Kapanadze Cousin - DALLY FREE ENERGY

[d3x0r]

See... it's all stabs in the dark... when a N magnetic field connects to a S field, does it cover the most area first... so the outer to the inner, or the inner to the outer?  if I have layers N N S S S S , is the first ... I dunno maybe if it's driven it does... just thinking that if it's induces... it's the inner and outer that participate first... but that's all beside the point.
----------
So I wound a kapagen coil on a ferrite bar... I wound several; and one works very well... and I swear it's 10x output power to input... being a power of 10 I assume I missed a scope probe or scope settings somewhere but I did double check, and triple when I filmed it...


with the ferrite bar, all grounds are basically equivalent... it oscillates at 225Khz (200-266 depending on where scope probes are)


all are bifilar twisted telephone jumper cable...
first was 16+16+8+8+4+4  or 16/8/4 for shorthand... there's 2 layers forward and back
it had moderate output at about 1.0Mhz...
I wound another that was more like 8/4/2 next to it... and it was useless...
Since I had the 2 wires I started testing on one filamnet, and got output, I could join them, and modify the frequency a little, but not a lot of improvement... if I connected them in series start-to-end, I got better output always... and so if 2 is good 4 is better right? 
anyway I turned another that was 36/18/9, this worked at 400Khz or so... and in series 200-266Khz *best output*
I turned another that was kind of narrow ... 16/8/4 so I could fit it in the C part of the core and increase inductance further... it had some output, but was kind of weak... quadfilar.. so I figured since 2 in series was better, 4 in series could recover and be the same inductance... but shorter; 4 in series is bad... like 0 output
I wound another that was 32/16/8 quadfilar... 2 filaments is about 250-300Khz (similar to best output) 4 filaments is bad... 1 filament is better than 4, and 2 is better than 1.. so the 2 pair in parallel is good...


I changed driving methods with the best output coil and was able to maintain a similar output.  1) I had a wide coil of 6-8 turns distanced from it a ways...a couple different ones maintaied the same output.  I also put a C magnet core over it and was able to put other peices of ferrite to make a complete magnetic loop  and drive a coil on the core ... the kapagen coil output was good.  (frequency is lower... but eventually I managed to separate the parts, and retune it until the c core was detached and it still worked... so it's not a input resonance that's doing it... and doubling /halving the turns on the input had no effect on the output.. the input was a signal generator +/-3V 10mA; I'm going to assert that it can supply 10mA because of a calculation later...


so it is only +/-24V (48V total) a 1uF cap charges in 250mS ... which is like 30mW....  250mS to the 48V is pretty good...
-----
Now; I put 2 10 ohm resistors in, and was able to bias one scope against the + signal and a 10ohm resistor to give a 200-400mV current sense... (20-40mA... the signal generator may be more than 10mA); the input voltage was represented as only 1V on the scope... and given that the signal should be 3V; have to multiply by another 3 for some math...




the other I put between the load and it's ground... and had 24V at the bridge rectifier
---------------


Re the attached picture... normally a resonant coil, opposing ends of the coil are different voltages... I find that this coil both ends are the same voltage... which can still go through a bridge rectifier.
When I was testing before... I had a single phase that was offset by 90 degrees...
it's kinda like the middle 'top-dead-center' gets skewed by the winding+counter-winding.... so the phase that is expressed at the end is offset in the sin wave...


https://www.youtube.com/watch?v=wWEHDVfvY08


Removing the scope probes I was able to tune it to a higher voltage output later... 

[Void]

Concerning this drawing, there are more connections that does not make sense to me like the Kacher secondary being connected
to ground at the bottom without being connected to the base of the transistor.
Would it work this way?

Hi Itsu, no the kacher will not work without a feedback connection from the kacher
secondary back to the transistor base. I noticed that as well but just thought that might be
because Ruslan was just drawing a quick sketch and didn't bother to show every detail.
All the best...

[Void]

Here is something to consider. Ruslan wrote the following in the realstrannik forum (Google translation):
"Desirable to remove the energy pulse is short (150-250 ns) This podbiraetsya when configuring. 
And it's better when the frequency of removal rate multiple of the frequency of the pump.  Ie  Kacher 1.5 MHz, 15 kHz will eat !!!"

So, if we pulse the capacitor with say an IGBT at 15 kHz, that is a period of about 66.67us (66.67 microseconds).
If we take the maximum pulse width from Ruslan's recommended pulse width above of 250ns, this
gives us a duty cycle of (250ns/66.67us) x 100 = 0.37%. That does not appear to be very easily achievable.
What do you guys think? Assuming we can achieve our 250ns driving pulse to the IGBT gate at around 15kHz,
would an IGBT fully switch properly at that short of a driving pulse width on its gate? You are going to need a very fast
IGBT in the very least. Is Ruslan just talking theoretically here but actually using a wider pulse width himself? I am
inclined to think so.

P.S. I see that there are some IGBT's that have on and off switching times that are very fast,
but only to collector-emitter breakdown voltages of about 1200V max or so. For example the
APT25GP120BDQ1G has a max C-E breakdown voltage of 1200V and a Ton of 12ns and
a Toff of 70ns. Unfortunately we would need a fair bit higher C-E breakdown voltage rating than
that since when switching around 1000V through the primary winding of the toroid transformer
the back pulses are likely going to be pretty high voltage, even if using a snubber. Maybe you 
could put two or three lower voltage rated ultra fast IGBT's in series and switch them
in sync, but at a 250ns pulse width circuit propagation delays are going to be critical to
get good switching timing sync. Did Ruslan really go to all that trouble, or am I overlooking something?

All the best...

[corry]

To understand the goal: http://www.walter-fendt.de/ph14e/stwaverefl.htm

[d3x0r]

The good ferrite coil has an inductance of 1.2mH; the other coil I have that's a toroid core with 6 turns is 1.1mH on one winding... I have another loose winding on the toroid to put the signal generator to...  so if it were just a standing wave, replacing it with the same inductance would have resulted in a similar behavior; was unable to get appreciable voltage in any near frequency.