I found a way to make a primary, that is series resonant and has impulses on its maximums.
mosfet body diode was the problem, by is solved by putting fast diode between mosfets and power supply.

I used MUR8100e before, but found an even better one.
it is listed as BYT 12P/1000
https://www.reichelt.de/nl/nl/fast-recovery-diode-to-220ac-1000v-12a-byt-12p-1000-p6356.html?&nbc=1

but st electronics calls it : STTH1210D

ok, I put a cooling body on the diodes, and did a test
L3 parallel tuned with 6nF gave 2700Vpp (caps stayed cool, but switches sparked a bit) :P

L2 tuned with 61nF, yellow is voltage sine.
L2 was 275V pp

F=92kC
input power, 1.12A X 2x20,4V=45.7W

impulses are 640nS (pretty fast)
 impulse voltage around 675V pp

So it works. Still not ideal, but good enough for testing.
I didn't push the impulses to higher voltages, as L3 already was at around 2700Vpp and started sparking over at the cap switch tuning board.

duty cycle is set to 48% might be needing more or less to be ideal. if it is to fast, the impulse looses power through the opening mosfet.

nice

here are some pictures, of the setup running
yellow L2 parallels  with 61nF series
green L2 current (61nF)
blue L3 voltage with 11nf parallel

I wonder what happens when I close couple L3 to L1.

l1-l3----L2

Quote from evostars on October 2nd, 2020, 09:56 AM

I wonder what happens when I close couple L3 to L1.

l1-l3----L2

I just did that.
L3 not tuned.

I added L4 to L2. close coupled.

L1-L3-----L2 - L4

and loaded L4 with lamp.
that also works. and very efficient.
L2 also is low impedance (series resonant)
which is ideal for power transfer.

I tuned L2 to a perfect sine (L1 impulses fully absorbed into L2) which gave the most power.

very interesting.
tuning L3 with parallel cap is next.

Very interesting this new setup.
L1-L3-----L2-L4 (see photo)
I made L3 parallel resonant, and still get good results. best with L2 fully resonant with no spikes (L1 impulses fully absorbed into L2)
I can also still tune to the higher resonant frequency.
But this time, the power input is very high.
I have seen that before, when the coil was flipped over.

Since L3 and L2 sine waves are out of phase, Is suspect, I need to flip over L3.
L3 probably is out of phase with L1, so flipping it over, will make it in phase again, reducing power draw. instead supporting it

I could also flip L2 and L4 around. Since I then dont have to disconnect the coils. (I got 2 coils packed together).
L2 becomes L4 and L4 becomes L2

https://youtu.be/7RxRnTo2iXg

I flipped L2/L4 over and reconnected them.

does still seem to work, but odly, now the L2 isnt switch at its maximum voltage any more. this should be right on top. It's probably all about tuning. When I remove all capacity from L3, L2 does switch on top  of its voltage maximum.

I could, still try and flip L3. But L3 is close coupled to L1, and this would make them counterwound, Aether flow never can be natural here.

Why is L2 offset... need to think this over.

This shows the offset. L2 isn't switched on it voltage maximum.
yellow L2 blue L3. they are infase now, as L2/L4 are flipped over, so they are tuned to the higher LMD resonant frequency.
L2 and L3 are both tuned with 61nF
Note that L3 is much bigger, having more energy. It is resonant from L1 and resonant from L2.
But L3 should feedback into L1 to reduce current draw from the supply. now it isn't helping at all, and it drags on L1.

tuning...

If I make L2 capacitance smaller, the impulse isn't fully absorbed anymore when tuned to the resonant frequency.

L2 and L3 are both resonant,
can it be they are not tuned to the same frequency, in other words, their capacitance isnt tuned to the same frequency ?

Earlier tests, L3 was made resonant from L2, so they always were in sinq (eh spelling?)
but now L3 is powered by L1 in trasnverse mode,

and L2 is powered by the half bridge directly.
So they can get out of tune, L2 and L3 need to be proeperly tuned to both their frequencies so they become one.
The LMD resonance can feedback from L2 in to L3 into L1, closing the feedback loop.

I noticed, with a distance of 25mm between L2 and L3
I could tune L2 to each frequency I wanted, and it would still absorbe the impulses fully.

So I reduced the distance to 15mm and now I do find a threshold. it is around 80kC/s
where L2 is 57nF and L3 is 51nF

with smaller capacity, the impulses is to strong and will over shoot the capacitor.
if the cap is bigger, the impulse is absorbed fully.

this was tested with 2x 24.8V dc input. and 28W load on L4

still a phase shift on L2.
I do not yet fully understand why it is not switching on its maximum.

the impulse kept not being fully absorbed if I increased or decreased the supply voltage

If I increased L3 capacity and retuned (lowered Fres) the impulse again wasn't fully absorbed

So distance,  L2 and L3 capacity determines if the impulse is fully absorbed
L3 can be a bit lower in capacity. (10nF)

best tuned (for impulse absorption into L2)
with L2 61nF, L3 51nF and distance 15mm

Probably larger distance would demand larger capacity (tested with smaller caps and stayed absorbed).

Ok. what does this mean? what does it say?
why does a closer distance and higher L3 capacity or smaller L2 capacity make it unable for the impulse to be fully absorbed into L2?

tested with L2/L4 flipped over, and in LMD mode, L2 L3 in phase.

I should still test reversing  L3

blue L3
yellow L2. notice switch is not on maximum

same but L3 is 41nF and power input is max
L2 still 61nf distance l2 L3 still 15mm

28W lamp is lit but not bright.

If L3 is not tuned with parallel cap,
LMD resonance isnt possible. only TEM resonance.
(or maybe at a much much higher frequency as L3 is resonant much higher).

Without caps over L3
TEM resonant mode if L2 is at 47kC/s
lamp than is easily bright, at 2x 8.2V and 1.86A dc input

L2 wave then still fully absorbs impulse whem series resonance is dailed in.

note that in TEM mode, L2 is out of phase with L1/L3 and needs to be reversed (didn't do that)

L2 has much higher amplitude 546V pp

I moved the bridge connections of L2 and L4 to the other side, since they now are flipped over.
this makes the insides (l2 l3) flat without wires. The connecting wires of the biifliar windings are in the outside.

again used 15mm distance, used nylon rings instead of cardboard rings.

now L3 is 25nF and L2 61nf
F=115.86 kC/s
why this big change in values?
the distance seems a bit more equal.

more importantly, now at this setting, the impulses are discharging L2. I must be doing something wrong here.

Tune again

Reversing L2 must have switched polarity of the sine with respect to impulse polarity.

So I should only reverse L3 to get it in phase with L1 in LMD mode.

this means I have to reverse L2 L4 back also...

Since I need to disconnect L3 any way, I will reverse L1
is that right?


L2 L4 are flipped. relative to L1 L3

I need L1 L3 L4 in the same direction.
only L3 flipped.

Yes. only flip L1.
Pfew...


but... if I flip L1 its magnetic polarity versus L3 polarity will change again.
yes. that's alright, as L2/L4 will be also reversed.
or...
we'll see...
can't think Clear. tired

Instead of physically flipping L1, reversed the wires of L1, since it is not resonant.

It does work, but at a cost. high input amps... not good.
But... it isn't tuned yet. this was predicted.

Still not sure about this.

reversed L3 relative to L1 L2 L4
as suspected no. doesnt work.
Yes it does work, but much to high input power.

Still the idea of a feedback loop is good.
Also, I do like the L2 as output, low impedance, perfect. and it works.

Let's for now remove L3 parallel resonance from the tests, and play a bit more.

To get LMD resonance, I need a secondary to the L2 primary. L3 is that secondary.

To tap of more from the primary. is that reasonable?
where is the amplification?
The though was, feedback was from L1 into L2 (series resonant impulsed primary) into L3 (secondary) back into L1.

so L3 should be in between L1 and L2, to create the loop right? or...
L1 in the middle makes no sense.
L2 in the middle, does make sense. (but only if output L4 is not used) ok. L2 in the middle.
Make the feedback loop from L1 into L2 and back into L1 again.
That seemed to  work before.
fine. but no parallel resonant capacitor placed over L3.

ok, I have L1 and L2 close coupled again, and L3 loose coupled to L2.
L1 impulses are fully absorbed into L2
L2 jumps in voltage at it's maximums.

L3 parallel resonant can make the 28W lamp very bright.

Without Any impulses.

So, L1 is pulsed by the half bridge, at L2 series resonant frequency.
L1 discharges straight into L2 which charges up, and L2 is again charging up L1 by it's magnetic fields.
or something like that.

BUT WHY IS PARALLEL RESONANT L3 POWERING THAT LAMP?

need time... :roll:

ok I want to go back to no parallel resonant coil.

Close couole L1 L2 L3, Load L3

back to basics

L1 recycles its back emf impulse into L2.
this is the bedini principle.
recycling from l1 into l2
re use of electricity


L1 close coupled to L2 takes it one step further. l1 recycling into l2.
L2 feeding back into L1.
a full recycling loop. but still losses from resistance / heat. no amplification

L2 LMD resonant with L3 gives amplification.

L1 into L2 into L3 back into L1 gave challenges, due to phase and polarity.
but it would be the ideal amplification loop. power could be taken out from L2 (low impedance)

impulse polarity was wrong.

But now I see if I tune down, the polarity of the impulse reverses.
still an option to make it work?

Why is L3 amplified with LMD resonance?
why does L3 heat up (power available?)

Right tuning is key.

also, new faster diodes and kool fins came in.
might also use extra cap for power between mosfets. inside of the diodes, from high side drain to lowside source. but not connected to ground.

I might also try with a single switch to create feedback. only high or low side switched.

I doubt it will work though, as half of the period is wasted away.

and what about L3. if it is loaded, voltage and current become in phase. that is amazing, but if L1 is the load to L3 (close coupled) then the L3 current is peaking when the impulse should be generated.

the current of L1 and L3 are then oit of phase. that is impossible. But it might be possible if L1 is TEM and L3 LMD.

I'll just have to try. and observe.
first make L2 a decent mach for the L1 impulse (harmonic), altough that isnt so important as L1 and L2 are not close coupled anymore. But L3 is, and has the same frequency.

somehow, I feel very excited when I think about this feedback loop. It makes sense.
I have had this before....
I learned to be open minded and not expect to much.

first I need to test those new fast diodes. and put a cap in. still wonder if I should groumd the cap. I remember having problems with ground. but impulse (that passes bosy diodes) charges the cap better with ground.

I also should make a full bridge rectifier for L4

Still have my doubts about the adding of the capacitor. The impulse indeed shows up at the drain and source, between the blocking diodes and the mosfets. not clear how much energy it really loses.
impulses passes the body diode, then is blocked. if I would charge up a cap, this cap would also discharge again, when the mosfet opens up again. So when the impulses charges the cap, it looses its momentum, although it is recycled it is not what I need.

I'm also not getting my head around the current polarities. I tested again with all coils not mirrored. then tested again with L2-L4 flipped over.

As L1 close coupled to L2 normally works fine, low power consumption, this whould be a feedback loop, so L1 and L2 magnetic fields assist.
But by adding a L3 coil, which is out of phase with L2, the current should also be out of phase, and when coupled should counter rotate, which drags, and needs much more power. And this is also what I see.

So I flip around L2 and L4. now L2 gets negative impulses from L1, on L2's positive maximum, and vice versa. So the dielectric field, is very temporarily discharged, and quickly charged up again. Now the power draw is less.

but... L2 and L3 should still be out of phase, but they are not!
L3 is parallel resonant, close coupled to L1, and L1 brings it into resonance by magnetic induction.
L2 also brings L3 into resonance, by dielectric induction. but, it seems it isn't strong enough. or else the phase whould have shifted.
 
this must be wrong. I must check polarity of L3, when L2 is far away.

!@#$@#$%#$%^ 
I need to get my head around this...
ok. back up again.
take a step back... chill
relax.
let it settle.
pffffffffffffffffffffffff

back up? me? ehh workaholic!

I like this setup, where L1 L3 are close coupled, and L2-L4 are close coupled. L3 L2 loose coupled. L2-L4 flipped over.

I still can dial in 2 resonant frequencies. Now the lowest TEM has L2 L3 out of phase, and the LMD has L2 L3 in phase.
positive impulse for negative maximum L2/L3 and vice versa.

only thing is, with LMD the impulse polarity on L2 is opposite to the voltage. it still gives power, but it seems to be a lot less.
But, the L3 coil (close coupled to L1) does rise to high voltages/currents.

I probably have to compare both setups, with L2 L4 flipped, and without.

it still is hard to make sense of it.
L2 L3 distance is around 15mm or somewhat more, since I have the bridge wires on the wrong side (lazy me)

I also dont have L4 fully rectified. I should not ground it, and fully rectify it.

I am able to light up the 28W lamp to full brightness, at 225V 124mA DC
which is always awesome to see. but... input is 40 watt.. so not very efficient. still. It is powered by a series resonant impulse coil.

I also believe, by flipping over L2 L4 I only reversed the sine polarity of L2, bu not the impulse polarity, and L3 reacts to the impulses, not so much the L2 sine wave polarity. So flippin over, does not really work?

and with TEM mode, now the l2 l3 are out of phase (before flipping it was in phase), which makes the impulse equal poalrity to L2 maximum, giving great output power. but again at a cost. but this time, I don't need to push in a lot of voltage. 27V gives high brightness, with LMD I need higher voltages 56V.

So this would make flipping wrong... but, efficiency does seem to have improved over not flipping. I really, need to do comparison tests. make a table, and analyse all the data.

finished designing a half bridge gate driver board for the impulsed series resonant primary.
ordered 5 at jlcpcb.com.

If it checks out. these will become available in the near future for fellow researchers.
Fully assembled.

Can also be used to drive a tesla coil, in the way Nikola Tesla intended it.