1/2 * L * I² 

Let's take as an example your 3 Amper peak current measured for your old L1 coil shown in your Reply #1 https://www.overunityresearch.com/index.php?topic=4312.msg99509#msg99509 (https://www.overunityresearch.com/index.php?topic=4312.msg99509#msg99509)

Your L1 coil inductance was 726 uH (without the L2+core), switching frequency was 1.4 kHz with about 9 % duty cycle as you wrote.

Using this online calculator  https://physicscalc.com/physics/inductor-energy-calculator/ (https://physicscalc.com/physics/inductor-energy-calculator/)  for the stored energy in coils, we get E = 0.003267 Joule.

Your ON time may have been 64.2 us (based on the 9 % duty cycle, and one full cycle time was T = 1 / f = 714.2 us).  So the L1 coil received 3 A current pulses at each 714.2 usec.  Multiplying the stored energy,  E = 0.003267 J by the 1400 Hz input frequency (or dividing it by T=714.2 us), we get 4.57 Watt, this is very close to your scope's Math calculated 4.63 W result for the input power by multiplying the instantaneous DC input supply voltage and the input current (2nd scope shot in your Reply #1 above). 

So the available magnetic energy (at the switch-off moment of the input current) in your L1 coil was 0,003267 Joule. This amount of energy calculated by the formula does not include the switching and the coil (and core) losses, these should be separately considered to get the full input energy (and power) taken from a power supply. These losses are relatively low or can be kept low. 

Gyula 

Quote from: gyula on 2022.07.22, 09:55:41

Hi Itsu,

This approach is the best to use but it needs to have a correct, two channel digital oscilloscope. 
There is another approach for those who have analog or other scope types with no Math function.

If a home experimenter, using his scope can measure the peak current his coil draws at the switch off moment and knows the inductance of the coil, then he can calculate the instantaneous energy (and power) the coil just possesses.

Here is the well known formula to calculate how much energy a coil receives from an input current:
   E = 1/2 * L * I² 

Let's take as an example your 3 Amper peak current measured for your old L1 coil shown in your Reply #1 https://www.overunityresearch.com/index.php?topic=4312.msg99509#msg99509 (https://www.overunityresearch.com/index.php?topic=4312.msg99509#msg99509)

Your L1 coil inductance was 726 uH (without the L2+core), switching frequency was 1.4 kHz with about 9 % duty cycle as you wrote.

Using this online calculator  https://physicscalc.com/physics/inductor-energy-calculator/ (https://physicscalc.com/physics/inductor-energy-calculator/)  for the stored energy in coils, we get E = 0.003267 Joule.

Your ON time may have been 64.2 us (based on the 9 % duty cycle, and one full cycle time was T = 1 / f = 714.2 us).  So the L1 coil received 3 A current pulses at each 714.2 usec.  Multiplying the stored energy,  E = 0.003267 J by the 1400 Hz input frequency (or dividing it by T=714.2 us), we get 4.57 Watt, this is very close to your scope's Math calculated 4.63 W result for the input power by multiplying the instantaneous DC input supply voltage and the input current (2nd scope shot in your Reply #1 above). 

So the available magnetic energy (at the switch-off moment of the input current) in your L1 coil was 0,003267 Joule. This amount of energy calculated by the formula does not include the switching and the coil (and core) losses, these should be separately considered to get the full input energy (and power) taken from a power supply. These losses are relatively low or can be kept low. 

Gyula

Gyula,

thanks so much, so like we can calculate the energy in a capacitor (different formula of course), we can with your formula (E = 1/2 * L * I²) also calculate the amount of energy in a coil.



So we can add a 4th method to measure / calculate the input power:


1)  using a voltage and current probe at the PS entry of the circuit.
    This shows i have 36V DC @ 129.7mA rms AC for which the scope math function calculated the mean power to be 4.42W

2)  using a method presented by Jagau "Urms_pulse=Vp√D" where for:
    voltage my t1 is 72us and T is 770us thus D = 0.093, Vp is average 32.9V, so giving Vrms to be 10.06V rms
    current my t1 is 72us and T is 770us thus D = 0.093, Ip is 2.8A, but triangled, so halved for square so is 1.4A, so giving I rms to be 427mA rms   
    Power then is V rms x I rms = 10.06 x 427mA = 4.29W

3)  A 2th way to calculate I rms in the above 2) method is to use the triangle formula "Urms_triangle_ = Vp√t1/3T" which gives 2.8A x  √72/3*770 = 2.8A x 0.1765 = 494mA rms
    This multiplied by Vrms 10.06V gives 4.97W

4) Gyula's presented formula E = 1/2 * L * I² which for the data presented in method 2) gives 1/2 * 0.000951 * 2.8² =  0.00372792 Joule  (using 951uH for L1 and running at 1.3Khz 10% duty cycle).
    Divided by the 770us period, it shows 0.00,372792 / 770 = 4.84W.

So we got input powers: 4.42W, 4.29W, 4.97W and 4.84W by different measurement / calculations.

Each method has its own parameters, and although i tried to measure them as best as i can, none will be exactly right due to several fluctuations thus these outcomes.

But overall, if i would have to pick the most correct one i will vote for method 1.

Regards Itsu

From my original Melnichenko circuit which shows a nasty BEMF spike i made a LTspice simulation too, see screenshot.

It also shows a nasty spike of 900V with 5V input green trace (voltage across L1).
Blue trace is the current through L1.

.ASC file attached.


Itsu

My IGBT being a 1200V one and using my high voltage probe i turned the PS up so i had a 1127V BEMF pulse on L1 which was reached at 18V input.
The lamp1 (120V / 4W) was not measurable on and the lamp2 (12V / 5W) was very dimly on (7.6V p2p).

The yellow 7.6V pulse across lamp2 STARTS at the beginning of the white BEMF pulse, so not at the shown position!!

So the fact that Lamp1 was not to be measured, this circuit is almost impossible to analyze.

Not sure how to tackle this.

Itsu

Hi Itsu,

During these latest tests, what is the input power to the circuit?  Is it also around 4.6 - 4.7 W?  Peak input current?  Just to learn more about the details in this situation.
IF Melnichenko used such circuit indeed (not only in his drawings shown in his videos), then he should have faced the same issue...  :D 

Gyula

Gyula,

no, with a 1150V BEMF pulse from L1, at 18.5V input, we have about 52mA input running making a calculated mean power b y the scope of 950mW .

Itsu

Okay, thanks.  This means that the AC impedance of the L1 coil is higher than the DC resistance of the filament, using the same ON time for the IGBT switch as earlier. 

You wrote: "So the fact that Lamp1 was not to be measured, this circuit is almost impossible to analyze."

Well, just use a noninductive resistor load instead of the light bulb and see the power in the resistor then. 


Gyula

Hmmm, the inductive reactance of my 951uH coil @ 1.5KHz calculates to be around 9 Ohm plus 0.3 Ohm dc resistance.
This is way lower then the 120V/4W lamp which has a cold filament resistance of 385 Ohm and 3.3K when on full.

The 4.6 - 4.7 W measured earlier was with the modified melnichenko circuit, not with the present original melnichenko circuit.

Anyway, i can substitute an induction free resistor  for this 120V/4W bulb and measure the power with and without lamp2 switched on.

Itsu

Okay Itsu,  thanks for the further pieces of info and the correction, I agree now.  8) 

Gyula

Using a 100 Ohm 1% induction free resistor as lamp1 and a 12V / 5W lamp as lamp2, i have a 200V BEMF spike across L1 @ 36V input voltage.

FG:1.5KHz @ 10% duty cycle


The input power in that situation is:

with lamp2 load off: 3.65W
with lamp2 load on:  3.73W


The power into the 100 Ohm lamp1 load is:

with lamp2 load off: 3.2W
with lamp2 load on:  2.3W


The power into the lamp2 load is:

with lamp2 load off:  0mW
with lamp2 load on: 710mW


So in the original melnichenko circuit using a 100 Ohm 1% induction free resistor as lamp1 we see a similar behavior as with the by Jagau modified melnicheno circuit.
At 36V input voltage, the lamp2 load when switched on takes / shares its power from Lamp1.

Efficiency with:

Lamp2 load off: 3.2W / 3.65W                = 0.87 = 87%
Lamp2 load on: 2.3W + 710mW / 3.73W = 0.80 = 80%

As both the original and the modified Melnichenk circuits show similar results (cop < 1) using similar components / setup, one thing that could be the cause of these ordinary results is the components / setup used for the coils.

It could be that the secret to the special Melnichenko effect is hidden there.

I did use several different L2 coil setup's, but none showed anything special.


Taking a look at the patent (attached below), and especially the items 0009 and 0010 where the coils are described does not enlighten me much:

[0009]
The simplest version of the device is one magnetization coil (inductor) and one ferromagnetic core with a removable winding on it.
The inductor coil can surround the core coaxially, but due to a specially increased diameter, it does not cover a significant or large part of the magnetic field of the core.
For example, the diameter of the coil is almost equal to the length of the core plus its width if the coil is considered round and the core is approximately round or square.
The coil can be both square and rectangular.
The dimensions of the core and the shape of the core, as well as their number, can be any.
But the dimensions of the coil are chosen so that a significant part of the magnetic energy of the ferromagnet is closed inside the coil without the coverage of the turns by inductive coupling.
This is achieved by selecting the distances from the wires of the inductor coil to the surface and sides of the core.

[0010]
The coaxial arrangement of the core in the coil is the simplest technical option.
For generation, it is enough to divide approximately two-thirds of the energy of the magnetic field of a ferromagnet from the coil and inductor.
This is achieved by selecting the dimensions of the diameter or sides of a rectangular coil, taking into account the length and width of the core itself and the distances from the wires to the surface of the core.
If the core is made in the form of a wide pack narrow steel plates, then the width of the coil is taken into account where the greatest magnetic scattering of the field occurs along the steel charge.
On top of the coil or in layers interspersed with turns of the inductor, if necessary, according to the scheme, a winding is wound to regenerate magnetic energy into capacitors.
The width of the winding on the core itself is also taken into account, since space and volume are needed for the formation of a ferromagnetic stray field inside the coil.
If a long core is inserted into the inductor coil partially and with only one end, then its ferromagnetic field is a vortex and its magnetic equator is shifted closer to the middle of the core.
The magnetic field of a ferromagnet is, as it were, taken out of the coil by its vortex, which sharply weakens its magnetic coupling.
Since any coil is a magnetic dipole, two long cores can be placed along (and across) the axis of the coil.
In the axial position, it is optimal to insert two relatively long cores into the coil with their ends and bring out the main magnetic vortices of the ferromagnet.
Open free ends can be closed with magnetic shunts or narrow corner and end protrusions can be made, which sharply reduce the demagnetizing factor.


Perhaps a better starting point is the diagram shown in Melnichenko his video at the start of this thread where he shows this diagram:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45083;image)


The formula above the both coils and the layout of the coils may give some idea of the setup used:  D=2R=L + 2a
"D" being the inner diameter of L1 (10cm in my case), "L" being the length of the core and "a" the outer diameter of the core (2cm in my case).

This computes that L (core length) in my case must be 6cm.

The layout shows that the L2 coil itself covers 1/3th of the core in its center, thus 2cm and using 60 turns (3 layers of 20 turns of 1mm litz wire).

I will be building such a L2 coil and start making some measurements with it.

Itsu

New L2 coil made according to above specs, see picture below.

L2 core 6cm long @ 2cm diameter, coil in center core 2cm long 60 turns (3 layers of 20 turns) 1mm litz wire, 0.5 Ohm DC resistance

L2 standalone                 372uH @ 1KHz
L2 inside L1                    369uH @ 1KHz

L1 standalone                 726uH @ 1KHz
L1 with L2 inside             798uH @ 1KHz
L1 with L2 inside shorted 729uH @ 1KHz

Coupling coefficient K      0.29     ( https://www.e-magnetica.pl/calculator/magnetic_coupling_coefficient )


I will do some measurements with this new L2 coil.

Itsu

I did some measurements on this new L2 coil setup:

Using a 100 Ohm 1% induction free resistor as lamp1 and a 12V / 5W lamp as lamp2, i have a 228V BEMF spike across L1 @ 36V input voltage.

FG:1.5KHz @ 10% duty cycle


The input power in that situation is:

with lamp2 load off:  4.52W
with lamp2 load on:  4.52W   so identical.


The power into the 100 Ohm lamp1 load is:

with lamp2 load off:  3.76W
with lamp2 load on:  3.50W


The power into the lamp2 load is:

with lamp2 load off:    0mW
with lamp2 load on: 200mW

So in the original melnichenko circuit with a matched L2 (video), using a 100 Ohm 1% induction free resistor as lamp1 we see a similar behavior as earlier and as with the by Jagau modified melnicheno circuit.
At 36V input voltage, the lamp2 load when switched on takes / shares its power from Lamp1.

Efficiency with:

Lamp2 load off:  3.76W / 4.52W = 0.76 = 76%
Lamp2 load on:  3.50W / 4.52W = 0.77 = 77%

Itsu

Hi Itsu,

Thanks for your devoted efforts on this setup.

With using this D=2R=L + 2a formula, you managed to test yet another size ratio for the coils as Melnichenko indicated. It is strange the efficiency has not improved but become even less than earlier tests.

Of course his patent application includes some pages on core material properties, sizes, many suggested variations too but no practical results as usual.

Gyula

Thanks Gyula,


yes, the results with different sized cores / coils for L2 seem to all point to an efficiency around the 70 - 80%.

All my L2 cores where made of ferrite, so i can try to use iron as that is what is being mentioned in the several video's  from Melnichenko.

Itsu

New L2 coil made using an iron core, see picture below.

L2 core 6cm long @ 2cm diameter, coil in center core 2cm long 60 turns (3 layers of 20 turns) 1mm litz wire, 0.3 Ohm DC resistance

L2 standalone                 167.8uH @ 1KHz
L2 inside L1                    168.3uH @ 1KHz

L1 standalone                 726uH @ 1KHz
L1 with L2 inside             748uH @ 1KHz
L1 with L2 inside shorted 725uH @ 1KHz

Coupling coefficient K      0.175    ( https://www.e-magnetica.pl/calculator/magnetic_coupling_coefficient )


I will do some measurements with this new L2 coil.


Itsu

Hi Folks,

In his patent application WO2022075876A1, Melnichenko wrote about a few things with which the output energy can be increased.

For instance, in Section 0017, Page 6, see PDF file https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45080 (https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45080) he wrote this:

"You can also use anisotropic steel specially made for increased magnetic dispersion not along,
but across the axis of the core with the highest magnetic permeability. For this, the axis of
maximum magnetic permeability and susceptibility is made at an angle to the axis or across
the geometric axis of the core itself, which sharply increases the transverse component of the
magnetic field of the core across its axis. The stray magnetic field across the core axis through
its side surfaces increases sharply due to the fact that the magnetic anisotropy of the steel
does not coincide in direction with the geometric axis of the core. This factor of the magnetic
anisotropy of the material can be used to increase the energy of the free magnetic field of
ferromagnets and the generation efficiency. This is most applicable in electrical steel and high
power applications and where steel grades with pronounced magnetic anisotropy can be used.
This anisotropy of the steel across the axis of the cores can increase the energy of the stray
magnetic field outside the inductors by several times."

In Section 0033, Page 10, he wrote this:

"Another option is to divide the core in height into shorter cores with dielectric gaps between
them to significantly separate their magnetic vortices and magnetic energies.
With this separation, the magnetic field of the cores is closed in a closer zone in distance and
the common magnetizing coil can be made much narrower than with a longer core or cores.
The total magnetic energy removed from such an assembly is several times greater than from
one longer core, since each core in a stack has almost its own separate magnetic energy. In
such a device, each core in the stack must already have its own separate removable winding
to convert its own magnetic energy."

In Section 0034, Page 10 and 11:

"For better magnetization, each such core in a stack is additionally divided into separate
narrower cores in cross section. Removable windings can be on each such core or stack of steel sheets.
Eddy current losses are also reduced by dividing a common steel sheet bundle into narrower and shorter
steel bundles. At the same mass flow rate of the ferromagnet material, the total the magnetic flux
and magnetic energy of such an assembly of shorter and narrower cores can be tens and hundreds or
more times greater than a large whole core of the same mass of material and volume."


In Section 0041, Page 13, he wrote this:

"The use of permanent magnets for magnetization can give the effect of remagnetization of the core and
this can give almost the full amplitude of magnetic induction. This makes it possible to increase almost four times
the total power of the core at a given frequency with the full swing of the magnetic induction amplitude.
This effect is applicable to all devices with unipolar operation to increase generation efficiency."

The use of permanent magnets to speed up the change of the core's magnetic field was also included in an earlier patent application by Melnichenko, WO2017209652A2, see the full description here: http://rexresearch.com/melnichenkoferromag/melnichenko.html (http://rexresearch.com/melnichenkoferromag/melnichenko.html) and I quote the relevant text from it:

"The reverse mode allows you to get rid of the reaction of the current in the secondary winding during magnetization,
but does not interfere with the conversion of magnetic energy of the ferromagnet during demagnetization.
It should be noted that the energy in a ferromagnet is stored in the form of a magnetic elastic interaction energy of the domains,
and this quantity also depends on the initial induction and the external constant magnetic field. When the core is magnetized by
permanent magnets (or currents), it is possible to work on a cycle not of magnetization, but already demagnetization of the core
to zero or of the remagnetization of the core in general in the opposite direction. But the principle of the reverse stroke remains here,
this is the accumulation of magnetic elastic energy in a ferromagnet and then the return of the magnetic induction to the primary state.
The core magnetized by permanent magnets can be demagnetized by the magnetic field of the current and even reversal in the opposite
direction to the magnetic field of the magnets. This allows you to significantly increase the maximum amplitude of the magnetic induction
in the core of a ferromagnet, almost twice (maximum) than just on the magnetization cycle from zero induction.
Also, the presence of a demagnetizing field of permanent magnets makes it possible to reduce the time of decay of magnetic induction to
zero when the magnetization current is turned off and to reduce the residual induction in the core.
The magnets can be located both sequentially at the ends of the core, and parallel to it."

In Section 0049, Pages 15 and 16, this is written:

"You can also use the repeatedly accumulated magnetic energy of the magnetization coil due to
the energy recovery system back to the power source. To do this, several techniques are used. The simplest is a special power recovery
circuit in the form of an oblique diode-transistor half-bridge (oblique transistor-diode half-bridge) with two switches and diode circuits for
the recuperation current. My invention lies in the fact that a removable winding from a ferromagnet is introduced into the diode chains and
not only the recuperation of the magnetic energy of the coil, but also additional magnetic energy of the free field of the ferromagnet goes to the input."

These are interesting thoughts, the question is how these statements hold in a practical circuit ?  I would appreciate reading comments on these energy increase possibilities.
Unfortunately, Melnichenko has published no measurements to back up his statements.

I have attached below all the Claims of his patent application WO2022075876A1.

Gyula

Gyula,

great research   O0  and good to know that Melnichenko has stated in his patents the reasons he thinks what is causing the amplification of magnetic energy.

Also that he mentions that adding one or more permanent magnets to the core can increase this effect.

I have a hard time visualizing how he used his PM's as no video's are showing this, but start experimenting.

If he used solid steel or in strips is not clear with me, so for the time being i will test with my solid iron FE-57 core, but expect heavy eddy currents.

Itsu

   Itsu:
   I have always used egg shaped magnets on the yoke cores and on most other ferrite cores. They do help, but at a higher expense of input power.

I think if the core used has very low remanence (residual magnetism) specification like say soft ferrites,  metglas,  then the use of a permanent magnet may not be needed at all.  This is because when the core is able to lose its magnetization the moment the outside magnetizing field disappears, then there is no need for a PM to help 'flush' the core. 
Iron and steel cores may have unwanted magnetization remaining when the outside field becomes zero.  Say one end of a core remains magnetized with an unwanted (and weak) S pole, then placing a (weak) N pole of a PM close to the core end will speed up the demagnetisation process and the core's S pole will become first zero and then gets weakly biased to have an N pole from PM.
So the next magnetizing field we apply should defeat first this weak bias field, this needs some more input as Nick found.
See how Doug Konzen applied such method in a pulse motor (very likely his cores had certain residual magnetism):
https://sites.google.com/site/alternativeworldenergy/re-guaging-magnets-in-dc-pulse-motors (https://sites.google.com/site/alternativeworldenergy/re-guaging-magnets-in-dc-pulse-motors)

In the Melnichenko setup discussed here you could attach a weak ceramic magnet with one of its poles to one of the ends of the core (test which pole helps), either directly or via a non-magnetic spacer, especially in case of the solid iron rod core you test now.  Question is how much loss the solid rod causes in itself from which any improvement can still manifest by the use of a PM. 

Gyula

I did some measurements on this new Iron L2 coil setup:

Using a 100 Ohm 1% induction free resistor as lamp1 and a 12V / 5W lamp as lamp2, i have a 293V BEMF spike across L1 @ 36V input voltage.

FG:1.5KHz @ 10% duty cycle


The input power in that situation is:

with lamp2 load off:  4.55W
with lamp2 load on:  4.37W   Less when L2 on!


The power into the 100 Ohm lamp1 load is:

with lamp2 load off:  3.89W
with lamp2 load on:  3.97W    More when L2 on!


The power into the lamp2 load is:

with lamp2 load off:    0mW
with lamp2 load on: 32mW     Very marginal power!

So in the original melnichenko circuit with a matched IRON L2 (video), using a 100 Ohm 1% induction free resistor as lamp1 we see a different behavior as earlier and as with the by Jagau modified melnicheno circuit.
At 36V input voltage, the lamp2 load when switched on does NOT take / shares its power from Lamp1.

Efficiency with:

Lamp2 load off:  3.89W / 4.55W = 0.85 = 85%
Lamp2 load on:       4W / 4.37W = 0.92 = 92%         Max. efficiency up till now!

Itsu

gyula has asked me to comment on Melnichenko's statements he posted in Reply#40 above.  The set up Itsu has just tested can be modeled in FEMM using the axisymmetric method which is a true 3D solution.  I have done this modeling the core with greater mu in the r dimension (the radial dimension) than the z dimension (the length dimension).  I used mu = 10,000 for the radial dimension and mu = 100 for the length dimension.  There is no overunity in that situation.  Melnichenko is correct in making observations that the anisotropy of the core will affect the magnetic energy stored for a given ampere turns of current, but I can find no evidence that this can lead to OU.  My opinion on his other statements of dividing the core into individual sections each with its own winding is that also will not produce OU.  I will have a go at doing a simulation of such an approach and report back.
Smudge   

Hi Itsu,

Interesting results with the solid iron rod core. The less input power draw when lamp2 is ON indicates the existing eddy current loss in the core is partially compensated by the load current from lamp2. When you have time, please check L1 coil current and voltage waveforms when lamp2 is OFF or ON, whether anything manifests on them.

Smudge, 

Thank you for taking the time to go through Melnichenko's statements.  I think he mentioned the use of the possible anisotropic property of a core because such property can help store (and capture) more field from a given amount of input energy versus a normal core which has no such property. 
Regarding his statement on the use of many individual core sections each with its own winding may be the conclusion coming from finding eddy current losses when he used solid steel or iron cores, so by dividing such solid cores can surely help reduce or eliminate the losses. 

Thanks,
Gyula

Gyula,

The L1 current and voltage waveform hardly change when switching on or off the L2 load.
There is a slight increase in Voltage (p2p from 298 to 302V), current and thus power.

Using a stack of ceramic magnets on the iron core does not show any deviation on the L1 current / voltage, but when monitoring the L2 voltage / current, we see a small (±2mW or so) deviation on the power,
2mW low when using 1 side of the magnet stack, 2mW higher using the other side, but with 32mW total this is very marginal.


Itsu

Thanks Itsu for checking the waveforms.

Yes, the small deviations caused by the magnet stack poles indicate that either the iron core losses hamper what the "regauging" effect by the magnets could improve on the output or the iron core does not need much help to lose quickly its induced poles when the input current is switched off. 

Gyula

This may not be relative to this thread so please delete if this be the case.

Many times in patents or other IP material, it's what is not said that can become the key to understanding as compared to what is said.  This could be the case with Melnichenko's devices.

For example, the paper below I did several years ago regarding core remanence in ferrite material shows the different energy levels to reach a given inductor current depending on the core remanence.  To be clear, the core was first setup or run with the appropriate 1 of 4 schematic and then was run with a "single cycle" using the NZR schematic.  Yes I know remanence is misspelled on the schematics. 

I have a transformer device using RLE that exhibits gains from 1.05~1.15 that has always puzzled me as to the source of gain but now I believe it is due to retained positive core remanence between cycles.

With iron having a much larger remanence than ferrite, Melnichenko may have used a similar setup with single cycles to do his tests.  Jus' saying!

Regards,
Pm

Thanks, PM,   this remanence could play a role, but this asymmetry would show up in a hysteresis curve i guess?

Not that that easy is to check.

Itsu

Quote from: Itsu on 2022.08.02, 08:47:26
Thanks, PM,   this remanence could play a role, but this asymmetry would show up in a hysteresis curve i guess?

Not that that easy is to check.

Itsu

Actually, the more I consider this, I think that the possibility of remanence playing a part in Melnichenko's device is not probable.  The reason being the core material IMO is not reaching a non-linear or saturation point in the hysteresis curve.  IOW, the hysteresis curve will be very flat on itself and would thus exhibit very little remanence.

Pm

New BIGGER L2 coil made using 3 stacked ferrite toroids as core, see picture below.

L2 core 8.6cm long @ 5cm diameter, coil 8.6cm long 66 turns (3 layers of 22 turns) 3.4mm stranded wire, 0.3 Ohm DC resistance

L2 standalone                 660uH @ 1KHz
L2 inside L1                    655uH @ 1KHz

L1 standalone                 726uH @ 1KHz
L1 with L2 inside            1083uH @ 1KHz
L1 with L2 inside shorted  595uH @ 1KHz

Coupling coefficient K      0.67   ( https://www.e-magnetica.pl/calculator/magnetic_coupling_coefficient )


I will do some measurements with this new L2 coil.


Itsu

Please check edited data below in bold



I did some measurements on this new BIGGER L2 coil made of 3 stacked ferrite toroids as core

Using a 100 Ohm 1% induction free resistor as lamp1 and a 12V / 5W lamp as lamp2, i have a 219 / 211V BEMF spike across L1 @ 36V input voltage.

FG:1.5KHz @ 10% duty cycle


The input power in that situation is:

with lamp2 load off:   3.2W
with lamp2 load on:   3.2W   


The power into the 100 Ohm lamp1 load is:

with lamp2 load off:   3.2W
with lamp2 load on:   2.1W   


The power into the lamp2 load is:

with lamp2 load off:    0mW
with lamp2 load on: 855mW     some light in the 12V / 5W lamp


At 36V input voltage, the lamp2 load when switched on does take / shares its power from Lamp1 once again.

Efficiency with:

Lamp2 load off:  3.2W /  3.2W = 1      = 100%            Max. efficiency up till now with L2 off
Lamp2 load on:  2.95W / 3.2W = 0.92 = 92%         

EDIT:

The 100% efficiency in the above efficiency results should have warned me that something was wrong.

I tried for several days now to remeasure this circuit, but i am not able to come to the reported 100%, so very probably
i made some mistake somewhere in the measurements.

The new measurements over several days are all very close to each other so my new results with this bigger L2 coil should be:

The input power is:

with lamp2 load off:              3.36W
with the lamp2 load on:       3.36W 

the power into the 100 Ohm lamp1 load is: 

with lamp2 load off:           3.04W
with lamp2 load on:           1.98W

The power into the lamp2 load is:

with lamp2 load off:          0mW
with lamp2 load on:    894mW

Efficiency with:

Lamp2 load off:             3.04W /  3.36W = 0.90       =   90%                                                   
Lamp2 load on:              2.87W /  3.36W = 0.85       =   85%     


Itsu

Hi Itsu,

Congratulations on achieving the 100 % efficiency, I assume you were surprised to measure that.  This is a great progress from the 70-80 % efficiency range you experienced so far with the previous tests.
Thanks for all your arduous efforts.

Gyula

Hi Gyula,


yes, I was surprised about the 100%, I will redo these tests tonight once again to be sure.

Itsu

Above measurements redone, now checking for 2 decimal digits (for what its worth), but basically it had the same results.

The power into the 100 Ohm lamp1 load is "with lamp2 load off:   3.18W" instead of 3.2W, which makes the efficiency with "Lamp2 load off:" 99.6%

Itsu

Doing some screenshots on my present "ORIGINAL Melnichenko" circuit, see diagram.

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45131;image)


First screenshot is showing the voltage / current across / through L1 (trigger point) in yellow / green together with
the voltage / current across / through  L2 in blue and purple.

My interpretation of the signals are noted in the screenshot.

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45127;image)

The second screenshot is of a frequency sweep while monitoring the voltage across L1 in yellow and L2 in blue while the circuit was unpowered.
I loosely couple (1 turn) a sweeping FG with L1 while recording the voltage across both coils.

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45129;image)

Never i have read that Melnichenko uses resonance to accomplish his free energy, but its nice to know where the coils resonance points are and
as can be seen they are way up in frequency like 100KHz for L1 and 680KHz for L2.
Sweep is from 10Hz to 1MHz, so basically we have 100KHz / division.
The red line is the used 1.5Khz pulse frequency.

I have up till now with all my Melnichenko experiments seen NO sign of any abnormalities on the scope nor as an effect in the lamps.
So i guess its time for other replicators to show their circuit, their in- / output measurement results, their signal screenshots and their abnormalities noted, especially
with COP > 1.   

Until then i will work on my present original Melnichenko setup as it has the highest efficiency up till now and trying to improve on.


Regards Itsu 

Itsu,

Thank you for your comprehensive professional report.

I can basically confirm your findings. I varied the input frequency for maximum lamp brightness then observed that just the addition of a core diminished the inputs lamp brilliance. The addition of a core and coil/lamp diminished it still further as the output of coil two came at the expense of coil one's output.

I was waiting for certain people to publish the secret, hasn't happened.

Ron

Hi Ronee,

thank you for revealing your results, i did see them also on aboveunity.com.

Yes, i hope more people who are replicating and have better results come forwards with their results, so we can see where we go wrong.

Regards Itsu

Concerning the polarization of L2 in the above screenshot with the voltage and currents, it deliberately is chosen that way so that the power induced into L2 comes from the "demagnetization phase" aka "collapsing magnetic field phase" aka the "bemf phase" of the cycle, as that is, according to Melnichenko and Jagau, where the free energy comes from.
Jagau correct me when wrong!


The below screenshot is from when the L2 connections are swapped, so showing the other polarization effect of L2:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45135;image)




We see that now the Power into L2 is induced during the magnetization phase of the cycle which is, according to melnichenko and Jagau, NOT where the free energy comes from.
Jagau correct me when wrong!

If those signals are still wrong, even when they are coming from a close replication of the circuit used by Melnichenko, then the circuit is wrong or the effect claimed is not there.

Itsu

Concerning replications:


When an inventor (Melnichenko) or a replicator (Jagau) claims a certain effect (like COP > 1 by both),  and REQUESTS other replicators to join them to confirm this claim,
it is IMHO the DUTY of the claimant to provide the replicators with ALL information needed to come to the same result (COP > 1).

If you want to entertain people by presenting puzzles to them, you should not REQUEST for replicators to join, but puzzle enthusiasts.


Throwing bread crumbs and expecting the replicators to figure out how it works without showing the used circuit, some in- and output measurements, components used, etc.
will quickly raise the suspicion that the claimant does NOT have anything special to show and is provoking the replicators to connect the dots and find the effect for him.

I know Jagau is not such a person and there are probably some other factors involved why he is so reluctant to provide details of his working circuit, but this leads to
nothing and only gets the honest working replicators frustrated (not to mention the costs involved).

We are all in this for the benefit of humankind, so playing games on "who is the best" or "who is light years ahead" or "who knows it all" is childish the least and bordering to unacceptable behavior IMO.

Itsu

Quote from: Itsu on 2022.08.09, 10:55:38
Concerning replications:


When an inventor (Melnichenko) or a replicator (Jagau) claims a certain effect (like COP > 1 by both),  and REQUESTS other replicators to join them to confirm this claim,
it is IMHO the DUTY of the claimant to provide the replicators with ALL information needed to come to the same result (COP > 1).

If you want to entertain people by presenting puzzles to them, you should not REQUEST for replicators to join, but puzzle enthusiasts.


Throwing bread crumbs and expecting the replicators to figure out how it works without showing the used circuit, some in- and output measurements, components used, etc.
will quickly raise the suspicion that the claimant does NOT have anything special to show and is provoking the replicators to connect the dots and find the effect for him.

I know Jagau is not such a person and there are probably some other factors involved why he is so reluctant to provide details of his working circuit, but this leads to
nothing and only gets the honest working replicators frustrated (not to mention the costs involved).

We are all in this for the benefit of humankind, so playing games on "who is the best" or "who is light years ahead" or "who knows it all" is childish the least and bordering to unacceptable behavior IMO.

Itsu

Itsu,

I wholeheartedly agree with your statements above! 

Regards,
Pm

Quote from: partzman on 2022.08.09, 13:46:28
Itsu,

I wholeheartedly agree with your statements above! 

Regards,
Pm

   Itsu:
    Thank you for your efforts. I also agree with your statements.
And, I am sad to see that all the different tests of the different free energy devices have resulted in zilch.

   I am about to disassemble the Stalker replication that has been sitting on my bench for several years, which was still waiting for the missing link, never to be found.
   A sad day for me, as well.
   Thanks, again.

   NickZ

Thanks Partzman,  thanks Nick,

Nick, i am sorry to hear you are going to disassemble your Stalker / Ruslan replication, i know how you believe(d) in it.
Mine is put in a box ready to reassemble when needed for some reason, you never know,  time will tell.

Thanks for joining me building one then.


Itsu

Folks,  here are some FACTs against Chris's claims he wrote here https://aboveunity.com/thread/melnichenko-s-effect/?order=all#comment-9ba8d63d-398e-4072-aec2-aeec0134ad63 (https://aboveunity.com/thread/melnichenko-s-effect/?order=all#comment-9ba8d63d-398e-4072-aec2-aeec0134ad63)

I include his relevant text here:

"Captainloz gave you 100% of the data, as has Jagau, and you still cant make it work!

In the end, we have given you 100% of the data! Its all here on THIS Website!
    Polarity of L1's Voltage changes at Mosfet T_Off, the CURRENT Does NOT!
    Polarity of L2's Voltage and Current do NOT Change!
    This is purely a DC System and there is no AC Component here!
    Voltage is Generated in L2 at Mosfet T_On and the Potential of BOTH L1 and L2 increase to VMax over Time t, at this point the Mosfet is turned Off, T_Off, and the Demagnetisation Phase occurs!
    The Demagnetisation Phase must always be longer in Time t than the Magnetisation Phase!
    Through this entire period, L1 and L2 oppose each other!"


Claim 1:
"Captainloz gave you 100% of the data, as has Jagau, and you still cant make it work!"

Fact:  Chris here means the POC setup for which Captainloz measured COP of 2. Captainloz used his precision current shunt resistor (metal strip through-hole resistor) at the 800 kHz frequency range, where the U shaped metal strip manifested 35 nH self inductance as per Itsu measurements on the same off the shelf shunt resistor.  Itsu used Vector Network Analyzer to characterize the frequency dependence of the shunt, see data here:
https://www.overunityresearch.com/index.php?topic=3691.msg85851#msg85851 (https://www.overunityresearch.com/index.php?topic=3691.msg85851#msg85851)  and the original 0.1 Ohm (at DC) manifested 0.23 Ohm impedance at 800 kHz, a 2x increase, it explains Captainloz COP of 2 "measurement" on his POC setup. Chris keeps deep silence on this fact.

Fact: Jagau has not given 100% the data, what he mentioned was "Melnichenko did not clearly reveal how to do it, so why is so much expected of us? Work a little, it does not kill anyone." He has not shown measurements like Itsu did on the input output-power relations, he only claimed 4.5 W in, 7 W out. This is not a complaint from me, just giving readers the facts.       

Claim 2:
In the end, we have given you 100% of the data! Its all here on THIS Website!
    Polarity of L1's Voltage changes at Mosfet T_Off, the CURRENT Does NOT!

Fact: He says the obvious: when the current in a coil is switched off, the polarity of the voltage across the coil changes to the opposite polarity of the input voltage (back spike appears) and the coil tries to maintain its current it has had at the switch-off moment. As the collapsing field goes down so does the current and its polarity does not change, a well known fact. 
And a scope shot by Itsu on his Melnichenko setup clearly shows this behaviour, see it in his Reply #57 here https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45127;image (https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45127;image)  for instance the yellow (L1 voltage) and green (L1 current) scope traces.  These traces show exactly the obvious behaviour, question is why Chris says this belongs to his given data? Any decent website dealing with coil current switch-off topic discusses this obvious behavior.
     By the way, neither Itsu nor anyone else said the current would change at Mosfet T_Off, why Chris mentions this as if it were said ??

Claim 3:
Polarity of L2's Voltage and Current do NOT Change!

Fact: The first part is not correct: the voltage polarity across L2 changes, see Itsu's blue trace (L2 voltage) in his scopeshot (referred to above) when the current is off in L1. This is because during the magnetization phase L2 senses an expanding and increasing magnetic field coming from L1 and at the current switch-off moment the magnetic field starts collapsing, hence the induced voltage in L2 should change polarity. Basic Faraday induction on the direction of moving magnetic field.

Fact: the second part of Claim 3 is incomplete, the L2 current is zero during magnetization (D2 and D3 diodes block current as wanted) and starts flowing at the moment of the input current switch-off in L1. So current changes from zero to a maximum value in L2 during the demagnetization, then reducing to zero as the collapsing field diminishing also to zero. 

Claim 4:
  This is purely a DC System and there is no AC Component here!

Fact: This is a pulsed DC system. Who said there is AC component here?

Claim 5:
Voltage is Generated in L2 at Mosfet T_On and the Potential of BOTH L1 and L2 increase to VMax over Time t, at this point the Mosfet is turned Off, T_Off, and the Demagnetisation Phase occurs!

Fact: This is also claiming the obvious, Itsu's scope shot clearly shows this, and this is the wanted operation established by Melnichenko, and Jagau also wrote this is the needed modus operandi. So what??

Claim 6:
The Demagnetisation Phase must always be longer in Time t than the Magnetisation Phase!

Fact: Yes, this is also achieved by Itsu, in the scopeshot referred to above the magnetization time is around 63 us and the demagnetization time is around 90 us for L1 voltage and current, and around 110 us for L2 voltage and current, this depends on the load resistances too. So what ??

Claim 7:
Through this entire period, L1 and L2 oppose each other!

Well, this needs explanation what exactly Chris means here.  Because, during the magnetization time L2 is not active magnetically (for it has no current) so the coils cannot oppose each other through the entire period. Entire period for me means the time from one switch-on moment to the next switch-on moment. Of course, the input current in L1 magnetizes the core in L2 but the two coils cannot oppose each other through the entire period. 

Anyway, these are as Chris says the 100% given data, so where are the data?

On Chris's forum, there are 3 other replicators, besides Jagau, who actually built the Melnichenko setup under discussion as Jagau described. Why is it then that they achieved the same 75% to 93% efficiency just like Itsu did? Where is the famous 'helping each other' claimed so much and so often on Chris's forum?

One more thing worth mentioning: IF the POC setup works with COP > 1 performance, then where is its practical application? What does Chris use it for during the long years? What do the many 'successful' replicators on his forum use the POC for?

Since the many years he has come up with such 'aboveunity' claim but it has never ever proved by him or by anybody with correct measurements, why is that?  IT is ridiculous to have fear from showing measurements.
This situation is like he has been shelving a COP > 1 device for years now (but of course if it indeed performs like that). The setup surely works, the question is its input and output power ratio.
   
And Chris can call me or Itsu or all others as trolls, trouble makers,   it does not matter, he has been doing this for years with anyone asking for proof (and he bans any member on his forum who does not 'behave'). The right questions should always be asked and he avoids answering them correctly at all cost. Why is that?  Just for fear, or ?

Gyula

Gyula,

Well stated and true!!!

Pm

Quote from: Itsu on 2022.08.09, 18:51:21

Nick, i am sorry to hear you are going to disassemble your Stalker / Ruslan replication, i know how you believe(d) in it.
Mine is put in a box ready to reassemble when needed for some reason, you never know,  time will tell.

Thanks for joining me building one then.

Itsu

:'(

Gyula,

thanks for the extensive point by point analysis, this is a skill i lack still and very useful.


Like Chris, i would like to see some different kind of signals coming out of this circuit (less input and more output preferable), but the scope does not lie and presents these conventional signals belonging
to this conventional fly back circuit.

I can not magically change those signals to something more promising, it is what it is.

I tried both the original Melnichenko circuit and the modified by Jagau circuit with several coils and cores, but they all stay within the conventional 70 to 90% efficiency range.


I still have some modifications pending on the L1 and L2 coil / core, presented by Jagau, so we will see if it will change much.

Regards Kees 

Hi, everybody.

Firs Itsu, I'm sorry to bring you in my theme on Chis forum, but it was my last post before i got kick out with no reason, and a have to know how naïve some people can be. For me the only proof of OU is a self runner. No scope no measurement nothing else. I spend a lot of money for a device which should be so easy and cheap ( 20$), for nothing. When you start to ask question
you got banned and got a stamp as a troll. I did need a lot of time to finally figure out who is who. I reading your journey with
big interest, thanks for sharing all and for all your time and money invested in the research. But unfortunately your last post here is right on the target.
Wish you all the best, Editor The Cook

Hi TheCook,

i agree with you about: "For me the only proof of OU is a self runner.", that's the ultimate proof.

But as nobody ever has shown a self runner for me to replicate, i have to lower the standard and replicate circuits / devices for which people claim they are over-or above unity.

But as a non Electronics trained person, i have to rely on the cooperation and information of the person claiming this.

If this information stalls somehow, then i can come a long way especially with the help of the many knowledgeable people on forums like this one, but ultimately the one making
the claim should step up to the plate IF that's what he is looking for in the replicators.

If you have something interesting, i would suggest opening a thread here and present your circuit, so others can join in and advise you or even replicate it.

Regards Itsu 

Quote from: Itsu on 2022.08.10, 20:35:49
"For me the only proof of OU is a self runner.", that's the ultimate proof.

I'd argue that a principle-of-operation and operating constraints are even more important than a working device.

A working device is a 'miracle' and unless it is understood it will always remain an oddity/serendipity.

  Someone might show a simple circuit with a coil apparently producing power, but that doesn't mean someone can understand or recreate it.    Perhaps the builder left out the fact that one of the coils was wound with iron wire and acts like a magnetic amplifier, or that it only works at voltages over 1500v, or that it requires a bias current to be injected just prior to triggering?  Or that one of the coils must be wound CCW, or that two coils must be of precisely equal lengths.  Or perhaps the builder doesn't even know how it works, they just know a circuit configuration that works most of the time. ^-^

You can't just show someone under the hood of a modern car and expect them to be able to build a working Model T or understand internal combustion. ;D

Hi Hakasays,

good point, if someone is ONLY showing a self runner for me to replicate i would have nowhere to begin, so there must always be additional info to be given for a replication.

A principle-of-operation and operating constraints would be a good start  O0


Itsu

I made a LTspice sim of my present original Melnichenko circuit.

The results can be seen below (i am zoomed in in the trace plot and i have reversed the currents to match my screenshot):
A bigger picture can be seen in the 2nd attachment at the bottom.

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45197;image)

Please compare this with my earlier made screenshot in post #57 here:  https://www.overunityresearch.com/index.php?topic=4312.msg99826#msg99826

It has the same colors and probe points.         


(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45127;image)


The sim file (.asc) is attached below, so anyone can play with the polarities of L2  and / or the L2 diodes to see what happens.


Regards Itsu

Last 2 posts removed being off-topic.


New L2 coil made of 3 coils in series each on a ferrite toroid with 5 mm gap in between as suggested by Jagau.     40 turns each so 120 turns total.

L2 standalone                 1093uH @ 1KHz
L2 inside L1                    1090uH @ 1KHz

L1 standalone                 726uH @ 1KHz
L1 with L2 inside             944uH @ 1KHz
L1 with L2 inside shorted  559uH @ 1KHz

Coupling coefficient K      0.64   (https://www.e-magnetica.pl/calculator/magnetic_coupling_coefficient)


I will do some measurements with this new L2 coil.

Itsu

Unfortunately, Jagau has decided not to share any more his findings on his Melnichenko replication due to several reasons, among others being a breach of trust.

I am not aware when i breached that trust, but i apologize for that and have to respect that decision.


I will stop my replication here, but hope that others will join him at his forum to continue to work on this Melnichenko replication.


Regards Itsu

I tried to make sense of what is on public display at that site and I can only find "emotional facts". If that is even a word?

I think that it is important for that site to contribute to the crashing mag field hypothesis or you may be seen to be betraying the

"great discovery" that is revealed there.

It reminds me of religious teachings .

You did an excellent job of duplicating and any body would be willing  to help you if only they were able.
Well done .

There is a much better project here if you get the time and don't mind high voltage .

3D, thanks for your insights.

The projects you're referring to are from Hakasays i guess, they are indeed very intriguing and of high standards, no doubt, but i don't think they are concerning overunity per se,
or you must consider partially lightning up a 20W fluorescent tube driven by a 0.7W FG   O0  that, but it will be hard to measure me thinks.

Regards Itsu

I agree


The steap project .

There was some misunderstanding between Jagau and me, which now is cleared up.

I will continue with this replication here in cooperation with Jagau and his thread on AU.com which is not an ideal situation but the best we can do right now.


Itsu

Hi Itsu,

one idea for you if you like - look for 1/2 wave resonance of your secondary winding and try all setup on this frequency.

Regards,
Alexey

Hi Alexey,

thanks, I was and am doing some resonance testing on L2 (10uF cap parallel to L2) as some Melnichenko video's show such a tuning cap.

Did you come across any information on this tuning cap in one of his video's (1/2 wave, 1/4 wave)?

Itsu

Quote from: Itsu on 2022.08.19, 10:19:19
Hi Alexey,

thanks, I was and am doing some resonance testing on L2 (10uF cap parallel to L2) as some Melnichenko video's show such a tuning cap.

Did you come across any information on this tuning cap in one of his video's (1/2 wave, 1/4 wave)?

Itsu

Itsu,

No, I am not following Melnichenko's videos, I do not have respect for people who behave like him and prefer not waste my time.

It is just an idea. When you have ferrite core in the coil wave resonance will be in kilohertz region without capacitors. It might be be not easy to find, but it is there. Once you tune to this frequency(or harmonic) you will get additional gain due to this resonance, in addition to regular flyback operation. This will give some FE if you not loading device too much.

Regards,
Alexey

Alexey,

thanks, a remarkable reply, not sure what you mean by that first part, but you will have your reason.


I did make a sweep of the both coils earlier ( https://www.overunityresearch.com/index.php?topic=4312.msg99826#msg99826 ), and found the L2 resonance (without parallel cap) to be around 680KHz.

This is way above the pulsing frequency of 1.5Khz, which is the 453rd subharmonic, so i guess not really realistic to use, but you never know.

Regards Itsu

Quote from: Itsu on 2022.08.19, 19:20:20
Alexey,

thanks, a remarkable reply, not sure what you mean by that first part, but you will have your reason.


I did make a sweep of the both coils earlier ( https://www.overunityresearch.com/index.php?topic=4312.msg99826#msg99826 ), and found the L2 resonance (without parallel cap) to be around 680KHz.

This is way above the pulsing frequency of 1.5Khz, which is the 453rd subharmonic, so i guess not really realistic to use, but you never know.

Regards Itsu

Itsu,

When looking for 1/2 wave resonance you need short secondary and look on current e.g. with current probe, also I think you need slow scan, it is easy miss this resonance.
It is possible to drive with short pulses 1.4us (1/680K) and still have frequency 1.5KHz. Other possibility is
to drive with pulse bursts e.g. 5 or 10 1.4us pulses, then pause for 1.5KHz.
This is probably going too far away from Melnichenko description...

Regards,
Alexey

Thanks Alexey,   sweeping with the current probe sounds i good idea.

Normally i sweep for 10 seconds, so fairly long.

I will see what i can do.   

Itsu

My latest post concerning the replication was about the again new coil L2 consisting of 3 coils in series with gaps in between and all data which I repeat here:

=====================================================================================================================================================
New L2 coil made of 3 coils in series each on a ferrite toroid with 5 mm gap in between as suggested by Jagau.     40 turns each so 120 turns total.

L2 standalone                 1093uH @ 1KHz
L2 inside L1                    1090uH @ 1KHz

L1 standalone                 726uH @ 1KHz
L1 with L2 inside             944uH @ 1KHz
L1 with L2 inside shorted  559uH @ 1KHz

Coupling coefficient K      0.64   (https://www.e-magnetica.pl/calculator/magnetic_coupling_coefficient)


I will do some measurements with this new L2 coil.

Itsu
=======================================================================================================================================================

The measurement results of this (original melnichenko) setup are:

Using a 100 Ohm 1% induction free resistor as lamp1 and a 12V / 5W incandescent lamp as lamp2, 36V input voltage.


FG:1.5KHz @ 10% duty cycle

The input power in that situation is:

with lamp2 load off:   3.97W
with lamp2 load on:   3.97W   


The power into the 100 Ohm lamp1 load is:

with lamp2 load off:   3.2W
with lamp2 load on:   2.15W   


The power into the lamp2 load is:

with lamp2 load off:    0mW
with lamp2 load on:  1.01W     some light in the 12V / 5W lamp


At 36V input voltage, the lamp2 load when switched on does take / shares its power from Lamp1 once again.

Efficiency with:

Lamp2 load off:   3.2W /  3.97W = 0.80  = 80%           
Lamp2 load on:  3.16W / 3.97W = 0.79 = 79%         

Itsu

Jagau pointed earlier in his thread to a new coil setup in which we have the L1 air coil in the middle, and 2 L2 coils with T-cores inserted joining at both sides of this L1 coil.

I 3D printed such a former with the L1 air coil in the middle, a 3 mm gap on each side, then the both L2 coils which have the T-cores (part of E-cores) inserted, see picture.

The coils measure:

L1 80 turns (5x16) 0.8 mm diameter magnet wire, L2 48 turns each (3x16) 0.8 mm diameter magnet wire.

L1 (air) measures 116uH with both T-cores inserted in the L2's and resp. 108uH and 104uH when removing the L2 cores.
L1 measures resp. 107 and 100uH when shorting L2's.

L2's measure 176uH each.

So we have a low coupling factor K of resp. 0.28 and 0.37.

I will do some testing with those and look into the use of a "tuning cap" parallel to the L2's for power factor compensation and / or resonance tuning.

This testing will be done using the "modified by Jagau" circuit as shown here: https://www.overunityresearch.com/index.php?topic=4312.msg99530#msg99530 but with the new
L1 / L2's coils and a 12V / 5W incandescent automotive bulb as Lamp2


Itsu

Hi Itsu,

You have been making great progress in testing the various coil constructions, thank you for your kind efforts and for sharing the results. 

Gyula

The measurements using the new combined L1/L2 coils is on hold right now as there is some discussion about how to accurately measure the input power.


Up till now I used the filtered DC input Voltage and current at the entry of the circuit right after the DC Power Supply, see "screenshot 1" probe points:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45338;image)

The result is a straight DC voltage (yellow) and a positive fluctuating current (blue):

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45340;image)


Using this as input power, and measuring the combined powers of the loads (lamp1 and lamp2) as output power, yields an COP of around 0.8, measured over several circuits and coils.


By using an alternative input method where the input is measured somewhere else in the circuit and then calculations are made for input power, the COP should be > 1.

 
I am waiting to get a clear picture where the Voltage and Current probes should be placed, so I can make these alternative input measurements and do the appropriate calculations to measure/calculate the input.

"Smath solver"program will be used to make the calculations.

Itsu

Itsu,

In a periodic operating circuit which you have here, and there is no gain being produced by the circuit, it will not matter where you measure the input if all the energies are added.  For example, where you are currently measuring the input power will yield the lowest COP relative to the output because you are not considering the losses in the cmc. 

If you  move the voltage and current probes to the right of the cmc, you will now yield a slightly higher COP but will still be conservative or COP<1 unless the circuit under test is OU.  This is because you have eliminated the losses in the cmc but, if all the losses are considered when measuring from the left side of the cmc, the COPs will be identical.

This will also apply if the current probe is placed on the left or right of C1.  However, the re-charge of C1 must also be considered which will be difficult to measure especially with larger values of C1.

Regards,
Pm

PM,

thanks, I agree, but I understand that the voltage probe (yellow) and current probe (green) possible need to be positioned as in the below diagram (need confirmation still):

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45346;image)

This way we not only omit the losses in the CMC, but also of the 24K resistor across C1, the Vce(on) loss (few volts) of the IGBT Q1 (IRG4PH30K) and even the FG loss.


Itsu

Itsu,

The position of the current probe as you show it does not include the current on the negative side of the power supply to the circuit.  The current probe should be positioned ahead or to the left of the junction of the anode of D1 and L1.  Or, the current probe could be placed on the emitter of Q1.  However, Q1 is instrumental in producing the periodic operation of the circuit and therefore should be included in the overall power/energy analysis to be really honest about it.  This would mean even including the gate drive energy to Q1 which could be independently measured.

Pm

An additional remark:

The current probe when placed in the position Itsu indicated in green (in his Reply # 91 above) measures the flyback current of L1 during the demagnetization time if one wishes to learn about the flyback current value.
At that position the current probe cannot and does not consider the current on the input side of course.

Gyula

To minimize the loading effect of the probe, clamp it at the low or ground end of a component lead when possible. This method also minimizes noise or stray
signal interference. This is a Tektronix recommendation

On this side you have the magnetisation


Si Itsu you are correct on the low ground side

Guys,  thanks,  good discussion.

So lets see what the different positions of the current probe does with the shape of the current signal.


I have put the current probe in 3 positions in the circuit, see each diagram and the accompanying current waveform.

1st is the current probe at the anode of the diode.

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45350;image)
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45352;image)





2nd is the current probe right of the junction with the diode
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45354;image)
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45356;image)







3rd is the current probe left of the junction with the diode
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45358;image)
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45360;image)




What we see is that the first position shows the demagnetization current only.
The second position shows both the magnetization AND the demagnetization currents.
The third position shows the magnetization current only .

Itsu

Quote from: Itsu on 2022.08.27, 11:24:02

...

What we see is that the first position shows the demagnetization current only.
The second position shows both the magnetization AND the demagnetization currents.
The third position shows the magnetization current only .

Itsu

Hi Itsu,

Very clear and telling measurements, thanks for sharing these. 

Gyula

I think we can assume that for input power measurements we should put the current probe left of the junction with the diode (position 3) to measure the magnetization phase current, so we can collect some data to calculate this input power.   Yes?  /  No?

I use this diagram with the probe positions green: current probe, yellow: voltage probe, purple: my normal input power probe points:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45388;image)
 
I toke several zoomed in screenshots showing the data we need to calculate input power.


Screenshot 1 shows the timing relations of the full cycle (1500KHz) and the on-time pulse:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45390;image)

 
screenshot 2 shows the measured average voltage (yellow), current (green) and power (red) of the ON-TIME pulse of 62.5us:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45398;image)
 


Data on screenshot 1 shows we have:

a full cycle of 667us = 1500Hz 
an ON-time pulse of 62.5us
a duty cycle (DTC) of 9.37%


Data on screenshot 2 shows we have In between cursors A and B:

an average pulse voltage (Vpave) of 33V,
an average pulse current (Ipave) of 1.158A (roughly half of Ipeak 2.313A)
an average pulse Power(Ppave) of 38.52W in this 62.5us pulse ON-TIME.


From this we can calculate the average power over a full cycle as:

Pfave = Ppave x DTC = 3.61W.



As a double check we can use multiple cycles  (6)  for measuring the average input power confirming the 3.6W input power like:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45394;image)

And finally we can use the earlier proposed method (here: https://www.overunityresearch.com/index.php?topic=4312.msg99687#msg99687 ) (method 4) to calculate the stored energy in a coil using this formula:  E = 1/2 x L x I²       
https://physicscalc.com/physics/inductor-energy-calculator/
For L= 945uH (L1), I= 2.3A we get 0.0025278J (Ws), so for 667us we get 0.0025278/0.000667=3.79W
Which is close to the 3.6W but deviates somewhat due to the 945uH being measured outside the circuit at 1KHz while we work at 1.5Khz etc.

Measuring the input power, the way I normally did, can be seen in screenshot 3:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45396;image)

Where we have an average input voltage of 36.1V (yellow), an average input current of 107.4mA (blue) and a calculated average input power of 3.87W (red).

This 3.87W input is some 260mW more than the 3.61W calculated earlier and can be explained by some losses like:
# the heat loss in the 24K resistor (P=U²/R =  36²/24000 =) 54mW
# Vce(on) saturation voltage loss of the IGBT (typical 3V according to the IRG4PH30K datasheet)

So I think we can conclude that this 260mW loss is marginal and that the different shapes of the voltage and current signals caused by the filtering components has no influence on the energy supplied except for some losses and certainly will not cause the COP to go from <1 to >1.
The earlier calculated efficiencies of the used circuit of around 80% still stands IMO.

The next question will be if the by me used output probing points and procedure (measuring loads Lamp1 and Lamp2) are the correct ones for measuring output power.

Any comments are very welcome!


Itsu

Hi Itsu,

You do a great job, thank you for evaluating the circuit in such a detail.

Quote from: Itsu on 2022.08.30, 09:26:00

I think we can assume that for input power measurements we should put the current probe left of the junction with the diode (position 3) to measure the magnetization phase current, so we can collect some data to calculate this input power.   Yes?  /  No?

Yes, where you indicated the probe (position 3) with the green symbol to check the L1 coil current, it is ok. And the voltage probe measures the voltage amplitude across the coil.
 

Quote

The next question will be if the by me used output probing points and procedure (measuring loads Lamp1 and Lamp2) are the correct ones for measuring output power.

Any comments are very welcome!

Well,  I would suggest using resistors (instead of light bulbs) like you already used a 100 Ohm resistor for lamp1.  However, it would be better to seek for a power match to get the highest output, this may involve the use of higher than 100 Ohm values.  And this could also be done for the lamp2 load.  Not an easy procedure, the loads surely interact when they are changed. 
By using resistors,  the measurements reduce to checking the voltage levels across the load resistors, aiming for the highest output level across them, no need for pulsed current measurements for the loads. Knowing the resistor value and the voltage across it makes it easy to calculate load power.
I understand that using light bulbs is a visual indication and an increasing brightness also shows an increase in output but to learn about the actual power the lamp currents should also be measured, giving room for more error.

Gyula

Gyula,

thanks for your insights.

I must confess that I already use a 100 Ohm as lamp1 load, so this diagram is somewhat outdated, but this does not change anything in the measurements.

I agree that doing some output tests using all resistors instead of lamps makes it easier, as is finding the perfect match to minimize reflections and losses.

But still there is some discussion about the correct way to measure / calculate the input power into such circuit, which needs to be settled first before continuing.


Regards Itsu

On AboveUnity.com, member Jagau showed here:  https://www.aboveunity.com/thread/melnichenko-s-effect/?order=all#comment-a8c6092b-8baa-44cf-883f-af0100e188c8   
an alternative way to arrive at the input power for this kind of circuit (Flyback converter in DCM mode) by pointing to this video https://www.youtube.com/watch?v=zunuqbap4Yk

In there is explained how to calculate the input power Pin and output power Po of such a circuit, but for now we are interested in Pin only.

My interpretation of this video is that at 1 min 22 into this video it is shown that Pin = Vi x <Ii>   for which <Ii> is Iinput average, our Iave.

Vi is the power source, in the video a battery, but in our case a 36V PS which is dropped to 33V after the IGBT.

In this video is, up to 6 min 23, explained that Iave can be calculated via   D² x Vi x T / 2Lm   where:

D = duty cycle
Vi = power source
T = period time
Lm = inductance coil

Using the data from my above measurements, we get:

Vi = 33V
D = 0.0937
T = 0.000667
Lm = 0.000945

Solving this into Iave we get:

Iave = 0.0937² x 33 x 0.000667 / 2 x 0.000945    =   0.102A

So Pin = 33 x 0.102 = 3.37W, which is close to my earlier calculated input power of 3.6W.

Comments?

Regards Itsu

Hi Itsu,

Your interpretation from "the Flyback converter in DCM mode" video (you referred to above) on the V_i (input DC voltage) and on the <I_i> average input current is correct. Using the 33 V for the resulting supply voltage your scope measures across the coil and the calculated average input current of 0.102 A,  your input power is indeed 3.37 W.

So this is a correct method for calculating the input power of a flyback converter operated in discontinuous conduction mode in which the Melnichenko circuit under investigation is said to be working.   

In the post at the aboveunity forum you referred to above, one of your scopeshots is included which shows both the input current and the flyback current.  You took that scopeshot when the current probe was placed to the right side of the diode and the coil junction.  You inluded it in your Reply #95 here https://www.overunityresearch.com/index.php?topic=4312.msg100330#msg100330 (https://www.overunityresearch.com/index.php?topic=4312.msg100330#msg100330)  I mean the 2nd schematic and the 2nd scopeshot.  When we deal with input power calculation, it is not correct to refer to a scopeshot which shows the input current and the flyback current together.  Your 3rd schematic in Reply #95 shows the current probe position for the input current and 3rd scopeshot shows the actual input current which is to be considered for input power evaluation.

Thanks for your kind efforts in testing and reporting your measurement results. 

Gyula

As mentioned here:  https://www.overunityresearch.com/index.php?topic=4312.msg100114#msg100114  i had build yet another coil combination suggested by Jagau consisting of an air coil L1 joined at each side with 2 ferrite T-cored coils put in series aiding.


The data of that setup i repeat here:

L1 80 turns (5x16) 0.8 mm diameter magnet wire, L2 48 turns each (3x16) 0.8 mm diameter magnet wire.

L1 (air) measures 116uH with both T-cores inserted in the L2's and resp. 108uH and 104uH when removing the L2 cores.
L1 measures resp. 107 and 100uH when shorting L2's.

L2's measure 176uH each and 381uH when in series aiding and 322uH when in series bucking.

So we have a low coupling factor K of resp. 0.28 and 0.37.

I use the below diagram, so we have still the 36V input, and the FG is still set to 1500Hz @ 10% Duty cycle:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45460;image)



I use for Lamp1 load a 220V / 25W incandescent bulb and for Lamp2 load the 12V / 5W incandescent bulb, see picture:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45462;image)

Both lamps are fairly brightly on.

I made my usual input power measurements right after the PS and for the output power measurement i measured the power in both lamps and added them together.

Input power shows 36V @ 790mA on the PS while a separate DMM in A mode shows 790mA also.
The screenshot of the scope input power measurements show 27.6W calculated via the math function (instantaneous voltage x current averaged):

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45464;image)

The power through Lamp1 load is measured by a Fluke 179 DMM across it (smoothed by the big C2 cap) for the voltage measuring 189.6V and the current probe in the lamp1 lead measuring 114mA average
calculating to be 21.6W, see screenshot 2:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45466;image)


The power through Lamp2 load is measured via the scope and calculated to be 2.25W, see screenshot 3:

(https://www.overunityresearch.com/index.php?action=dlattach;topic=4312.0;attach=45468;image)


So the lamp loads consume 21.6 + 2.25 = 23.8W while the input was 27.6W, so we have an efficiency of 86%.

Let's see if we can tweak this somewhat.

Itsu

For the ones of us that still are in denial of the truth, here I calculate the input power as advised by Jagau here:

https://www.aboveunity.com/thread/melnichenko-s-effect/?order=all#comment-a8c6092b-8baa-44cf-883f-af0100e188c8

Pin = Vi x <Ii>  where Vi is the supply voltage and <Ii> is the average input current being calculated by D² x Vi x T / 2Lm   where:

D = duty cycle
Vi = power source
T = period time
Lm = inductance coil


Using again my data, I get:

Vi = 33V
D = 0.0937
T = 0.000667
Lm = 0.000116

Solving this into Iave we get:

Iave = 0.0937² x 33 x 0.000667 / 2 x 0.000116    =  0.833A

So Pin = 33 x 0.833 = 27.5W, which is close to my earlier calculated input power of 27.6W.

Itsu

Hi Itsu,

You have made very good tests and posts.  And you use the formula accepted by Jagau and by Chris too and your calculations confirm your recent measurements on the new L1 - L2 coil set.

Now the funny situation is that Chris wrote in his post here  https://aboveunity.com/thread/melnichenko-s-effect/?order=all#comment-69b1e1ae-2ee7-4b27-a02f-af050155c7f5 (https://aboveunity.com/thread/melnichenko-s-effect/?order=all#comment-69b1e1ae-2ee7-4b27-a02f-af050155c7f5) the following:

"Itsu is basing his experiment around the PBE, or Power Balance Equation: P_In = P_Out  which does not hold when Power is Generated!  Excess Charge is separated and Accelerated!"

Chris "forgets" that in the video  https://www.youtube.com/watch?v=zunuqbap4Yk (https://www.youtube.com/watch?v=zunuqbap4Yk)  (from which Jagau took the formula you have also used) the first step to arrive at the formula started with
P_In = P_Out  i.e. with power balance.   Go figure.    C.C
 
And of course, Chris criticizes your measurements too in his above referred post and this is not the first case. When you started this thread in July, and you showed your scopeshots and detailed measurement results in Reply #6 here:  https://www.overunityresearch.com/index.php?topic=4312.msg99546#msg99546 (https://www.overunityresearch.com/index.php?topic=4312.msg99546#msg99546) ,  then Chris made this post here:

https://aboveunity.com/thread/melnichenko-s-effect/?order=all#comment-97eb4b3d-522d-4551-9bd5-aece00395f36  you surely remember. 

He objected your scope power measurement result (its Math function to multiply two channels) because the two scope channels were set to show RMS values for the voltage and current. So he objected this in spite of the fact that the Math channel was set to show the Mean power value of the two multiplied waveforms, that is what always counts.   And you had to prove with different channel settings https://www.overunityresearch.com/index.php?topic=4312.msg99598#msg99598 (https://www.overunityresearch.com/index.php?topic=4312.msg99598#msg99598) that the Math function considers only the value of the power which is set in the Math function menu, and it was set to Mean by you in all measurements here.

And Chris also criticized the shape of the waveforms your scopeshots showed on his Melnichenko circuit.  He referred to Jagau's scopeshot Jagau demonstrated back in February here:
https://aboveunity.com/thread/melnichenko-s-effect/?order=all#comment-5e3dbc41-9773-48e0-88a0-ae3f011c4617 (https://aboveunity.com/thread/melnichenko-s-effect/?order=all#comment-5e3dbc41-9773-48e0-88a0-ae3f011c4617) 

Jagau clearly described that in his setup back then he used a two transistor oscillator run from 3 VDC to evaluate initial circuit behaviour and his scopeshots were taken at the base of oscillator transistor Q2 etc, etc. His scopeshots were ok for an oscillator circuit running at around 55.7 kHz his coil inductance and stray capacitances dictated.
The waveforms of an oscillator circuit can differ very much from that of a switched coil circuit so indeed Chris compared apples and oranges (while he wrote this same to you).  Go figure.   C.C   :D 

Anyway, thanks for sharing your results.

Gyula

Itsu, this is what can be interpreted as being disrespectful by some but the many here thank you for asking the question and doing the work.

Un fortunately there is a lot of this in this quest but ignoring facts and not clarifying, other than to say"he has not seen the light " is about as unhelpful as one can be .

This begs the question WHY does this phenomenon occur so much as it is actually standing squarely in the way of good sharing and research.

Are we all dreamers ?  I dont think so . 

Edit I just read the thread again and if the negative current is in fact confusing our attention wrongly then let us be educated ...by a self runner .

And here is the real kicker ....I hope that I am completely wrong and the video showing the Jagu self runner is not just a youtube hit grabber .

Gyula,

thanks for "putting the finger on the sore spots".

It's well known that snippets of my posts (and others) are being pulled to highlight what can be used there, and other snipped or complete posts are carefully being avoided.
It's going on for light years already.

I try not to pay much attention to what is written there, only what Jagau and the other replicators there are saying.

Instead of the endless repetition of the mantra's about negative power and certain triangle shaped signals backed up by so-called enlightening video's and lectures I rather would like to see (and hope for) "homemade" circuits with clear diagrams showing the used components and probe points and explained screenshots of the signals.

Then we can compare notes and see where i or others go wrong if not reaching the cop > 1 goal.



3D,

thanks for your words, yes, I hope for some more helpful responses too, as what now is being replied is an endless repetition of useless words with which I cannot do anything practical.

Showing a self runner would be the optimal "convincer", it would turn me into a believer but only IF I am able to replicate that in good cooperation with the builder.

Itsu

@Gyuala @Itsu

I haven't followed this thread much but what you say, Gyula, about Chris' critique of Itsu's work caught my attention. I think you are 100% right, Chris is wrong.

Chris also says "RMS is Specifically designed for AC, Alternating Current Only! It can NOT be used for DC, Direct Current, at all.". This is ridiculous, an RMS value applies to any form of current, including DC. This guy, besides being vindictive, is not competent to the level of the lessons he claims to give and the people he criticizes, which raises questions about his own measurements.

It saddens me to see your work and your person, Itsu, the object of unfounded criticism. I have always seen you take great care with your measurements, and with method. I can only encourage you to continue ignoring Chris' useless opinion.

F6FLT, thanks a lot, your words are much appreciated.

I will continue to concentrate on what the other replicators of the Melnichenko effect on AU.com have to say.

Regards Itsu

Itsu - I see you are using a single strand wire for your coils. Have you tried using Litz wire for the L2 coil? I'm assuming that L2 is the coil you are using to harvest the collapsing magnetic field.

These words purported to be from Steven Marks resonated with me:

QuoteLet us say that you have a magnetic field perhaps it is only a small permanent magnet. Now, you have a single copper wire twelve inches long. If you move the magnet across the surface of the wire from left to right at a certain speed you create an electron flow which is DC and it has a power potential based on how strong the field is and how fast the magnet if moved. So, if you increase the size of the magnet or the speed it moves you create a larger flow of electrons, larger as in higher voltage or more currant.

Everyone tells us that the earth's magnetic field is measured as being too insignificant to generate any useable power, that is not so.

Let me give you something to think about...  If you had a short wire and you moved a magnet across it you would always have limited potential because the length of wire was so short. OK now what if we increase the length of the wire to many miles in length even with a very weak magnetic field moving across the wire you still have a much greater potential flow of power available.  If we put it into a perspective of power per inch it may be easier to understand.

If you have a small magnetic field moving across a wire twelve inches long it can generate an electron flow equal to lets say one millivolt per inch. If you move the magnet twelve inches at the same speed you get 12 millivolts as you transgress the twelve inches of wire. Understand that I am trying to convey a principal that you can understand for use in the future.

So, you have a wire twelve inches long and you can make 12 millivolts moving a magnet across it. If you have a wire 1000 feet long and you move the same small magnetic field across the length of it you can create much more voltage potential perhaps 12,000 millivolts lets say.

So, you have managed to generate a significant amount of electric power with a weak magnetic force.

OK, how does this help us? where am I going with this?

Suppose you have 1,000 pieces of wire twelve inches long and you run the same weak magnetic field over them all at the same time..... you get the same flow of electrons.

If the wires are run in series then you will get the 12,000 millivolts etc. If you connect the wires in parallel you will get higher current but lower voltage.  However, the power potential is the same whether you run the wires in series or parallel.

Have you also thought about using a more permeable magnetic core to concentrate the magnetic field in L2?

Material	Relative permeability
Air	1
Ferrite	10 to 2,300
Iron	5,000
Metglas	1,000,000

Hi lfarrand,

yes, if you look back into this thread, you will see that i used magnet wire (thick and thin), stranded wire but also litz wire.
I also used several cores for L2, like several ferrites, but also iron.

None of them showed any abnormal behavior and efficiency was around 0.8 to 0.9.

Regards Itsu 

Itsu and others, have you come across the tuning fork idea where you have a core toroid but instead of the magnetic
flux perpetually running around in one direction it has a slit tuned to a resonant frequency that's in tune with sound
vibration like a tuning fork and then place your winding around it and tune it to the same frequency or harmonic.

Try that idea you might get some where.

Sil