M.E.G replication attemp

[patmac]

I've been trying to replicate it based on JlNaudin MEG 2.0 by using a grain oriented ferrite core.

Results are strange, by using a varistor seems OU but I don't know exactly why.

Test:

By using 24volts X 0.4amps = 9.6 watts

Output rectified with 1n4007 diodes then a 47ohms resistor to measure current and Varistor.

Varister perfectly reads 360 volts and 47 ohms reads 1.5 volts then:

1.5volts / 47 ohm = 0.031 amps

360 X 0.031amps = 11,16 watts.

My question is: Varistor is buring 11,16 watts? or simply is a negative resitor.

Here the video showing my MEG taken with my webcam
I can support my results with scope shots sooner.

YouTube - xc.avi

I've my ears open.

REgards

[gyula]

Hi,

You have not mentioned what measuring instrument you used for getting 1.5V across the 47 Ohm? Unfortunately when you have a full wave diode bridge rectifier without a puffer capacitor, then a digital voltmeter can fool you (because of the positive half waves). It is not easy to perform a meaningful measurement without a true rms voltmeter when you have the full wave rectified output pulses without any filtering like a puffer cap does.

Have you found the varistor body as warming at least a little? A 11W dissipation in it should make it pretty warm, right? (switch the setup off when you touch the varistor!)

What is the switching frequency approximately? because the 1N4007 diodes are low frequency types, they are ok for now but later if you really have more out than in, you may wish to use the fast recovery version like UF4007.

I never understood why Naudin used a nonlinear load? I fully disagree with it and it makes the output measurements rather foggy and uncertain... it is against any logic. I guess this was the gateway to come out from the situation... Years ago I built a MEG with ferrite cores, taken from line output transformers used in CRT television sets but got a decent 87% efficiency... a COP of 0.87. I used normal resistor loads.

rgds, Gyula

[patmac]

@Gyula

Exactly, varistor get warm on about 40* C, the output was measured with digital multmeter and scopes waveform seems confirm this output.

Test #2 taken today for answer you:

Power input drawing measured from Power Supply analog

24v x 0.35amps = 8.4 watts

I'm avoiding use DMM so please look scope shots

first scope shows voltage on output with varistor connected using 1n4007 bridge diodes and 47 ohm resistor in series with varistor.
Oscilloscope tuned:
Time div / .1 ms
Volt div / 100 volts


second scope shows cur drawing taken from 47 ohm resistor
Time div / .1ms
Volt div / 5 volts

Varistor is really pretty warm I can feel pain by putting my fingers in 5 seconds. I would try to measure temperature with DMM of my brother maybe tomorrow.

[gyula]

Hi,

No much need now for temperature measure.
I assume the scope zero line is in the middle horizontal line in the second picture and somehow I suspect those spikes has less than 1.5V effective value.
If you could span out the time base so that time division 1us or 10us then the area under the spike borderlines could be estimated much better, the area could be 'arranged' to be an equivalent rectangular with horizontal coverline instead of the spike shape. This ought to be made with the voltage spikes in the first shot too, unfortunately... by the way where is the zero line in it?

Gyula

[patmac]

I traced the 0 volts line

Hi,

No much need now for temperature measure.
I assume the scope zero line is in the middle horizontal line in the second picture and somehow I suspect those spikes has less than 1.5V effective value.
If you could span out the time base so that time division 1us or 10us then the area under the spike borderlines could be estimated much better, the area could be 'arranged' to be an equivalent rectangular with horizontal coverline instead of the spike shape. This ought to be made with the voltage spikes in the first shot too, unfortunately... by the way where is the zero line in it?

Gyula

I not sure, but look 5 volts div, then the peak then reads about 15 volts peak so look bottom of the wave form, easily reads 25% of time, so if we take only 5 volts easily we have: 1,25 volts. And we're discarting a good part of this waveform. So I've noticed incredibly a cheap DMM seems to be more precise than many people claims.

Look Naudins use a nonlinear component for avoid sceptiscism, a good non linear component is ideal to get a perfect sine wave on output, so his digital tektronk equipment can reads precisely a RMS form and calculate the output.

I think ohm law is tricky by using negative energy. I don't know. Tomorrow I can retune the circuit and take the scope shot that you suggested to me on about 1us-10us.

Please I'm not forcing a OU results here this not justify to say this device is OU only is my appreciation about the waveform.
Rgds

[gyula]

Hi Patmac,

Ok on the spikes amplitude, I did not consider too much the upper 5V peak, (from 10 to 15V, the very peak part) being very narrow in time, so the real and effective voltage amplitude can indeed be somewhere between 1 to 1.5V.

On Naudin: I respect your opinion but I have a different one...

On Naudin's output sinewave: he did a frequency sweep with the TL494 pulse oscillator to find a kind of parallel resonance in the secondary coils of the transformer, on the 1.3 Henry secondary he searched for resonance with the coils self capacitance and found it at 34kHz.
( http://jnaudin.free.fr/images/meg21dg.gif 'P1' varies frequency)

SO it is NOT the nonlinear component (varistor in your case or his 100kOhm "conditioned" resistor) that helps make sinewave at the output but finding LC resonance (a 1.29H coil surely has got about 17pF self capacitance to resonate at 34kHz).

Gyula

[patmac]

Hi Patmac,

Ok on the spikes amplitude, I did not consider too much the upper 5V peak, (from 10 to 15V, the very peak part) being very narrow in time, so the real and effective voltage amplitude can indeed be somewhere between 1 to 1.5V.

On Naudin: I respect your opinion but I have a different one...

On Naudin's output sinewave: he did a frequency sweep with the TL494 pulse oscillator to find a kind of parallel resonance in the secondary coils of the transformer, on the 1.3 Henry secondary he searched for resonance with the coils self capacitance and found it at 34kHz.
( http://jnaudin.free.fr/images/meg21dg.gif 'P1' varies frequency)

SO it is NOT the nonlinear component (varistor in your case or his 100kOhm "conditioned" resistor) that helps make sinewave at the output but finding LC resonance (a 1.29H coil surely has got about 17pF self capacitance to resonate at 34kHz).

Gyula

It's possible notice diode bridge is not working correctly if we look negative part, I've tried varying the frequency with varistor is possible get a sine wave but without diode bridge....

Naudin talks about the danger of misreading results on Tektronik equipment when waveform is not sine perfect.

[gyula]

Of course there are no any diodes at the output when you look for resonance, use a high value resistive divider (say 1MegaOhm - 10kOhm) directly across the output coil and probe with the scope across the 10kOhm resistor to see how the waveform and its amplitude is changing when you change the pulse gen frequency with P1.
When you find a sinewave-like waveform, the output amplitude goes up and in case your varistor is connected it may clamp the peaks, making the waveform distorted again, so first do not use any varistor to find sinewave-like output...

AC Measuring equipments are calibrated for sinewaves, that is ok.



Gyula

[patmac]

Of course there are no any diodes at the output when you look for resonance, use a high value resistive divider (say 1MegaOhm - 10kOhm) directly across the output coil and probe with the scope across the 10kOhm resistor to see how the waveform and its amplitude is changing when you change the pulse gen frequency with P1.
When you find a sinewave-like waveform, the output amplitude goes up and in case your varistor is connected it may clamp the peaks, making the waveform distorted again, so first do not use any varistor to find sinewave-like output...

AC Measuring equipments are calibrated for sinewaves, that is ok.



Gyula

Yes I agree 100%. Tonight I'll upload the test without rectifier to get sinewave and don't worry my shots for measure voltage are using 1mohm+10kohm resistors . This is far better to protect the equipment too.

[patmac]

Of course there are no any diodes at the output when you look for resonance, use a high value resistive divider (say 1MegaOhm - 10kOhm) directly across the output coil and probe with the scope across the 10kOhm resistor to see how the waveform and its amplitude is changing when you change the pulse gen frequency with P1.
When you find a sinewave-like waveform, the output amplitude goes up and in case your varistor is connected it may clamp the peaks, making the waveform distorted again, so first do not use any varistor to find sinewave-like output...

AC Measuring equipments are calibrated for sinewaves, that is ok.



Gyula

@Gyula

I've experimented big problems with my internet connection, rains is destroying all here in my country. But I'm here again

First shot Voltage shows some like sinewave 5 div /100

Second shot Current shows some strange effect 5 div.


The test was without diodes only coil output with resistors voltage divider for take the voltage and 47 ohm resistor to measure current

Both shots was taken on .2 ms Time /div

[gyula]

Hi Patmac,

Thanks for the scopeshots. Would you explain where you placed the 47 Ohm for current measuring, into the primary or the secondary coil? The load was only the resistor divider across the secondary, 1MOhm + 10kOhm, correct?

If you could span out the time divisions from .2ms/DIV to 10 or 20usec so that say one or two full current wave period is shown on the display, it would be better. Also indicate zero line or state if it is always in the middle horizontally.

The current waveforms indeed look strange, I assume you can influence the shape by tuning/changing the input pulse frequency AND changing the amount of input current by reducing input 24V DC. Normally for a transformer secondary coil with having roughly sinusoidal voltage waveshape the current shape is also roughly sinusoidal with the resistive loading unless the transformer core manifests already nonlinearity.
Just thinking: the permanent magnet in the middle biases the core towards one direction, ok? it means that the operating point of the core got shifted from the zero middle point (unbiased case) to either upwards or downwards alongside the B/H curve, depending on the magnets inserted polarity direction. Now, when the input pulse appears, there can be two directions the operating point can move further alongside the curve from that of the point the magnets shifted it: further up or further down. In case the flux of the input current helps the magnet flux, i.e. adds to it then the core gets even closer to saturation and in case the flux from input current works against the magnet flux, the operating point gets move to the opposite direction from the point the magnet statically shifted it, in this latter case the control range can be much wider for the core but you have to sacrifice some input power to work against the permanent magnet flux...


rgds, Gyula

[patmac]

Hi Patmac,

Thanks for the scopeshots. Would you explain where you placed the 47 Ohm for current measuring, into the primary or the secondary coil? The load was only the resistor divider across the secondary, 1MOhm + 10kOhm, correct?

If you could span out the time divisions from .2ms/DIV to 10 or 20usec so that say one or two full current wave period is shown on the display, it would be better. Also indicate zero line or state if it is always in the middle horizontally.

The current waveforms indeed look strange, I assume you can influence the shape by tuning/changing the input pulse frequency AND changing the amount of input current by reducing input 24V DC. Normally for a transformer secondary coil with having roughly sinusoidal voltage waveshape the current shape is also roughly sinusoidal with the resistive loading unless the transformer core manifests already nonlinearity.
Just thinking: the permanent magnet in the middle biases the core towards one direction, ok? it means that the operating point of the core got shifted from the zero middle point (unbiased case) to either upwards or downwards alongside the B/H curve, depending on the magnets inserted polarity direction. Now, when the input pulse appears, there can be two directions the operating point can move further alongside the curve from that of the point the magnets shifted it: further up or further down. In case the flux of the input current helps the magnet flux, i.e. adds to it then the core gets even closer to saturation and in case the flux from input current works against the magnet flux, the operating point gets move to the opposite direction from the point the magnet statically shifted it, in this latter case the control range can be much wider for the core but you have to sacrifice some input power to work against the permanent magnet flux...


rgds, Gyula

Explication of Test Setup:

This very last shots was taken with Input 21V @ 0.3A, reading the analog power supply look the attach.

The very first shots was taken by rectifying the output, but to avoid confussions I've removed the diodes bridge two last test.

All test setups the output current was measured with 47ohm resistor on Secondary Coil in series with the LOAD (Varistor).

With or Without diodes Varistor gets really warm impossible resist the heat in the fingers.

The logic circuit is a TL494 reducing the square wave a little (35-40%)this produces me better results. This TL494 is connected to my lead acid batt @12 Volts fall apart from input source to avoid frequency change when I change the input voltage.

Really I think Naudins use a Carbon conditioned resistor or MOV because this produces a near perfect Sine Wave, in my case is near, maybe I need test with many MOV types and specifications. Sine wave is usefull to use another inductor to reconvert the output energy and try self loop for example. But is really usefull to precisely calculate the COP; Naudins talk about this:

THS720P measurement error*with non sinusoidal waves

If we look his output shots we can confirm the near perfect Sine wave. But output is really good too when this wave out of shape. I can change the wave by changing the frequency or the Voltage both methods can produces results identical.

Regards...

[gyula]

...
Really I think Naudins use a Carbon conditioned resistor or MOV because this produces a near perfect Sine Wave, in my case is near, maybe I need test with many MOV types and specifications.

Hi Patmac

I would like to understand why you think that using a Carbon conditioned resistor or a MOV produces near perfect sinewave at the output in this setup?

Sine wave is usefull to use another inductor to reconvert the output energy and try self loop for example.

Could you explain how you connect another inductor for reconverting output energy for looping? On inductor you mean a coil (either with or without a core) or you mean another transformer?

IF you allow one more question, could you recall what was the highest frequency you went up during your testing your setup? Your present waveshapes show about 6-6.4 kHz frequency.

thanks, Gyula

[patmac]

Hi Patmac

I would like to understand why you think that using a Carbon conditioned resistor or a MOV produces near perfect sinewave at the output in this setup?




Could you explain how you connect another inductor for reconverting output energy for looping? On inductor you mean a coil (either with or without a core) or you mean another transformer?

IF you allow one more question, could you recall what was the highest frequency you went up during your testing your setup? Your present waveshapes show about 6-6.4 kHz frequency.

thanks, Gyula

Because MEG produces high voltage and high COP seems depend of impendance on the output, so you need avoid load the output when waveform voltage is relatively low. So a MOV close the circuit when voltage is high so this seems create the sine wave. A inductor can be a wall transformer. The frequency is a interesting thing but when you can change the voltage input you get greats results too, so with lower voltage you can operate with higher frequencies.

Hey using a non linear load like MOV or carbon conditioned IS NOT MY DISCOVERY AT ALL, this is proven by Naudins and others replicators and obviously Bearden. Seems a nonlinear load in this gadget seems to be a Negative Resistor.

[gyula]

Hi Patmac,

I have read a paper on varistors in general and turns out they have a self capacitance just because of the material and manufacturing process and this capacitance ranges from some hundred picoFarad to some nanoFarad, depending mainly on the working voltage they are designed to. See Figure 3, Page 4 in this PDF file http://www.eeel.nist.gov/817/pubs/sp...ounce%20BW.pdf

Here you can find data sheets on varistors if you do not have data on your own varistor: Description Varistors, Varistors Datasheet search, Varistors datasheet, Varistors datasheets, Varistors data sheets - ALLDATASHEET.COM

If you do not mind testing, I would like to suggest removing the varistor and placing a 600-800pF capacitor directly in parallel with your secondary output coil and find again resonance by varying the input frequency. This way you surely can get a sinusoidal voltage output when you find resonance. Use at least four 250V rated capacitors in series to get a 1000V rated capacitor if you do not have any high voltage type in that picoFarad range.

I believe the near sinewave shape waveform is created by the self capacitance of the varistor but the varistor behaves as a switch too and when it is ON, it shunts its self capacitance therefore resonance is ruined for those ON time instants, this can explain why the waveform is not purely sinusoidal. Your secondary coil has 1 Henry inductance as you indicated in your schematics and this can resonate at around 6.1kHz with 680pF parallel capacitor I used this calculator:
Pronine Electronics Design - LC Resonance Calculator

Of course you have to use the tuning potmeter of the TL494 oscillator to sweep through the frequency range and tune in the maximum output amplitude.

If you wish to use a wall transformer at the output, how would you like to connect it? Maybe using several wall transformer with its primaries in series to make up for the 800-1000V voltage amplitude?

rgds, Gyula