c5 and c6 are for the dc offset of L2
I always made them big.
but since they now charge to slow, I want to make them much smaller.

C6 can become as small as C2.
this will result in L2 becoming half wave resonant, with both c2 AND c6.

I never tried it before but it should not cause any problems.

so c5 c6 and c2 will be equal in size.

I still have a lot of 10nf caps, so I can already build it.

My head went crazy because I couldn't see what was wrong with the circuit.

but there was nothing wrong! the caps just were to big and charged to slow. thats all!

I cant remember how the resonant frequency reacts when going from quarter wave to. half wave resonance. If it slows down that would be great, but I think I will need larger capacity to get the frequency down enough.
we'll see.

Good that I let it go for a day.
full moon in virgo was very nice and productive today :blank:

I made 2 more 60nF 2000V dc caps.
One with diodes, for the dc offset (C5),
 and the other at the end of L2 (C6)

and it works fine!

tested with 2x14.2V dc @ 0.37A dc input from the PSU
at 135.3kc/s
700 volt impulse giving a -650V dc offset.

So I need more capacity, to get the frequency down to equal levels as when only C2 would be resonant (quarter wave)

this test still had 20nF parallel to L3.
c5=c6=c2=60nF

When turned off, the voltages quickly discharged back to safe levels which is also nice.
Those large caps with high DC voltages always scared me.

These will still bite, but wont be able to rip my finger off.

next... Ill hook up the parallel mosfets again, and see how fast its charges.
since the caps are a lot smaller, it should be quicker.
But I could already see that if I made a larger time base scope shot.

yellow is C2 L2
orange is C6 L2
both ends of L2 are now resonant.
Close coupled L1 doesn't seem to bother C6

I am tempted to short C6 out while it is running, to see what the 700V discharge looks like

ok I did a test to see the charge rate.
psu is set to 2x13.1V @0.33A  which is weak, around 600V impulses

100us per division, so its still fairly slow on the charge, but much better then before.
starting to look like what I want to see.

I can give the impulse more force, by slowing it down (more L1 windings)
Which would also mean a better match for L4.
I think it would be best to make L1 L2 L3 L4 equal in size.
make L1 unifilar to speed it up, while the rest is bifilar.
then measure and match L4 if it is to slow (reduce windings)

more inductance on L1 would give a bigger punch , and a faster charge of the DC.
I dont want to discharge the dc once every hundred impulses. I am maxed to 25 right now, altough I could easily make it.

I think, I need to max out this test, give it full power, and see how it charges.

it is going in thr right direction. happy with these results!

jikes...
I just in time remembered, I only have 1000V capability with the single diodes (mur8100), I need to double them first, to be able to get 2000V impulses.

When discharging the c6,  the impulse voltage will never reach it. pretty easy to break if not handled properly.

oh well, lets just test to 1000V

still half power, 0,42A @2x 15.9V

Weird to again see the two hickups at the starting voltages.

good enough for some discharge tests. lets hookup the  SW2(switch 2) b MOSFETs

first tested with 2x5V then went up to 2x 8,9V

I was lucky, as I didn't tune the pulse for the sw2, but it was aligned with the impulse.

from the looks of it, the dc is completely discharged, and then bounces back, and rings.
So probably need a bit more time to keep the sw2 engaged.

At least, I now see the DC levels drop. Still very low levels buthell yes! its there.

very happy to finally see this. been working long to get this.

long way to go, but now I know it can be done

the impulse on L2 (yellow) is still clipped, as it is charging the dc caps, but thats no problem, L4 will still ring from it

scr321 is the pulse for the B gate, It's a little bit wider, and a bit more delayed

scr320 orange, shows the C6 L2 discharge from this pulse
green shows the other side of C6, at the 4uF cap some ringing here that could be improved.

C6 clearly rings.
Goal is to get a shift in DC, to a more positive level. not there yet...

I probed (orange) the L3 L4 connection,
Still untuned, but the amplification can already be seen with these low power levels (2x6.2V 0,2A dc PSU)
it is in the middle of the screenshot, where also the sw2 discharges C6

I'm still using the epoxy coils, and I noticed the L2 L3 are not 1.5 but 2.5mm2
L1 is 4mm2 and l4 is 0.75mm2

L4 is now to. slow compared to before,
because it is now grounded on one end, and had L3 om the other end.
before L4 was grounded on L3 and open ended on the other side.

So now I want new coils.
L1 must be unifilar with mostly inductance (windings) and low capacitance (not bifilar)
All coils can be made with the same speaker wire, 1.5mm2 and the same length, enough for a diameter that fits in the vacuum pan (for epoxy)

L1 will join both windings into one, 3mm2 winding (parallel connected)

the other are bifilar. L4 can be adjusted if needed. if L4 would still be to fast, then L1 would  need to loose some windings.

1.5mm2 will raise the resistance, so less current will flow. the q factor will be lower.
impulse will be slower. and lower in voltage.
but also the frequency will drop because more windings.

If the amplification works then voltage mist be able to rise so Inneed headroom (L3 L4)

I also need more power to charge the dc up faster. thus more power from the impulse.

I still have 2 equal size coils but I think I should make 4 new ones that are bigger.

once tuned I can again play with coupling distance and see what gives most power.

then an epoxy cast followed by retuning and power comparison (see if it is worth the effort)

I also need 4 parallel mosfets to get the discharge faster, as it looks slow right now.
I want to order a large batch of mosfets. but they are expensive...

lots to do.

I wonder If I can come up with the same trick that created the voltage impulse.
using the sw1 body diodes to redirect the second half wave to c4.

if I want sw2 to only pass dc in one direction...
then NOT flow back. a diode would make sense.
But I dont want to discharge through a diode.
So first I need to understand where the flow back originates.
placing a diode between L2 and C6?

L1 produces the impulse.

what if I short L1 at V max?

slow first half of the impulse, quik second half of the impulse!

can add a parallel cap to increase the difference

edit: can't put a mosfet parallel to L1 because the body diode conducts first half. this is the same problem as with the half bridge.

but it can be done after C2.
just leave the dc offset out.
short c2-L2 to ground at V max of negative impulse.

edit 2
in effect, I short L2!
c2 shorts out to c4, which is the same as putting a parallel cap over L2.
so sw2 drain at sw1 drain
and sw2 source at c2-L2
body diode then conducts positive half of L2 resonance.... nooooo good
so I do need the dc offset to get the resonant voltage on one polarity
then it can work. by shaping the upgoing flank of the impulse to discharge faster

edit 3
sw2 body diode will always cause problems.
there is another option.
use the ripple.
I have had it before, whereby the ripple on L2 (just before sw1 starts conducting) is tuned to the closing of sw1, whereby the ripple is cut in half due to the sudden conduction.
no impulse is then present.
and the ripple should be made large enough.
sw1 should then be a parallel switch.
hmm... I could try it

edit 4
if sw2 is parallel to C2 it would also work.
I thought I needed to use c4, to bypass sw1 body diodes. but that was wrong.
straight from c2 L2 discharge to c2 L1 is still possible

I've been recording the tuning proces of my 4 coil setup.
going step by step, already up to an hour of tuning. needs lots of editing.

what again suprised me, is seeing the L3 L4 voltage AND current going way up to 2500v 15A at 160kc/s when I'm near perfect tuning.
this is with 4 equal size coils, 2cm between L2 L3. L4 grounded like I do these days.

I have 60nf on L2, and started out with c3 20nf parallel to L3 (series to l4)
that gave resonance at 97kc/s with only 5A
signals clearly showed L4 ringing on the L3 slower sine wave

but after tuning by only making c3 smaller to 5nf the frequency went up to 160kc/s
normally with smaller caps you dont get larger currents but niw the sinewaves almost became perfect and gave 15A

also the different points had different distribution of amps, outside rim of l3, inside rim of L3 both gave around 15 A,

while at the lower frequency larger cap, the outside rim of L3 gave much larger currents then the inside rim of L3.

And again, L4 current in phase with voltage.

so this system has a combined resonance effect which is truly amazing.

Will continue tuning to get it perfect, and record everything.

Still without a load, but I really should do a tuning cycle with load also. crazy how this works. :D

Maybe I don't even need a cap discharge! :shocked:

I wonder, if the sudden stop of changing voltage at the end of the half wave, which is the voltage impulse,
is causing the inertia of the moving volume of aether to keep flowing.

ah my words dont flow. Im still effected by the coils they knocked me out of lead yesterday. still recovering.

but the dV/dt is high, and then suddenly the voltage change stops. whould that not be enough to get the water hammer effect?

the inertia of the moving volume that eants to keep moving?

I really need to use a load, and see if I can tune it.

it's time for startimg to design PLL
phase locked loop,
or im other words,
an autotuner.

this IC seems to be a good choice:
CD74HC4046AE

https://youtu.be/VOHCvKXW9Qw

https://youtu.be/RkG0hdWMMNc

https://youtu.be/fol7f_X-e98

After recovering from another dose of radiation sickness, I did a test again.
This time keeping the system on as short as needed, and keeping my distance

I first rebuilt the circuit again, to produce negative voltage impulses. I did not connect the DC offset module, altough I do intend to use it again, as it should increase the displacement current.

I'm preparing to do a loaded test, whereby I intend to tune the output coils (c3).

I used max input power: 0.92A @ 2x 32.2V DC
frequency was 151.1kc/s  which still had some ripples eating away at the impulse, but this gave max output at L3/L4

I forgot to measure the impulse duration and L4 ripple, will do that later... I want more time between tests to not crash from the radiation.

C3=5.5nF (22x 1nF (or20?) caps in series parallel, for 3500V ac capability)
C2= 50nF (5x10nF parallel)
L3L4=2128Vpp
impulse is 1450V down, and 1700V up  ripple is 1,5x 400V
green=current yellow=L2 volts; orange is L3-L4 voltage; blue is mosfet on off
scr324: L2 current (green) 13A pp
scr325: L3in=12A pp (green)
scr326: L4in=5.64App
scr328+scr327:L3out=14.64App 371Vpp (250Vpp without L4 ripple)

Current on L2 can be tuned better, depending on sub harmonic.
So First, L2 needs to be tuned by C2 for max current in L2.

damn my energy crashed again after the test.
I should tune at lower levels, until it is perfect and only then power it up higher.
or even better.
keep a load on it

Its at least very clear now that L2 must be tuned for max current.

https://youtu.be/jE6wvCZIGEE

I am recording the tuning process
one thing is very nice
L3 outside rim clearly shows the ripple of L4, especially when detuned by making C3 twice as big, while driving at the same frequency as before.

I loaded L4 and this again was not giving power, very disappointing, but it would need further tuning.

I also wonder if it would be better to load L3
and also wonder about putting L4 on the inside.

since the impulse barely swings L4.

so, more work ahead.

Since L2 has the impulse, Why put L4 on the outside, with L3 in between?
So I placed L4 in between L2 and L3.
So now L4 is loose coupled to L2, and L4 is tight coupled to L3, which is on the outside.
MAkes a lot of sense to me.
I just swapped the connections of L3 and L4. which means the old L4 now has the parallel cap, and the old L3 is ouside rim grounded. Easy to do.  but L4 is now a bit bigger in diameter, and L3 a bit smaller.

The ringing on L4 seems more pronounced.
Right now the L1 is still to slow for L4, and needs to speed up by removing some of its windings.

The following measurements where tuned to the maximum at 148.0kc/s using the same caps as before, 60nF on L2 and 5nF on L3
scr332 shows the L2 impulse and ripple, in yellow, and in orange the L4 ripple, on thec3  L3 L4 inside rim junction.
power used: 0.97A 2x32.2V DC
-1500V impulse @ 323ns scr330 F1=1547kv/s (not presice)
L3L4C3=2kVpp
L4=1948ns for 3.5 periods, so half period is 278ns which is16% faster then L1 impulse (323ns) F4=1796kc/s

12th sub harmonic of L4=149.7kc/s So this is very near perfect tuning. Still 2kV is relative low, and the ripple of L4 is present.
Should I flip L4 over, to make them in phase?
lets try and see

I flipped L4 over, still in the middle of L3 L2.
These results are better.
L3 sinewave has now also flipped 180 degrees, which is unexpected. but it makes the phase of L3 L4 good for the sub harmonic coupling/ amplification.
So L4 is now counter rotating compared to L1 L2 L3
Still the same for the rest, 5nF C3 and 60nF C2

scr334
Frequency has gone down to 145.8 kc/s from 148.0kc/s, dont know why
power in went up to 1.02A @ 2x 32.2V dc but that is very sensitive to frequency settings.

scr335
impulse=-1750V (250V higher) @318ns  (1575kc/s approx)
L4= 3235ns for 5 periods=323,5 for a half wave, so L1 is now a bit faster (GOOD!)
L4 F=1545.6kc/s

scr336
L3=174302Vpp, L3+L4=2083.8Vpp

So L1 now is a bit faster then L4 which is good.
the L4 ripple is bigger, and continues.
sub harmonic of L4=
10th=154.56kc/s
11th=140.51kc/s

I feel loading should not be over L4 anymore. but strickly over L3. So NO GROUND. isolation transformer might be needed for ground referenced loads. This challenge, would also make it posible to use a negative dc offset of L3 L4 .

I like where this is going

I increased C3 from 5 to 5.5nF which gave a resonant freqeuncy of 139.9kc/s a pretty perfect sub octave of L4, but the ringing continues... maybe the sub octave is to low.
I tuned this so that the ringing seemed to cancel out in the second half of the L3 wave

I probed the current on the coils: ir= inside rim or= outside rim
scr338=L2 ir
scr339=L3 or
scr340=L3 ir
scr341=L4 ir

at this lower freqeuncy L4 current is again clearly in phase with voltage
same settings as previous test.

I tried to tune to a sub harmonic again, but measured L4 ripple first, and yes it changed again.
now at 1479.29kc/s

so 10th=147.79kc/s

I tuned it to 148.7kc/s
 with 4.8nF on C3
and 60nF on C2
 still 17mm separation between L2 L3
the L4 measures 1481.04 kc/s so that is a near perfect 10th sub harmonic.
Still it wont form into a single sine wave.
My experience tells I need to go higher to the 9th (164.6kc/s) or 8th (185.13kc/s) sub harmonic

I can give L4 a small cap to ground.
give that cap a negative dc offset.

discharge it at the time of the impulse when iy is maximal negative.

L2 then has a positive dc charge.
L4 a negative dc charge.

together they are double the voltage of the impulse (a bit less)

discharge should be fairly easy. L4 to ground.

the negative rise would flow first from L4 then continue to L2 becoming positive again.
then suddenly comming to a halt at max dV/dt when the body diodes conduct the second half of the resonant half wave of L1.