Logic would say:
I use tem mode,
so when I tune to LMD mode,
I can flip L3 back over.
Voltage are then out of phase again.

problems with this:
frequency is very high due to close coupling.
thus voltages remain low, and need very high voltages from the psu, but that can be done.

capacity can also be increased to lower the frequency, but this will again, also lower the resonant voltages.
But, it will increase the currents, which can be good.
 the current probe is not very accurate above 100kc/s but should pick up the resonant frequency.

another trick could be to just use one coil for high voltage (small cap)
and the other for high current (large cap).
to get the frequency down.

L3 small cap gives best results from previous research with voltage impulses.

very curious what the current impulses will do with L3, when the cap is small enough

the square wave capacitor could also still be tuned, as it tunes the L1 curent impulse, which can be used for close coupled L4, which should match the current impulse quarter wave.

I gave L2 62nf which made it basically non resonant.
L3 kept its 9nF. and was still flipped over.

scr 483, is at a lower harmonic of L3. zoomed into the L3 at the time of the discharge of c4
Notice that large positive overshoot in orange. clearly the fast displacement energy isn't a match for the the L3 resonance.

scr484 is zoomed out, at the same frequency of 71,97kc/c clearly can be seen we are at the thrid sub harmonic of L3.
and the overshoot is also visible.

scr485 is at the L3 resonant frequency of 214.3kc/s
Still the overshoot is present.

now lets try the reverse, Make C3 big, and C2 small.

this is with C3 large and C2 small
same sizes, 9nf for c2 61nf for C3

I really should also look at this with the current probe.
as L3 now produces 64V pp with 61nF and at 221 kc.
This is with a little more power from the psu.
L2 does now become resonant.
clearly showing the L3 coil needs to be flipped back, in the higher LMD mode, to get it out of phase again.
that is scr486

scr487 is at the lower third harmonic at 74.1kc/s
second harmonic doesn't work, due to to charge and discharge displacement currents, that only line up with odd harmonics.

ok to get a clearer picture, I will flip L3 back. and test again.

fo rrefference, I didn't have the LMD mode of c2+c3=9nF
So here it is.
Very high frequency of 287 kc.s
psu was at 2x19.7V @ 0.46 A dc
L2 not really resonant in this mode.
L3 (orange) just barely showing a sine wave, of 68.6V but it does appear to be out of phase, so indeed should be flipped.

I really prefer how it looks when L2 is also resonant, adding to the voltage difference of the cap plates.
Maybe... I should stay at the TEM mode for now

I made C2 smaller to 7nF and C3 bigger to 11nF
input PSU: 2x8.4V  @0,5A (8.4W)

I need to compare this with the equal 9nf each setting.
f=72.89kc/s (TEM)

L3 orange is 789Vpp

to compare, C3 is now smaller, 7nf
C2 is larger 11nF
F is 72.81kc/s (the same)
psu is the same, 0,5A 2x 8.4V

So with c3 smaller I would expect it to become higher in resonant voltage.
but... it is slighty lower than previous test.
lets compare voltage difference between L2 L3

note that L2 with its larger c2 11nF but only a little... not worthy of mentioning

Difference between L2 L3:
 850V with both tests

Very cool to play with this.
I got it tuned to 98kc/s now
 4nf and 5nf not sure. tired...
this way voltages become higher.
the smaller cap should have more voltage less current. but it seems the other way round.
also the c4 square wave cap seems to discharge when L2 is resonant.
I have - 400V but then the discharge is only - 200V after a half period. where is that voltage going? is it becoming part of the series parallel resonance of L2 C2 C4? need to check...
tomorrow.
heat wave is coming.
need airco

Ah! now I understand

C4 is small en is quickly charged.
but C2 is relativly large, and charges slower.
so do I need to make C2 smaller, so that C4 stays charged and doesn't deplete by c2?

They are not really parallel but it could befv thef reason why C4 discharges to. half voltage within a half period

or maybe I dont get it.

here, green is the voltage on the C4 cap, should be a square wave, but it empties in a half period.
So... does it flow into the parallel resonance?
Can I do someting about it, so it retains the voltage better?

hmm...

should L2 change phase? and... how do I do that? it could only be pushed out of phase by L3, but then L3 needs to couple to L1? hmm weird.. no that would bring counter emf problems back into the game.

So... use L4 extra coil? I must play with it any way...
Give it large capacity and see if I can couple l4 back to L1.
That would be really fun

maybe the diode is kaput. easy to test

but I have seen C4 discharge before. and it always happened when L2 was resonant.

since L3 is out of phase and a capacitor plate, would the increased tension pull C4 empty?

adding L4.

first close couple it to L1. without connecting it to L3.
Tune it to the right frequency, and see if the phase matches with L3. if not. phase reverse it by flipping it over.

I now realise it doesnt realy make sense to want to give L4 a big cap. since it will need to be tuned to the right frequency.
high resonant currents are fun... but... if it isnt tuned... then... what? no amps

it is interesting, as L1 now also produces a current impulse, since it now is a a quarter wave.
So I could tune L4 to this quarter wave ...
hmmm

I removed L2 L3 L4 and their caps from C4,
to see if the cap still discharges.
and it does not
green is the voltage at the mosfet source.

So the voltage must be feeding into L2/C2

redraw of the circuit, to clear things up.
I left L4 out, as it is to soon to play with that.

I also decided to remove C1, and relabel the square wave capacitor as C1

I could also use wima capacitors instead of the ceramic disk ones that in now use for C1.
But I feel it wont matter.

The resonant coil cap, seems to suck the life out of C1, which makes sense, as the energy needs to come from somewhere.
I just want it to come from the ambient... so... Do I need another diode? in-between C1 and L2?

same test but now with C2=C3=4nf (3x1+10x0.1)
f=103.19kc/s
PSU=0.71A 2x19.0V =27W

a bit more power this time.
The psu amps fluctuated a bit. temperature of the mosfet around 27C
Didn't run it long.

green=outside rim L2/C1/source
yellow= inside rim L2/c2
orange= inside rim L3/c3

deltaV= 1180V between L2 and L3 during discharge, which is only 130V
look at the huge voltage drop of the green line right after the L1 discharge... What is going on there?
looks like a overshoot of more than 300V.
The mosfet body diode, should be blocking, but is it?
The MUR8100EG diode, works fine in other tests.
I might need another diode in series between source and L1?
Would it be possible that the mosfet body diode is leaking away the voltage?
I should have a -500V discharge from C1...

I could also replace the c1 cap. but... it works fine, when the coils cap is removed...

SO WHY IS IT DISCHARGING?

It could also be, that the L1 discharge is to fast, and produces to high currents. which can be tamed with a parallel capacitor over L1. but again... it seems to be able to produce a perfect square wave, without the coil cap.

or is it. I never tested with these high settings.

I wonder... If C1 didn't discharge, would the voltage restore to its max negative value?

ok, now I am getting some insight.
I increased both caps to 9nF again, and tuned it to 70,9 kc/s
with a low power setting, I almost got the same values, as with a higher power setting.

2x16.2V @  0.95A  =31W     SCR496  (current again fluctuated)
2x11.4V @  0.67A  =15W     SCR497 (current stable of PSU)

So barely more voltage, with twice the power input.
it seems, the diode might be the flaw.
luckily I ordered some new heavy weights which will arrive next week.
I hope they will keep up with the high currents.

Else I am forced to put a parallel capacitor over L1, to slow down the discharge current to more acceptable levels.
In the end I feel it is the high current from the L1 discharge that overloads the diode, which also prevents it from proper turn off.
And this might also cause the leakage of voltage from C1

BUT AGAIN WHY DOES IT STILL PROD?UCE A PERFECT S?QUARE WAVE without L2 L3.... arrggg

https://nl.mouser.com/datasheet/2/308/1/MUR8100E_D-2316385.pdf
peak repetitive forward current square wave 20kc/s T=150C 16A
ONLY 16A?!! definitely need another diode
non repetitive, 100A

its rated 1000V DC but that is not what I need.
it should be fast. I wish I could parallel diodes. but they don't work like that (no current sharing)

since it is so hot over here, I might use the heat to cast some epoxy, as it is more liquid, and thus lets the air bubbles go easier.
L2 L3 counter rotated, sticked close together.

if all else fails....
SPARKGAP
no diodes...

or even better. an auto oscillating relays.
OH man... there has to be a solid state way

slowing the impulse amps down seems the best path right now.

Slow charge fast discharge was always the goal.

yep... parallel cap over L1.
the current is to fast to measure right now any way. so I should slow it down.
I could even charge it over a half period. hmm, that would definatly also demand a much higher input power, to get a proper voltage to charge in C1.

trumpet format is pretty is there a circuit do that or a wave form generator?

Quote from securesupplies on June 9th, 2023, 07:02 PM

trumpet format is pretty is there a circuit do that or a wave form generator?

Yes can be made with a power supply that gradually increases in voltage.

Like using a large capacitor to power the resonance,
while slowly charging up that large capacitor.

and, after its charged, discharge the large cap to start over

I remember playing with a grounding series cap before. when it was to small it would follow the resonance.
since it was on the opposite side of the resonant coil
 it would show the opposite polarity voltage change.

So... I need to enlarge the C1 capacitor.
and im doing so, it will also charge up slower as the L1 resonant frequency will be influenced.

thus, I dont need a parallel capacitor over L1, as C1 already does do the trick.

still, it will demand higher voltages from the PSU to charge it up.
but totally worth it, as the discharge will provide more displacement current.

so, out with the 615pF ceramic caps, and in with the wima caps. I still have a bunch of 330pF which I can add in parallel. hmm maybe I  can leave ceramic disk caps in, as they are pretty much hard wired in.

I added 4x 330pF 1600Vdc wima fkp1 caps,
in parallel with the 615pF ceramic disk caps.

increasing it to nearly 2nF.
Not a  big increase but enough to see the difference

ok it works. the diode now keeps working.
still some decrease in dc voltage, but that will reduce more, when I increase capacity.

scr498 : PSU 0.59A 2x13,4V=15.8W
scr499 : PSU 0,92A 2x20,3V=37.4W

now the resonant energy increases, and the diode blocks the reverse voltage with more than doubling the input power.

1814V between L2 and L3 at time of the discharge.
discharge is slightly more, to around -300V.
discharge ringing is also more intense. I might need to dampen that, with a resistor.

But first, increase the C1 cap even more, to see if I can stabilize the DC voltage.

this was at 103.32 kc/s with 2x 4nF for c2 and c3

I now replaced the 4x330pF with 4x1000pF increasing C1 from 1.9nF to 4nF
This is clearly to much. everything is ringing madly, and the DC component is almost lost.
increasing power will only increase ringing, not giving the DC more voltage in C1

So, I can decrease C1 back to 1,9nF with the 4x 330pF which seemed to work fine.
And instead, I can decrease C2 from 4nF to 2nF.
while increasing C3 from 4nf to 8nF,
to keep the frequency low enough.
I always aim to stay below 100kc/s at the poly poly propylene caps don't really like it above, and the current probe also stays linear below 100kc

L2 will have a larger resonant voltage from the smaller cap, while L3 should have has less voltage, altough I have seen L3 increase more than L2... somehow?