made a little video on the H bridge, just a teaser.
this is a double half bridge (making it a H bridge) run out of phase.
The potential of just a single half bridge is immense, but a full H... that should really kick some... POWER

https://youtu.be/bAf4dIMfB2M

since ground will now be in the middle between plus and minus,
I will need to use an isolated dc supply for the lowside switch also.

https://nl.mouser.com/datasheet/2/281/kdc_mgj6scdc-1019175.pdf

this is what I use for that
Murata isolated dc to dc supply
MGJ6D122005SC

I made a drawing how I set up my H bridge, and how it also works as 2 half bridges (out of phase).

to do so, i needed to isolate the gate drivers on the low side also. since the (bnc) square wave which drives it is referenced to zero volts ground.

I made a half bridge setup with 2 coils, and a dual power supply (earth zero between positive and negative).

one coil is producing impulses, the other coil discharges the impulse to earth zero by being series resonant.

I finally have the impulse on both maximums of the series resonant sine!
very proud of this result.

the circuit still rings far to much, so I will probably need to rebuild the h bridge with smaller leads and higher gate resistance. maybe even a source resistance (to the gate driver).

still having trouble with driving the mosfets proper, but startimg to get the right results.

screenshot is a 784Vpp Sine wave (series resonant) with impulses on its maximums.
around 74kHz

this acts as the primary of a true Tesla coil.

it's made by a halfbridge circuit, and a symetrical power supply (coils grounded in the center ground.

I'm trying to switch my SiC mosfets, by making use of a isolated dc dc converter of murata,
MGJ6D122005SC

The 4 outputs are ringing as shown on the photo. 1V/div

4 pins output of which 1 is unused (causimg problems maybe?):
-5V
0V
20V

-5 and 0 ; and the  0 and +20 are both filtered by parallel caps:
47uF
10uF tantalum
1uF 3x7r
100nF
(so that 2 times)

I have 680 ohm 2w load over 0V and 20v
and 180 ohm load over - 5V and 0V
to give the 6W converter someting to work into. the SiC mosfet needs far less (using it for the - 5 output, but considering 2W version for less troubles).

pfff....
should I do something with the unused output pin?
maybe use a series resistor to create a better low pass?
going to find and read an article about power supply filtering.
or look for a 2W version with just 0 and 20V

EDIT:
www came up with Pi filter, so basically add an inductor in between 2 caps ( on both -5 and +20V outputs). that might do the trick. I have low resistance (high power) 10uH that I used for the input, so I will use 2 of those on the outputs also.

solution :
use battery

insane much noise from the Murata isolated dc to dc converter is messing up my gate driver IC.on its in AND output.

causing the duty cycle to fail.

found the solution in a pi filter on its in and out puts.
basically a 10uH inductor on ALL its in and output lines. man that thing is so noisy.

fingers crossed. I hope this will soon work. pfew...

my pulse signal was 3v some how (I had it 3.3V before)

changed it back to 3.3V and now duty cycle works again.

Amazing how much frustration 0.3V can cause

got the basic setup working.
only blew up 1 diode, on the high side. used to block the impulse into the gate driver circuit, so it doesnt resonate.
wondering if I can place it out between the mosfet and coils, and place it between source and gate driver circuit, The diodes really get hot and need cooling (mur 8100e)

picture: left, top diode overheats much more. now with it in, only positive impulses, with it bypassed only negative impulses. ehhh what? oops need to think about this...
right might be a solution.

without diodes it oscillates on the gate (maybe I should check that again, and bypass both.
first a break. my brain isnt at full capacity

EDIT:
No won't work, body diodes of the mosfets would pass (some) of the impulse energy, and need to be blocked by the diodes. Solution is to use heatsinks on the diodes. TO-220 style. must have something laying around

since the 3rd coil is parallel resonant, and producing inphase voltage and current, it might be better to put a equal coil close coupled on top of it, and use that one for the output. this way, the load only sees L3 via L4, and the source L1 and L2 doesnt see the load L4 (maybe)

making it a 4 coil stack:

L1= impulse generator coil
L2=series resonant coil receiving impulses coil (primary)

--distance--

L3=parallel resonant coil (secondary)
L4=output coil.

my half bridge circuit now produces positive and negative impulses from the high- and low switch.

but they ar unequal in voltage. positive is 550V and negative is - 350v

I feel this is due to part of the negative impulse is seeing the source and also the battery that charges the source.

the positive impulse only sees the drain.

The only solution to get them equal might me to tap into the positive impulse and use  it to charge something.

I might create a negative dc offset on the L3  parallel resonant output coil. This would setup a dielectric field between L2(impulsed series resonant) and L3.
the impulses would then follow the field lines.

this is also what thomas townsend brown did, exept he used unequal capacitor plates, so maybe in the future I might change the L3 to bigger.

I could also try to make a bypass to charge the other battery. (of the low switch).

In the end I want to engineer the batteries out. then I would need a (fast) zener... troubles...

for now.. I first need to balance the impulse voltages

I realized I did only use one channel of my power supply, and used 2 capacitors to ground, which didn't have resistors over them.

So the halfbridge ground wasn't in the middle. the low side had more voltage, so the spike was bigger.

Now I use 2 channels regulated from my power supply in series.

et voila!
impulsed series resonance, with equal voltage impulses.

picture shows 69kHz from 60nF
will make it 50nF.
this was with 2x 6.7V (13.4v total)

Kick it! :)

73 kHz
2x11.3V=22.6V input on the half bridge.

1424V impulses positive and negative!

475Vpp resonant sine

12.1App resonant sine

impulse duration: 800nS

if I add the secondary this will all drop, this is only measured with the primary (L1+L2 close coupled)

impulse duration is a half wave

it needs to be tuned in harmonic relation

to series resonant L2?
or parallel resonant L3?
Both?

goal is a trumpet wave, or as CP Steinmetz in his 1914 second edition called it :"cumulative wave"

Also,
I looked at the current of L1, while tuning slightly below resonant frequency of L2. didn't look good. tuning above does look good like a proper sine wave (and impulses on L2).

tuning just above Fres gives slight impulses.
when voltage is pushed up, amps also rise in L1 together with impulse voltage.

but key is, L1 L2 close coupled with reduced power draw.
Thats why I need to look into the ratio of impulse duration and period of L2.
see when current draw is lowest. for max impulse voltage.

logic would be if the impulse created by L1 is a higher harmonic of L2. this should be a good starting point.

once this is achieved, L3 is again a harmonic of L2 AND L1
making it pretty hard but doable

What I also did do is tuning below Fres L2
the current Sine of L2 then jumps through zero at the impulse point.
Which makes sense.

Maximum impulse should be achieved when current is interupted at its maximum.

this square wave current, together with the (odd?) harmonic relation between impulse and L2 resonance seems key

measured impulse duration 790nS
is a half wave, full wave duration is double.
1divided by this is the resonant frequency of L1

L2 is the 9th harmonic octave lower, tuned by a 57.8nF series capacitor.

L1 gives 16A peak to peak sine
L2 gives a 26A(!) peak to peak sine.
L2 has a 990V pp sine with 1000v Impulses on its maximums

It looks as if the impulses are helping the current wave to grow, the current sine seems to "jump".

prolonging the impulses somewhat might influence the reaction of the current build up.

small duration would need more voltage on the impulse. (energy is surface area)

I tuned this so L1 is the 9th harmonic of L2

pricture: L2 series resonant voltage (yellow) and current (blue)

N Tesla always used series resonance. not only with his impulsed primary but also with his secondary.

I use parallel resonance for my secondary.

series resonance needs to be able to somehow discharge the series capacitor into the coil. simply adding a series capacitor does not guarantee series resonance, as the cap can charge up, but wont discharge by itself. it just will stay charged.

So we need a discharge method.

Put a load on it. Yes. maybe but that same load will hinder the resonance.
thats why I added a 4th coil close coupled to the 3th coil, for output.

parallel resonance is in principle series resonance, but close looped.

And lets not forget, Tesla's series resonant capacity (top load) is a single plate capacitor. It literally uses the ambient surrounding medium as "ground plate"

I am researching what happens with L1 and L2 coupled. as I saw before, the dc current draw drops. but there is more

I have a 500khz current probe, and I probe (blue) the series resonant current between L2 and its tuning cap (57nF).
the yellow is the impulsed voltage sine of L2.

What I can clearly see is the jump in current, right after the impulse.
I need to see if I can tune that jump in more prominently. based on impulse duration and L2 resonant frequency as a harmonic.

Another analogy came to mind. If we push a swing (in this case from both ends) if the push is long in duration, we can accelerate it enough, if the push is short in duration, we have to put A LOT of energy in that push (high voltage).

the push seems to help the current to build up.
It is only if I tune slighty higher then L2 Fres. when the impulse is just showing up on the voltage sine.

it only seems to work, when L1 and L2 are close coupled.

picture: see how the current sine blue, jumps

dc input 1A 30V
73.8kHz
16.5App L2
1000V impulses 790nS
L2 resonant sine 500Vpp

I am going to a some few hundred pF to L1 to prolong the impulse duration to 1uF
this means basically tuning the resonant frequency of L1 to 500kHz.
the 7th sub harmonic of that is 71.4kHz, I will tune L2 for that frequency, and look again to the current build up at the time of the impulse.

I will need to drive more current in, to get around 1000V impulses again, as the duration is longer.

I put 4x330pF parallel over the top half of L1 (source and middle of bifilar coil).

Expected a longer duration. but no
impulse is faster... ehhhh?

and current shows a high frequency ring

not what I was expecting

I have played enough.
parallel capacitance over L1 does change duration as expected, I think I did a measurment error. nothing special. parallel caps over L1 can be used to make the impulse slightly longer, in case it swings through the zero.

removed the caps over L1 again, as this is no problem becasue I have a low resistance setup, and the impulse can travel fast.

played some more, and came to the conclusion, the jump in current, is from the jump in voltage in the L2 series tuning cap. The impulse quickly charges the capacitor, which gives an increase in current.

I added 1nf to L2, total 58nF series now. Fres 70,5kHz. all looks good.
going to put L3 back on.

tuned for max current in L2, and max voltage in L3 (didn't use L4)
put a 230V 28W halogen lamp on L3, and measured voltage and current.
picture blue is voltage and yellow is current.
around 1A and 300V peak to peak. pretty much in phase (500kHz current probe has slight phase shift)

L1 and L2 close coupled, L3 is 5cm distanced from L2
might need to play a bit with distance.

I might need to make a L3 with thicker or thinner wire (equal mass) after I found the best distance.
Also, rectify L3 into cap, and then load it. (try to get DC for easier measurements)

this in phase 1A and 300V would indicate 300W, but the 28W lamp is not fully lit.

How can it be, that the lamp is lit but doesnt "eat up" all the energy present?

Strange.
L3 is my secondary, tuned by parallel caps. and loaded with parallel resistive load (lamp).
I would think current measurement in L3 isnt place dependend, but it is.

Near the lamp the current and voltage are in phase, and current is around 1A pp
but on the same line, if I measure between coil and caps, the current is much higher around 4A and 90 degrees out of phase, as expected with parallel resonance.

So the Load, brings the voltage and current in phase?
its the same line I measure...
Strange behavior

really need to get the rectifier and caps back in place. and not connect the load directly on the coil

I tuned, so the impulse charges the cap of L2 instantly. lamp still burns, no impulse!.

So basically all the energy of L1 that is discharged as inductive spik/impulse, now charges the series resonant L2 tuning capacitor (58nF)

als made the distance closer between L2 and L3, now 3,5cm, and used a rectifier and caps, to get DC into the lamp for easier power measurements

before impulse 59V after impulse 106V   increase due to impulse=47V
energy @ 59V=‭0,100949‬ joule
energy @106V=0,325844‬ joule
thats an increase of 0,224895‬ joule. that is a 222% increase in energy!

WOW :shocked:

and that's not only in the cap, but also in the series connected bifilar coil dielectric field.
I need to get that voltage higher, so it has more inpact on the dielectric field of the coil (its capacitance is fairly low)
So lower, the tuning capacitance, this should result in a higher change in voltage.