Kapanadze Cousin - DALLY FREE ENERGY

[Alfeen]

I received some parts today, and made a quick breadboard setup for this main generator too (4.6Khz) using the TL494.
Video to be seen here: http://www.youtube.com/watch?v=35OZREadOv0&feature=youtu.be
Regards Itsu

Hi Itsu ... and all,
I'm new to this forum, so let me briefly introduce myself. I'm a German engineer dealing mainly with electronics (including coils) during my job. Itsu, you actually recruited me for this forum (though unknowingly) because of the most commendable approach taken in your videos and the accurateness of your comments. Very good -- to say the least!

You had a question regarding the coax cable. Yes, BELDEN RG58A/U is really black PVC. But I could find the source from pictures 91684510.jpg through 91684518.jpg as posted by Edward_Lee. It was hard to read, but after a while I understood he really used *RG58U 50 Ohms (AUCAS Guangzhou Hscom Network Technology Co., Ltd.).jpg*, a white cable produced in China.

I'm still painstakingly working through your videos. A problem you had was explained in *Dally replication vco nano pulser*

I think Edward_Lee got the idea for his schematic(s) from *3985790_generator.pdf* page 3, an old (2003) document from a Siberian institute in the city of Tomsk written by A. Karaush, R.Potemin, S.Luyanov and O.Tolbanov. There you find the 7400 NAND gates to generate a rectangle and much of the rest of this stuff. You ended up with 220pF on Pin5 and a pulse width of about 100ns. To my understanding the desired overunity effect comes stronger when rise and fall times, as well as pulse duration itself goes down. But I may be wrong. At least the schematic says, rise time 1ns and peak 1kV. For the moment we are still far from that.

You said you had no success driving the mosfet with the two transistors. What type did you try? The schematic says KT972, which are Russian Darlington NPNs capable of 100MHz 60V 5A and having a current amplification of 750. A possible western replacement would be *BDW42G* which has power but can only do 4MHz, or *NTE48 50V 1A 1GHz*. State of the art, of course, are special mosfet drivers as you yourself already decided to use, like the MAX4420. Because we want rise times as low as possible it might help to try ixdd415 as a driver (3ns 15A) and IXFT30N50Q3 as the mosfet which can slew 50kV per µs @500V and 90 amps. But let us spare that for the moment to tweak efficiency in later experiments.

BTW, the three diodes later on in the schematic are specified in *3985790_generator.pdf* as VD1-VD3 - КД 522A (similar to our common 1N4148), having 50V 0,1A but a frequency of max 100kHz, which is a poor choice (or more probably intentionally misleading). These should be of the DSRD type, Russian ДДРВ.

I dare to question that your SCS106AG will lead to the desired sharp edges, because schottky is different from *Step Recovery*. Ask any anonymous alcoholic to get confirmation.

Ooops, little joke ;-) But, probably with some truth in it, because Step-Recovery Diodes are in a way much like alcoholics.

When their cathodes are connected to the negative voltage and a current flows, then they show almost no resistance and get drunk with electrons, so to speak. Thereby they become so saturated and thus oblivious to their environment (the voltage now slowly changing polarity)  that they keep drinking = the current flowing = conductive, even when after a while the voltage is fully reversed. Then the pub owner calls the police to get their bottle (the so-called intrinsic layer) emptied ... and all of a sudden they realize that their booze is all gone and switch off within picoseconds, faster than you can look. This gives us the desired sharp pulse. Normal scottish whisky drinkers -- excuse me, schottky diodes-- simply are not wired to behave like that. They instantaneously switch off as soon as their boss (or wife) appears at the pub's entrance door, meaning, as soon as their environmental tension (voltage) is reversed.

Finally, I try to understand what this coax cable is all about. I came accross an old publication from 1970 issued by one Genadin Mesyatz in Moscow. Chapters 1.1 through 1.4 explain the use of coax cables for creating high voltage nansosecond pulses.  Please find my rough translation of the preface below.

-----------------------------------
Preface
Interest in  generators of nanosecond impulses of high voltages quickly increases. In recent times they were used in quantum radio physics, in nuclear physics, in particle accelerators, X-ray analysis, in equipment for the high-speed photography etc. Extreme demand exists for the possibility of use of nanosecond generators for the decision such tasks in experimental physics, as creation of the powerful pulse lasers particle accelerators  and fast heating of plasmas.

Using high voltage nanosecond generators promises they are capable to provide huge amounts of energy (100J up to Megajoule) during short intervals in the range of 10 to 100ns and, thus, are power sources of enormous density. Already nowadays [published in 1970] impulse powers up to 10 Terawatt have become a reality.

Unfortunately, in the technical literature devoted to powerful pulse equipment, questions of formation of nanosecond impulses of a high voltage are poorly handled. This circumstance suggested the idea to the authors to impart their experience with the reader who is located in this area in the Soviet Union and abroad.

Considering the small volume of the book, we aspired to avoid bulky calculations, most to facilitate understanding of various ways of formation of nanosecond impulses and to define a technique to which the engineer should follow at a choice of the scheme and calculation of its components.

The first and second chapters are devoted to the main questions of formation of nanosecond impulses of a high voltage; the third and fourth chapters contain original helpful material in the theoretical basics, which is necessary for calculation and design of components for nanosecond generators. The final chapter is devoted to application of nanosecond impulses of high voltage. Authors gave main attention to those prospects who deal with application of powerful high voltage nanosecond impulses while quickly exploring the avenues of modern physics. ....


Introduction
The nanosecond pulse equipment is divided into equipment for medium impulses of a few hundred volts and below and to equipment for creation of high-voltage and powerful impulses of in the range of 10 000 — 10 million volts. The low-power nanosecond pulse equipment is helpful for experimental nuclear physics, the radio technician and computing equipment. The basic distinction between these two sections of pulse equipment is ruled by the character of used active elements. If in the first case such elements are tunnel diodes, fluorescent bulbs , low-voltage high-speed thyratrons. In the second case various types of spark gaps, ferrites, high peak power hydrogen thyratrons, lines with electromagnetic shock waves etc.

Methods of generating low-power nanosecond impulses are quiet often mentioned in periodicals and text books. But how these powerful nanosecond impulses can be achieved is poorly covered in technical literature. As an excuse we hear that, on one hand, prior to the beginning of the 1950's there wasn't great interest in such impulses, and on the other hand — great technical difficulties were encountered. The range of harmonics of nanosecond impulses extends up to the microwave band. Generation and transfer of such impulses therefore demands ultra wide band equipment. Moreover, how could transmissions of  such frequencies be obtained and maintained without undesired sparking, coronae and high voltage blackout. One parameter of strong influence is rendered by parasitic capacity and inductance of wires and other construction parts of the generator, which grow in size proportional to the size of the whole installation itself. Therefore requirements of shortening of the rise time and duration of impulses (especially when rectangular) became inevitable. But higher amplitudes need higher slew rates of --both-- voltage and current, while bigger physical dimensions lead to the growth of parasitic capacity and inductivity, slowing the whole thing down again. A vicious cycle, so to speak. *These contradictions can be resolved by application of coaxial designs*, solved with wisely tuned parameters and with the help of fast switching devices between the highly charged electrodes which shall have high electric durability.....
------------------------
So much for now.

[itsu]

Hi Alfeen,

welcome aboard and thanks for the info on the RG58U (white afterall he) and the translated preface of that article.

But we came a long way since your mentioned video, so i guess you still have some reading up to do.

Regards Itsu

[verpies]

Welcome to the forum Alfeen,

Do you have any experiences with MOSFET oscillations and parasitic inductances, that are beyond this paper ?

[mihai.isteniuc]

Welcome Alfeen,

I'm happy to see you here.

Mihai

[Alfeen]

Welcome to the forum Alfeen,

Do you have any experiences with MOSFET oscillations and parasitic inductances, that are beyond this paper ?

Hi, verpies

Yes, I was recently doing a tesla coil @ ~4MHz, primary driven with a  mosfet against an adjustable 30V to 70V DC supply. When the mosfet gate was conducted at the wrong moment I had awful non-linear ringing instead of the desired smooth resonance. See pictures attached.

9 turns of 0.8mm wire wound over a 3mm cylindrical former will pick up every available magnetic noise and is likely to fail as a countermeasure for the 90MHz ringing of itsu. A shielded coil like "COILTRONICS - DR1030-151-R" might probably help for small mosfets with low input capacitance. But it can take only 680mA. When switching the gate on/off a high current is flowing to/from the gate which may saturate the ferrite, change the resonant frequency --> rendering the intended filter effect useless.

The solution in my case was --> to provide proper timing at the gate.

Regarding itsu's schematic: When the mosfet Q5 becomes conducting, a current starts to flow into coil W1. Proportional to its rising speed a voltage is induced in W2, which in turn causes a charging current through the two diodes (in series) so that the capacitor C34 1nF/1500V becomes fully charged. Because the diodes are both conducting there should be no more than 2.4V across the coax. Additionally the coax is shorted at its other end. So almost no voltage should be observed across the coax at low frequencies.

I maybe mistaken, but Diode D7 does not make much sense to me, neither as a DSR nor as a normal diode. It would just consume energy. If it is there with the intention of protecting the mosfet Q5 from overvoltage or negative voltage, it should rather be a schottky diode, Cathode to W1, anode to GND. Some mosfet types have one such diode already integrated. Pleas correct me if I'm wrong.

Ideally, the mosfet switch off happens at the same time when the current through W1 reaches equilibrium. In this moment we observe maximum flux in the transformer and maximum saturation of the core. Windings, core and the 1nF should be tuned to meet this requirement. To feed maximum energy from the transformer into the coax its impedance should match the 50 ohms of the coax. Given, a C34 = 1nF is used, then the secondary W2 should have 2.5 uH (measured with an almost saturated core) and the resulting resonant frequency would be 3.138MHz. Saturation can easily achieved with an adjustable DC fed into ten temporary additional windings on the core. The static current creates static magnetic flux. Any overlaid modulation from an inductance meter will show that at high current the inductivity goes down. When inductivity does not increase any more although you still increase the current then you have saturation. However this frequency 3.138MHz is not interesting here, as it will never go into the coax by design. It could be 1MHz or 20MHz as well. Only the impedance match counts.

If the mosfet is kept switched on after the current in W1 stopped rising, it heats up itself and the coil, wastes energy and creates all kinds of undesirable ringing -- coupling even back to the gate and to the 74HCT00 U2 & U3. So, my proposal is to start with a Dirac-like pulse at the output of the 74HCT00, say C17 = 3.3pF at Pin5. Then slowly increase this capacitor. Pulse repetition rate could be slowed down by replacing C14 = C15 = 1nF with bigger values, thus giving the mosfet more time to calm down in between the pulses. These capacitors can later be restored to 1nF (or even less) after the ringing issue is solved.

In an early video (I did not yet have the time to see all), itsu connected the mosfet with colored cables having clamps on each end. I guess this is not the case any more. But still, connecting wires between U2 & U3, driver and mosfet should be as short and thick as possible, especially the GND connection. Litz wire is preferred and a large copper area on the PCB for GND. Every millimeter counts at these frequencies.

After the ringing problem is solved let us look into efficiency. To increase the strength (energy content) of the back EMF pulse in the secondary --> the core should probably have a gap of 0,5mm up to 5mm. This is because considerable magnetic energy can only be stored in an air gap, not in the ferrite core itself. A ferrite is like a short circuit to the magnetic flux. With a closed core you have much magnetic "current" (measured in Tesla) but almost no magnetic voltage (measured in A/m). The resulting power is the product of both. Integrated over time we get the amount of energy. Something multiplied by almost zero remains almost zero. The air gap acts like a resistor in Mr. Ohm's circuit. Current goes down, but voltage rises. On the other extreme current becomes zero and voltage maximum. But again power is zero. Highest power (and thus optimum energy transmission) occurs, when impedance of voltage source and resistor match. A Dremel Rotary Tool (diamond blade and LOW speed works best) can create such gap in a ferrite (under rinsing water, use goggles to protect your eyes!). We will have to find out by trial and error which width works best.

When the mosfet is switched off, the field returns to the coil (forming a back EMF pulse) and current starts to discharge C34 1nF. If  1N5408 diodes are used, I expect them, to block the current flow within nanoseconds of the voltage reversal on the secondary output. If it has the alleged step recovery effect, the current will continue to flow for a while until all electrons stored between the p-layer and n-layer are exhausted. As there is no special SDR intrinsic zone in a 1N5408 this won't take too long. So the negative voltage will rise almost immediately, but not too quickly.

If instead a real SDR is used, after the voltage reversal (start of the back EMF)  there should still be no voltage across the coax cable. Only when the SDR switches off and the current flow is sharply interrupted then the desired pulse with a rise time of picoseconds should appear on the coax, causing standing waves therein. Those will decay over time. Stranded copper or solid copper does not make much of a difference. The standing waves  last longer if the cable used has smaller losses at high frequencies. To increase the reported 800V to over 1kV steal a winding from the primary or add one (or more) windings to the secondary to get a higher transformation factor. Instead of magnet wire copper band (167-9332 rs-online.com) and thick layers of  insulation (Kapton  436-2784 or  436-2778 @ rs-online.com) may be used to increase surface and reduce high frequency losses in the secondary W2. Also, for C34 a low inductance high impulse resistant type like WIMA FKP 1 is preferred.

kind regards