I'm glad I did the flip or not video.posted below
It amazes me how the outside rim amplitude of L3 is so high.
and not only that, it is in phase with the L3 L4 inside rim.

How can that be? with c3 parallel to it.

I really should look at the current also.
but I did that in the previous video.
in and out side rim of L3 have almost the same amounts of high currents.

since the outside rim of L3 isn't really coupled to any thing, why is that amplitude so high? and current also! and IN phase.
with half wave resonance they are 180 out of phase like I showed last Christmas video.

So I wonder, is this from a longitudinal energy flow flowing perpendicular to. the wimdings,
and is this flow. also the reason why current is. in phase with voltage on the L4?

Very curious what the dc discharge will do to this.

first fine tune it a bit more.
measure L4 an get L3 on the right sub harmonic. and see if the impulse still fits.

Nearly there boys and girls.

L3 inside rim a has large current 6.2App and low voltage 618Vpp
L3 outside rim has small current 3App and high voltage 1300Vpp
there seems to be a phase change in current at the inside rim!
So L3 is becoming resonant while riding on the L4 wave.
This means L4 is already achieving series resonance with C3.
Awesome!

To make the current complete this is inside rim of L4

it looks that the current of L3 and L4 inside rims are now IN phase...

L4 frequency measured over 2 periods, 1358ns 736.377kc/s times 2= 1473kc/s
10th sub= 147kc/s
9th=164kc/s

right now I am tuned to 156.4kc/s
so I can ad a bit more capacity to C3, to get it to 147 kc/s

ok I added all 10x 100pF to C3, so no I have 5nF (4x1nF+10x100pF) 100pf makes a big change
this is tuned to 147.0kc/s
L4 is now measured over 3 periods: 2055ns 486.618 times 3=1460kc/s small difference due to error.
so... tuned again to 146 kc/s (10th sub)

since 100pF made a big change, And I noticed I wasnt at the mag with the 5nF c3, I tuned down to 146.4kc/s where it reached it maximum.
this is with 2x15.0V @ 0.41A from the PSU, so a low setting, already giving 1268Vpp on the L3 outside rim (green)
orange is L3 L4 inside rim.

impulse half wave is now 361.2ns 2768.55 kc/s devided by 2=1384 kc/s so it appear to be slightly slower which isnt good, but L4 is ringing enough, to be able to amplify from the dc discharge

ok... so where to place the dc discharge?
I cant tune it while running, as I risk getting radiated again.
or maybe I could. If I had longer BNC cables, I could place it outside the shielding

in theory it should become more positive with a negative to positive discharge, as the inertia makes it continue.

so where on the sine wave, do I want more positive? on the negative maximum, or on the zero where it becomes positive?

ok, so inside rim L4 to inside rim L3
with L3 flipped,
makes outside rim L3 resonant

But it needs to be inside rim L3 resonant

So, don't flip L3, connect inside rim L4 to OUTside rim L3.
then insode rim L3 becomes resonant.

edit
nope. I want L3 and L4 resonant at their series connection

I want to. see If I can tune it to a lower harmonic. with a larger c3

impedance, frequency dependent resistance.
the resistance to change, the inhibitor of flow.

L4 C3 series resonant, low impedance. the collector where the external energy flows into

L3 C3 parallel resonant. high impedance,
the store house, where the energy accumulates?

what is the discharge current through the parallel sw2 MOSFETs,
with 1nF and 1000V
seems it is much faster now, around 40ns at 200V but capacity is much smaller.

I need to attach a temperature sensor to my mosfet which should be easy
also place the IR camera so I can see how they behave.

also I want to tune to a lower harmonic,to get better current readings (phase and amplitude are distorted above 100kc/s)

then finally I want to discharge at the negative voltage maximum of C3 L3 L4 and see what happens. but need to test above 1000V

1nF cap 1000V discharged through 2 parallel mosfets having a Rds together of 11milli Ohm (but will be higher by connectors and wires)
1nF 1000V = 0.5mJ of energy
time constant is 1.10e-11, jikes... is that right?
second
milli -3
micro -6
nano -9
pico -12
So, 11 pico seconds? eh... I thought I saw 200V 40ns hmmm....
OK, forget about calculations, they never really reflect reality on these levels.
just measure and be careful

did tests up to 0,9A @ 2x28.2 V from psu at 144.1kc/s (5nF=c3)
temperature rose to 30C which is still low enough.
negative DC offset was around -800V while positive dc was around +1100V
impulse was max -1100V
I made several screenshots, while slowly raising the input power.
discharge was tuned to maximum negative.
I notice, the L4 did start ringing from it, which is what I expect to become cumulative

Note that I changed the V/div scales while providing more input power!

yellow=L2
green= negative dc offset
orange= L3 L4 C3 inside rim junction

Note again the ripples on L2, which cant be tuned away.
I did notice, the discharge can be tuned to a ripple, which then disappears

I dont like how its ringing after the discharge, but it is not to much.
also... those ripples on L2 bug me. they eat away at the voltage of the impulse and thus the dc offset voltage.

also, Notice how fast the 1nF cap is charged back up, to half its final voltage with only one impulse! that might need an increase of the 1nF

since the cap charges so fast, I could also discharge at each second impulse.
the impulse would then not have a real function anymore,
but L3 would be pumped at each flank.
which I was hoping to do with the half bridge circuit and failed.

Interesting.

just look at that, charge, discharge if it would be repeated like that

the 1nF is now 4x 1nF series parallel for 4kV
but I don't need that.
I also still have a bunch of ceramic disk caps, I believe 9x68pF in parallel
so I will place those at the outside rim of L4.

Then, I'm going to make the delay one step (if possible) instead of 5.

the L1 impulse then charges the cap, bij sw1
after which it is discharged by sw2.
hmm, 1 step would even be to much. I only need a half step.

oh boy... that means the whole delay circuit would be useless. and I need to design a new one.

or not, I could use my signal generator second channel, with duty cycle set to 1%

Image attached shows L3 sine wave at the zero volt crossing (where current normally is max) has a vertical linear voltage change.
This represents an inflow and outflow of energy (displacement current)
pumping energy in and out.
making a ring torroid?

ok, I have programmed my dual signal generator to give a 1% duty pulse on channel B which is 180 degrees out of phase with channel A. This way I dont need my delayed pulse circuit. I looked, and I dont believe I can use it for what I intend to do.
the pulse is now around 80ns just enough to discharge the cap. if not I can set it to 2%
edit... I see It is still timed wrong.

ok, I set the phase to 0 and duty to 1%, now it's right.
impulse is created, at bleu goes low. charges cap.
180 degrees later, pulse (green) discharges cap.

I forgot what the turn on time of the mosfet is. 80ns might be to low. If it doesnt turn fully on, then it will have higher Rds and run hot.

td(on) Turn-On Delay Time 142ns
tr Rise Time 27ns
td(off) Turn-Off Delay Time 72ns
tf Fall Time 25ns

https://assets.wolfspeed.com/uploads/2020/12/C3M0021120D.pdf

27+25=52ns rise +fall time
turnoff delay is less then turn on delay, eh... ok  so it turns on after a longer delay, and turns off after a shorter delay, which is half
ok. I'll use 2% duty
When sw1 is on, diode blocks positive so that is no problem
but when cap is discharged sw2 can remain on, so I could easily use 5% duty

I wonder if I realy need the L4 extra coil.

I could just ground L3 through the 1nF cap.

L4 would then be used as a separate output coil, close coupled to the power of the L3

L4 non resonant to be full rectified into a capacitor bank to be loaded.
L4 not connected to L3, so the dc component does not interfere.

Also
the driver of sw2 can be 49%duty as long as it is turned of when Sw1 closes and L1 produces the impulse.

the beauty is, we now have 2 current impulses, just like I intended to use with the half bridge circuit, but which didn't work due to the use of voltage impulses (which is impossible due to the body diodes)

also the voltage impulses gave a double in out and out in of displacement current.

now the current impulses give an in and an out of displacement current energy, on the right spots of the resonant sine.

its just beautiful.
if it works...

current and voltage of L3 are in phase at inside rim
yellow =L2
green = L3 current inside rim
orange= L3 voltage inside rim (and L4 inside rim voltage and C3 voltage)

input was 2x12V (low safe setting)

charge impulse is slower and distorted, 800ns it now needs to pass through 4 diodes (only need one)

I want a smaller cap and only 1 diode.

This was the first test with the half bridge pulsed system. charge, discharge continuous

what, if I had a series cap to ground with L3, and  parallel cap over L3.

to become series parallel resonant.

parallel resonant is easy, and mostly done by the high magnetic induction current of L2 primary.

but series resonant demands a charge and discharge of the series cap (to ground).
And this would provide the dielectric current.

L4 would again only be used for output couling. to not disturb the dc of L3.

damn this is getting more and more interesting.

As this also explains why my impulse is now so slow. the impedance doesnt match. But it it was series resonant again, impedance would be low. getting the cap charged up instantly.
or am I wrong?

only using one coil with 2 caps doesnt work. only the outside rim becomes resonant.
mainly also not working, as the diode is then connected to the to be resonant point.
ok. back to L3 + L4

edit: I might need to revisit this.
cap to ground on outside rim, not resonant.
L3 first without parallel cap.
then It should ring.
Then tune with c3 to sub harmonic.

ok, L4 is back in, like before.
L3 is flipped, and on top.

scr382 shows in green the cap charge and discharge, but not at the max current resonant point of L3 which was higher
sc3383 shows  in green the current of L3 inside rim, and voltage (orange) at that point

scr385 is tuned to the max resonance point @144.6kc/s, showing current and voltage of L3 inside rim , green and orange.
input power is 2x 14.0V (low)

So the discharge of the cap, now also discharges the resonance of L3, and charge= charge of resonance.
Is this ok? or should I flip L3 back.

I changed the setup a bit in placement, but no values are changed

ok, I flipped back L3 so it rotates like the rest of the stack, now its out of phase with L2 again.
but the dc discharge seems to still discharge the resonance.
lets compare this to the previous measurement

funny... when comparing scr384 with scr385:
orange L3 inside rim voltage did not change phase after flipping it over! hmm...
BUT green L3 inside rim CURRENT does change.

Now how can that be... voltage doesnt change., but current does. weird.

Since L3 flipped, gives a higher amplitude current AND
now I see... it changes the phase of the L3 inside rim  current (green)!
ah. ok. good. So:
L3 flipped gives current that is in phase with voltage, and is further amplified.
(already knew this)

So L3 needs to be flipped.

But ... then my question is...
What if... L4 is flipped, and not L3?
hmm

because.... L3 is not flipped in voltage phase. I want to see what happens with the sine wave when the cap charge discharge when it is phase reversed. so L4 should be flipped and L3 normal non flipped.