Itsu:
Interesting to see the 100 MHz spectrum for the primary winding on your yoke.
You have a very nice spectrum analyzer there. As Verpies mentioned, varying impedance mismatches
through the frequency range may be affecting the results, but it is still interesting to see.


Jeg:
Good work. I suspect any small ringing or spikes on the gate or drain signals is probably
no so critical, except in the case where it might potentially lead to component failure. Otherwise a little
ringing or some switching spikes is probably OK, considering that Akula didn't seem to go to
too much trouble in that area.

We also probably need to watch out for any high voltage spikes feeding back from the tesla coil, once that part
of the circuitry is made active.

     Jeg: 
          Looks very nice, and neat. 
          Maybe use two 12v batteries, if you can, instead of the PS. 
          What kind of earth ground are you using?

  Jeg: Ok, thanks for the reply.
   The reason I mentioned about the batteries, is because I notice some stray AC inductance on anything that's connected to my PS, even with adequate grounding. Which can at times also affect the signals read on a scope.
   Both my scope and PS are using the same AC ground from the grid.
   The earth ground is also very important, my 5 meter ground line, does nothing.  The 37 meter ground line into my water well does. But, it all depends on resonance, without proper resonance of the circuits, the ground lines don't do anything, or very little.

   Good luck with your circuits.  Looking good...

Hi Void
Yes I guess there is not any problem for little ringing considering a self running operation. It is just a matter of self satisfaction

Hi Jeg. Sure. Also it sure can't hurt to get the PWM driver circuit working really clean and stable and efficient.
As others have commented, it is in the very least a good learning experience.

I have not watched or analyzed your methodology yet, but the most common thing that can go wrong with these measurement is distortion of the tracking generator's output due to impedance mismatches and reflections.  The solution is to put a circulator or attenuator at the generator's output in order to isolate it from a misbehaving load.
Not normalizing the spectrum analyzer when the output cables are connected directly to the input cables (without the DUT), is another common rookie mistake (if an isolating attenuator is used - it should remain in place and participate in the normalization).

Applying DC to other transformer windings through a good choke, is another test that can be performed to see how the transformer's core responds at various magnetic fluxes.  Flux generated by such DC, has a different direction than the flux from external permanent magnets.

@All
Notice, how complex the frequency response of a simple yoke transformer is.
I expect the frequency response of the grenade coil to be even more complicated.

It is no wonder that it is so difficult to replicate with such unknown complexities.

If one of the "successful" Russian friends gave us a plots like this of their yoke and grenade - we could replicate this device in weeks.

I did use "normalize" to flatten the response when testing the connecting leads which in this case also includes the 3 turn in- and output coils.
I do realize that when removing the output 3 turn coil and connecting it to the primary under test (put) it could invalidate this "normalization".
I just don't see any other way to examine this primary.

I did not use a 10dB attenuator in the TG lead, so i redid the same test now with one.
I cranked up the TG output from -10dBm to 0dBm to compensate.

The below screenshot shows the result which is similar as the earlier one.

Itsu

I did use "normalize" to flatten the response when testing the connecting leads which in this case also includes the 3 turn in- and output coils.

Hi Itsu. Maybe I am misunderstanding, but if you normalized with the two three turn
windings on the ferrite core, it seems to me that the spectrum anaylzer would try to
normalize the response of the spectrum analyzer to be flat through the normalized frequency
range of 9 kHz to 100 MHz, which would normalize out much of the frequency response variation of the
yoke core. Don't you actually want to see what effect both the yoke core and primary winding are having
on the frequency response? If so, I think you need to just connect the spectrum analyzer probe to the
TG output cable (including the 10dB attenuator if you are using it) and do the normalization, and then
after doing this normalization to normalize out any frequency dependent imbalances in the cables and probe
and attenuator, you would then connect the cables to the three turn pickup winding and the primary winding
to see the frequency response through 100 MHz. Were you trying to normalize out the frequency response of the
yoke core intentionally?

Hi Void,
well, my dilemma is that  doing it your way (which i was meaning to do too), you ALSO measure the response of the 3 turn coil, and i ONLY want to know the response of one primary coil.
I hoped that doing it my way    i could eliminate (normalize) this 3 turn coil, but i am not sure if this will work that way, so i will try the other way too.
Itsu

Hi Itsu.  OK, I see what you were trying to do now. The problem is that in trying to
normalize out any frequency peaks or dips due to the three turn coil, you are also normalizing out
the frequency response of the yoke core. You can't completely win either way it seems. 

Here the normalize was done (as from the books) using the used cables and attenuator before connecting to the dut.
The result is again over a 0Hz - 100MHz range, but shows to my opinion the response of both the 3 turn secondary and one primary combined.
Itsu

Hi Itsu. That is why I was just connecting my probe to a short strip of aluminum foil stuck to the side of the yoke
when I was doing my spectrum analyzer tests. A single loop of wire through the yoke core may have
less impact on the frequency response trace as well. I am not sure what the best approach is for
trying to do these sort of frequency response tests however.

P.S.  Those peaks around 50 MHz and above 90 MHz are interesting.