using my series high side switched Mosfets, and my impulsed primary I made a "tesla"coil.
I used 0.4mm thinck (diameter) copper wire, to make a coil of 8cm high and a diameter of 8cm. so 1:1
L2 primary coil is a bifilar made of 2x14 windings of 0.75mm2 copper/ PVC speaker wire

The L1 impulse generation coil was underneath L2, close coupled to it. I? started with 15 turns of 0.75mm2 speaker wire (also bifilar).
But while tuning the impulse to the L3 secondary Tesla coil, I reduced the windings to only 8 turns.

L2 series resonant was tuned by a 117nF capacitor and I tuned above its resonant frequency, to get proper impulses.

Measuring L3 secondary was done by a high voltage probe, with a piece of copper tape connected to it. placed far enough away from the coil to not influence the (low) capacity of the secondary (so it wouldn't detune).

I added some copper plate to the top of the L3 secondary Tesla coil, to give it a bit more stability, and lower the Fr slightly.

The L1 impulse is super imposed onto L2, and tuned in duration, so it was slightly quicker than the L3 coil.

impulse was -3kV at 280.8ns duration giving a full wave frequency of 1.7806Mc/s
L3 Fr=1.6667Mc/s with a half wave duration of 300ns

I could add some capacity parallel to L1 to tune it even better, but it would need to be very small and I dont have smaller than 68pF. I could put some in series... maybe...

input power was around 62.4W dc: 1.04A x 2x 30.0V
The freqeuncy where the mosfets were driven at was 97.9kc/s which is the 17th sub octave

The purple spark was around 1cm visible in daylight, but when made dark it was twice as long

during tuning, I had one spark over from the L3 to L2, which damaged one wire, which I cleaned up and coverered with tape.

I made it, so that the secondary L3 could also be run in half wave mode, ungrounded. The L1 L2 coil would then be placed in the widdle of the L3 secondary.
The ends of L3 would then be brought close together for spark over.

Since Tesla didn't ground his secondary, but used both ends of his secondary I decided to do the same.

for this I moved the L1 L2 coil to the middle of the secondary L3 coil, and I connected the ends of the l3 secondary to a spark gap.

this is a very efficient method. I can easily produce big arcs with very low power.

Interesting to see this is much more efficient.

this makes me think about my other experiments.

less than 18W for 1 cm spark

The only thing is, when not grounded, the secondary coil becomes half wave resonant, doubling the  resonant frequency (in comparison to the quarter wave resonance when grounded), which means the impulse will also need to be twice as fast. So again, L1 would need to be tuned.
I only have 8 turns now, so I would then prefer to make a unifilar coil for L1 with more windings.

tuning the impulse to the secondary again, would probably make it even more efficient.

instead of making the L1 faster, I made a new slower L3 secondary.
twice the lenght so 16cm.

I need a bit more power to drive it, but the sparks are also very quickly available. So quick that I need a new spark gap for it, as to drive it fully up to the 3kV impulses, I now burn away the steel needles which were used for the 1cm spark gap.
I have som tungsten rods, that I will use to create a proper spark gap, of several centimeters.
Its a fun project.

I think I see an increase in effectiveness when the secondary is not grounded but both sides are resonant (half wave mode)
But that is not fair, as my quaterwave spark had no ground to spark to, so That will need to be compared, if it doesn matter, than I will just use the half wave resonant mode, as it is like Tesla used. But he used auto oscilating relays, so the primary had a ringing, not the efficient half wave I use.

After realizing, the large C2 series resonant tuning capacitor of the primary, was demanding high currents, I decided to make the C2 very small (68pF) and see if I can make L2 equal the (impulse) frequency of L1, and the secondary L3.

This should make it much more efficient

scr151
This first test shows the primary in yellow, and the secondary in orange.
They do not match up perfectly, but it already is a good start.

the impulse is the first negative halfwave of L2.

to me it appears that first the L3 secondary (orange) is faster than the L2 primary (yellow), but after five oscillations, this changes. very odd.

Since L1 has Less windings than L2, I will reduce the windings of L2, to make it faster.

scr152 shows the impulse ,probed at the source.
Since L2 now matches th impulse, I wonder, if the second half wave still is recycled.
If Not, then it is not as efficient, and probably needs more energy, as the resonant currents burn away the energy through resistance

I first removed 1 winding from my L2 primary coil, to speed it up.
Then I added 68pF parallel to the C2 series tuning capacitor making it 136pF in series with L2 primary. This slowed it down again, but also reduced the ESR.

the primary easily exceeds the 3kV pp , but the rating is for DC, so hf AC should be able to go a lot higher. I should keep a finger on the temperature of the caps to keep it safe.

scr161 shows the impulse at the source
scr160 shows the primary in yellow and the secondary in orange (capacitive coupled, so voltage is much higher)
tested at 95.7kc/s with 1.00A and 2x17.2V dc from the power supply (34.4W)

scr155 shows half wave duration of the secondary, with just one 68pF cap for C2 primary.
scre154 shows impulse duration (half wave of L1) which is longer than the secoandry.
I must compare the L2 first half wave of ringing, to the impulse of L1, and see if they are equal in duration.

yellow is impulse at source
orange is primary probed between coil and C2 (136pF)

The L1 impulse is clearly much faster than the L2 secondary.

With my first tests, I had a large C2 cap, and I had matched the impulse precisely to the L3 secondary coil.
it seems the secondary is now slowed down, Is that due to the L2 ringing?

I need to reduce the windings of L2 primary, to make it faster, and also reduce the C2 capacity back to 68pF

I removed in total 2 windings from L2, and I reduced the C2 cap again to 68pF.
Now L2 is to fast. I will need to see if adding another 68pf Makes it to slow again. I could then add 2x 68pF in series, to make it 34pF addition.

I compared the impulse speed to the secondary and they still match up.
Current draw is high, 1.18A @2x 19.9V =47W at 90.2 kc/s

scr 170 shows orange half wave of L3 secondary= 258.8 ns
scr163 shows yellow impulse halfwave is the same 258.8ns
scr167 shows L2 yellow and L3 orange. whereby yellow is to fast for orange