Right, but even then no sweet spot has been found..... yet.

You know that I don't think this system will work without gain medium.

However, in case I am wrong, I still have several useful objections:
1) In Dally's schematic a 220V light bulb is used at the output of the device (L4). Not a 12V bulb.
2) The nanopulse might not be narrow enough (yes it still can be improved a lot and you know it)
3) During fine tuning, a light bulb does not provide a visual feedback of the tuning response that is quick enough to notice, if the frequency response of the anomaly is very sharp.  Because of this, a scope should be used to catch anything anomalous and short happening at L4 (using one-shot realtime sampling triggered by some anomalously high trigger level).
4) In Dally's schematic a FWBR made out of D10-D13 is connected to the output of the device (L4) and this FWBR is feeding stiff capacitors C35-C37. This is very significant because stiff capacitors will allow a very large current to flow in L4. High current flowing in L4 changes the magnetic field distribution in the coil significantly. Such high current cannot flow with just the 12V light bulb, without those capacitors! 

I first tried with a 220V bulb, but no light at all.

Anyway, i wanted to show severall people that just putting the parts together not necessarily mean that the magic starts.
More work needs to be done and improving the nano-pulse (narrow it) is one of them.

Regards Itsu

@itsu
Currently, after some modifications, my nanopulser is producing 960 volts pulse. The modifications were:
1) using a MIC 4427 MOSFET driver with a single 10 ohms resistor in series with the gate of the irf840.
2) returning the toroidal transformer to 6 turns primary and 12 turns secondary. The toroid was obtained from the printed circuit of an Osram 9w CFL. I removed the plastic insulation and made the windings directly on it.
3) the capacitor in the output of the transformer is 1nF/ 2KV ceramic.
4) used two 1N5408 in parallel as the output diode, assembled directly at the input of the coaxial cable.
5) at the drain of the Irf840, I get a 140ns pulse.

My inverter is operating at 58.2 KHz, so I modified L2 to 160 turns, 18AWG and tuned it with 4 100 nF capacitors in series.
My L4 is about 59 turns flex wire, multi stranded AWG 12.
L1 was wound on a movable carton core, so I can slide it inside L2/L3/L4.

What I found out is that I can light a 60W incandescent bulb, with voltage that can vary from peak 120 to almost 160V ( sine wave, at 58.2 KHz), depending how much of L1 is inside the other 3. At 120V peak, input current is about 4.1A in the 12V power supply, which actually is 11.6v DC. Turning on the nanopulser, and sliding L1 out a little bit, I started to see some modulation on the envelope of of the sine wave, probably related with the frequency of the nanopulser, but no improvement in the output voltage or light intensity. Grounding L1 and L4 did not make any difference.
That's where I am, quite similar to you, despite difference in the frequency.

Can you try to increase the output cap till about 10nF? It works much better with higher value's with me.

Please also take note of the comments by verpies above, especially the point 4 concerning the output FWBR:
"This is very significant because stiff capacitors will allow a very large current to flow in L4."


Regards Itsu

I will try that. Capacitor up to 10n first.
PS: I added a 3.3 nF cap in parallel with the 1nF cap. The nanopulser amplitude is now 1920V, but there is an uncertainty on that, because I'm measuring it through a voltage divider. I don't have a high voltage probe, neither can afford one now, so I added in parallel with the diode one 180 ohms resistor in series with a 12 ohms. The voltage at the 12 ohm terminals will be 1/16 of the voltage on the diode, and I'm measuring 120Volts on it, so across  the diode it should be around 1920V.

@itsu
Your nanopulser seems fine and even more "perfect" than Dally had.
All you need to get its power source from same toroid(yoke in your version) so it will have relationship with current going into L1 coil.
When nanopulser fires there is loop created and what I really would love to see on scope is this:
http://www.youtube.com/watch?v=66-DV9kY2dM#t=0h13m06s
(the sinus wave is your primary current driving frequency, the spikes are BEMF. It is important to have them set into 180 degrees from current cycle)

Also there is very important moment when to take energy out of system:
http://www.youtube.com/watch?v=EVDNkF1npHs#t=0h6m25s
(when spark gap is firing in Tesla coil, you hold off from taking energy and allow free flow to charge secondary on explosive capacitor discharge. As soon as it stops the longditudal wave fires right after this discharge in EM field collapse you take energy out of all tuned coils (as many as you can place) around secondary coil.)

Sorry if it sounds greek in Youtube maybe someone can translate that fully into English.
P.S.> Some teaser: Romanov (the guy who explained all stuff) demonstrated self runner in the end of same video - http://www.youtube.com/watch?v=7eKi7ol12c4#t=0h44m0s

Cheers!

@itsu

I reviewed the measurements and the pulse amplitude is not 1920V ( it was too good to be true!). It remained the same, when I increased the capacitor. I was fooled by some ground noise.
Sorry for the bad information.

@itsu:
I increased the output capacitor, but no further improvement. I also connected the FWRB in the output and a 15uF/200V capacitor. I can't say that this modified the behavior. I'm using two secondaries in the inverter, one to power L1 and other, through a FWRB and filter capacitor, to power the MOSFET that drives the toroidal transformer. DC voltage at the 60 Watt/120V incandescent bulb is between 70 and 90V, depending on how much of L1 is inside the other three coils. As I said before, there is a particular spot where there is a low frequency modulation of the AC output, but I can't say this improves something.

You might put some ferrite (bars/rings) in the coil to see if that has any effect.

...and DC offset.

verpies:
Can you explain a little more your idea of the DC offset?
Thanks.

That is applicable only to devices containing gain media, such as mentioned by Itsu below:

You might put some ferrite (bars/rings) in the coil to see if that has any effect.

Do you still want to know about it?