As an (inactive) HAM radio amateur i struggled with standing waves on coax many times (to get them go away :-) )

Unlike with HAM transmitters and antennas, with this Dally device you have to maximize the VSWR by mismatching the coxial terminating impedance as much as possible - that's why I wrote: "shorted or opened".  (BTW: Tesla Coils have the same requirement).

This means that for my shorted 41 turns coax coil of 6 meters length i have to have a node every 14.6cm (6/41).
Having 2 nodes/wavelength, this means a wavelength of 29.2cm = 1GHz.   Correct?

Almost. You are forgetting that at 1GHz the full wavelength in your coax is not 29.98cm but 19.79cm because the velocity factor of your coax is most likely 66% of the speed of light.

That's the theory, but in practice it would be good if you could visualize the standing wave in the coax with a linear array of small cheap neon bulbs (e.g. NE2) working according to this principle or a capacitively coupled oscilloscope probe like that Cortland guy.

Attached is an old military document but very applicable to building stronger pulse generators out of multiple avalanching transistors.
Contemporary transistors are much faster than the ones available when this document was written.

All,

Concerning the main generator (4.6Khz) toroid, i found this diagram which contains some data on it:

http://realstrannik.ru/media/kunena/attachments/1226/dally1.GIF

meaning:

Ferrite Ring probably 4cm od

Windings 1 and 2 are  3 turns 1.5mm
Windings 3 and 4 are 70 turns 0.6mm

But looking at the original pictures like Hoppy already mentioned, this does looks more like a commercial
toroid with much more wire on it.

While waiting on my parts to arrive, i was experimenting with a toroid i had laying around, see picture
(2 turns primary / 40 turns secondary), but was blowing up my MOSFET driver chips 4420 (MOSFETS IRF630) :-)

Questions:

# these 3 (or 2 turn) primary, act like a short to the MOSFET's, right?  Should there not be much more turns?
# these primaries (1a and 1b) should be CW and CCW, to work in this push/pull configuration, right?
# the secondaries (ii and iii) can be either CW or CCW, right?

Thanks,  Regards Itsu

My expectation to the ferrite ring size is about 6cm od. BTW regarding 494 pwm IC, ' been playing on similar circuit for quite sometime... An hour ago, I tried to test with similar filter coil like what you have shown but can't get a good result with that kind of coil. I think the material used in those kind of filter coils does not much with the frequency or something else. I too have blown my mosfet while my modulator is already tested with my other known good trafos. In my personal practice, cannibalized pc psu main trafo is one of the bests for quick & easy testing if your 494 ckt is working good or not, but prior to that, waveform reading will help a lot.
   
About Dally's circuit, I'm just wondering about rectifiers used at the outputs. It appears that those bridge diodes are not fast recovery type... Can somebody verify that?       

Attached are my sample tests with my upc494 generator ... 
 

This is what happens when you pulse a coaxial cable.
http://www.youtube.com/watch?v=zrDxSM91Jcg

This video is about unterminated - this isn't exactly the same, because the center is connected to the shield.  Using my 6m length of coax with the end shorted, and sending a low frequency square pulse, and watching at the rise and the fall, the inducatance is expressed on the other coils as first a rise, then a fall as it's rising, since the reflected pulse is in the opposite direction.


 

http://www.youtube.com/watch?v=Il_eju4D_TM

But since the speed of light is approx 12"/ns, and signal propagation is 50% of that, 6"/ns ... so 39.37ns for the half reflection...... 80ns total to get back on the shield....  (the speed of propagation is just word of mouth,... and could have had something to do with the test equipment they had on hand)
as long as the pulse is short enough to fit within this...  could probably be even as wide as 10ns and work pretty good...

@Itsu

With that horizontal scope resolution you are blind to the influence of MOSFET's gate capacitance on the rise time and fall times of the TL494 outputs.
Please increase the horizontal ns/div on your scope and compare the rise/fall times with and without the MOSFETs connected.
A 1μs increase in the rise/fall time will not make much difference for 4kHz, but for 400kHz it will make a huge difference!
Just don't do anything blindly and observe...

As far as the torroid goes, you should measure its primary inductance by measuring the current through the primary winding using a small series non-inductive resistor and your scope across that resistor, while the primary winding is driven with rectangular pulses. 
Your testing setup must be able to withstand a short circuit. 
As you apply the rectangular pulses to the primary, the current sensed by the series resistor will increase with time from the beginning of each pulse.  The initial slope of this current trace will indicate the initial inductance of the winding.  The larger the inductance, the slower the increase of the current.

Soon you will be able to observe how quickly the current increases in time. For example if the current incleases at 2A/μs and your MOSFETs are rated at 6A max., then you will know, that with this ferrite and with this many windings, the increasing current will blow your MOSFETs after 3μs. 
This means that the maximum pulse width will not be able to exceed 3μs.  You can then play with your switching frequency and duty cycle limits in such a way that it never happens... or you will have to add more winding turns ...or change to a ferrite of larger permeability.

If you see that the slope of the current increases as the time goes by, when you apply the rectangular pulses to the primary, then you will know that the ferrite starts saturating magneticaly (saturation causes the scope trace to curve up usually in microseconds).
Without saturation, the slope of the current in an RL circuit should asymptotically approach the V/R limit like this: I_L(t)= (V/R)*( 1 - (e^(-t*L/R)) ) where:
e=2.718282,
V is the supply voltage,
R is resistance of the winding and MOSFETs and the current sensing resistor (or the internal resistance of your signal generator),
L is the inductance of the winding,
I_L is the current through he winding,
t is the time from the rising edge of the stimulating rectangular pulse.

It is useful to graph the above formula in Excel, etc...

As far as the direction of windings: if they are wound in the same direction, then for push-pull operation, the current should be in opposite directions in each winding as the MOSFETs switch on alternately.

@Itsu

With that horizontal scope resolution you are blind to the influence of MOSFET's gate capacitance on the rise time and fall times of the TL494 outputs.

Thanks for the info verpies.

i will start building a toroid using 7 prim. turns each first (easier to remove turns then to add) and do some testing.
Then decide to use drivers (max4420) or not.


I also received my LT1073's and mur diodes, so can start building your nanopulser.

Regards Itsu

 Sir Itsu, I think you have all info’s to build your toroid trafo but to answer & give you also few details on my pre-tested toroid, I just picked it from my old smps (on pfc circuit) with a dimension of about 5 x 3cm od. I just added a few turns to stand as a primary windings with NP of just 4T while I kept the original windings to stand for my secondary having a number of turns of about 43T x 2 @ approx 1mm dia each.  The primary wire is just an ordinary stranded wire taken from the output of an old pc smps. I cut 2 of them and twisted both ends then that end is used as my center tap. The direction is maybe a bit confusing to many diyer’s if we will not mention the right reference. If we will start from center tap, then one side will be CW while the other side will be CCW, but if we will just start from one end towards center tap ‘till another end, then all windings will be in one direction only.  Secondary is either CW or CCW is not a problem because we are dealing with push-pull topology except if we will sync the starting pulse to another circuit... My output reading with 127V/60W lamp load is around 200VAC in my analog meter but of course this is still considered false AC reading for a high frequency like this.     

You plan to remove few parts before the gate. As I know they are not drivers but a clamp circuit to improve the waveform going to the gate. I observed during my waveform measurements, it helps to chop those unwanted impulse that interfering the proper operation of the driven mosfets. So, I kept the circuit as it is and it is the same idea as Mr Dally’s recommended schematic.
Best regards,
576Pinoy_tech   

This is a little bit off-topic but it might help some experimenters who are pulsing inductors.

When a rectangular pulse transitioning abruptly from 0 to some voltage V is applied to an inductor (e.g. coil) and a resistor in series, the following sequence of events happens:
1) At the beginning (point A) no energy and no current is flowing.
2) Shortly after the rising edge of the stimulating pulse, the current increases linearly
3) Some of the energy of the pulse is converted into the magnetic field in the inductor and some energy is dissipated in the resistance as heat. At this point the energy flows into the inductor faster than it is dissipated by the resistor.
4) After the time equal to 0.69 Tau (point B) the energy flow (a.k.a. power) into the inductor reaches its peak and start decreasing afterwards, eventually reaching zero power and magnetic energy equal to 0.5*L*((V/R)^2) at Tau >> 5
5) However the current through the resistor keeps increasing non-linearly but monotonically and asymptotically up to the V/R limit and the energy flow (a.k.a power), dissipated as heat in the resistor, increases similarly up to the (V^2)/R limit.
6) After time equal to 1.15 Tau (point C), the magnetic energy accumulated in the inductor reaches the break even point with the total energy dissipated as heat in the resistor up to that point in time. Continuing beyond point C guarantees that more energy is dissipated as heat in the resistor than stored as the magnetic field of the inductor.
7) After a very long time the current reaches the V/R limit and the magnetic energy stored in the inductor reaches 0.5*L*((V/R)^2) limit but the energy dissipated in the resistor increases ad infinitum at the rate (a.k.a. power) equal to (V^2)/R.

Tau = R/L (a time constant)
V = The high level voltage of the stimulating rectangular pulse.
E_TOT = Total energy delivered by the supply to the series RL circuit.
E_L = Energy stored in the inductor as magnetic field
E_R = Energy dissipated in the resistance as heat
P_L = Instantaneous Power (energy flow) flowing into the inductor
P_R = Instantaneous Power (energy flow) dissipation in the resistance
i_L = The current flowing through the inductor (and resistor)

For transformers, putting a load on the secondary winding (e.g. shorting it) has the same effect as decreasing the inductance of the primary winding (L). As a result of this, the Tau decreases and the current in the primary rises faster with time.