The basic solid state hairpin circuit is explained here:

https://youtu.be/1Flj1i0zQ-8
see attached schematic

I want to advance it.

Since L3 only gets one impulse from L2 per period, Why not also pulse from the other side of L3, with a second L2 coil?

lets call them L2a and L2b. The impulses they create, would be alternating...  one (a) at the maximum, and one (b) at the miniumum (but seen from the other side of L3 this is actually a maximum also).
See attached picture.

this would require a single pulse generator, that drives the 2 separate L2 systems. This might be solved by Matts isolated gate drive circuit, or by a pulse transformer. I already made pulse transformers, and I'm not a fan of them, so I want to give the opto isolator a try.

I'm going to use 1EDI60I12AF as gate drivers. these are isolated, and need 2 power supplies.
Thanks Matt for advising to use these.

the specs of these are amazing, and it will be the first time I will use smd components.

part link, where data sheet can be found
https://www.mouser.com/Semiconductors/Power-Management-ICs/Gate-Drivers/_/N-41fcp?Keyword=1EDI60I12AF&FS=True

First time I soldered smd.
Wasn't that hard.

Good job Evo!   You must have far better eyes than I do--without 4x magnification, forget about it.

nice job,  that was super satisfying first time i did it too, -  concentration is heaps higher than my usual style of beer driven electronics.

Thanks guys! :thumbsup:

Also decided to step up the voltage capabilities to 900V by using 2 diodes in series and this Sic mosfet:
C3M0120090D

mosfet info page with datasheet

also ordered higher voltage tuning caps (1nf wima fkp1 2000v)
It doesn't seem to be needed but I rather play it safe.
the caps haven't changed their values although I have been crossing over the voltage threshold several times with many hundreds of volts. I think this is due to the difference in quality of the negative back emf.

still waiting on the final parts.

the back emf is limited in voltage by the mosfet. the SiC sets the maximum around 900V
but... would it be possible to step it up, with another coil, just there to capture the back emf and step up its voltage?
yes but then it needs to be injected into a series resonant bifilar coil. thats a bit harder.

or not. basically, this is done by coupling l1 and l2, but making l2 step up l1. hmm...

L1 is basically only there to provide back emf. this means, high inductance low capacitance and low resistance would be best. Not a bifilar, but a pure inductive coil. Single distanced thick windings.
a single pancake coil with distanced windings, would be perfect.

L1 would demand high voltage, as the current/inductance would be lower probably.

pressure is needed. bemf impulse pressure. the faster the better, the higher in negative voltage the better.

just ideas...

I changed the function of c1 and c2 around. this also means the L2 is connected in reverse, outside rim (grounded) on C2

It simply makes a more powerfull impulse in L2:

- C2/L2 connection (outside rim L2) gets to see the back emf impulse. giving a maximum negative voltage.
- While the C1/L2 connection (inside rim L2) is being resonant, giving a maximum positive voltage.

this outside negative with inside positive results in a bigger voltage difference over the whole of L2, during the impulse.

I also changed the negative DC offset connection to C2/L2

The negative impulse, now reaches a quicker maximum, the 500V maximum of the IRFP460 makes it impossible to step up the voltage higher, as the body diode starts avalance and cuts the voltage off.
To avoid this, I slightly slowed down the impulse, by adding C5 parallel over L2, 700pF.  Its small capacitance is charged up by the back emf, and this adds some time to the discharge, making the impulse wider, and lower in voltage.

Now the back emf impulse voltage is lower (and loinger in duration), I can Increase the current and voltage of the dc power supply again. resulting in a higher voltage resonant sine in L2

I added the changes to the picture below in RED

The resonant sine of L2 at C1/L2 now doesn't show the impulse anymore, but is a flawless sine wave.
The circuit still works the same and can light a 28W lightbulb,
But it does look, like it needs less power for equal light. I tested this without proper tuning. So when tuned better, I should be able to see how much more...

Another thing could be using a 900V SiC mosfet, and doubling the voltage of D1 by putting 2 diodes is series. But I havent done this yet, because the capacitors also need to be stepped up, to deal with these increased voltages

Here we go again.
I decided to put a fan on the mosfets.
the gate drive circuit will be on its own pcb, raised above the main radiant citcuit.
Already love it!

I was looking at a video of a smd soldering microscope comparison, and looked at the price of the best one. 280 euro on aliexpress. And probably more for custums duty tax.

Then I realized... My phone has a decent camera.
made some pictures with and without digital zoom.
pretty impressive. So for occasional smd work, Ill just use the phone (or a lense).

pictures says all, thats my pinky finger for size

been playing with the layout. These caps are BIG!
think this will be the layout. center green is low voltage gate driver circuit. it will be elevated.

to the right I added a buck converter that can handle 36V, it outputs 12V so I only need one external powersupply, (no battery needed) as I will also use a 12V to 15V isolated boost converter, thats the only part still in the mail.

Did several test with the new radiant circuit v2.0 much more power this way. But it als draws more power, low voltage is enough.

but for the buck converter to work it needs at least 13.5V to generate 12.

with room for the 12v to 15v isolated boost converter

bottom of the main board has no copper pads. I avoid  ground planes, as I dont want the impulse to loose voltage

I soldered the first gate driver for testing. I made the gate resistor very small 1.5 ohm, so its limited to 10A.
But it might oscillate a lot.

12v to 15v isolated power supplies came in today. no duty tax. tested them. can be regulated by changing input voltage. Gave the 15 some extra capacitors to really be able to give a 10A punch to the gate. 100nF ceramic, 1uf polyester and 22nf polypropylene. together 1.122uF

still got rome for an elco. but hope I dont need it

input only has a 22nF and the 220uF from the buck converter (connects to main power supply max 36 V).

hmm...
first I forgot hooking up the - input signal to ground of the IC.
Then I hooked it up correctly, and the signal on the input side looks good.
Voltages on the input and output side are correct 10V and 14,5V  well within range.
mosfet checks out.
 all connections are good.
 
But no switching signal at the gate.
Can't find anything wrong.

only thing left is the IC is dead... but I see no reason why it would have died...
I have not socketed it, so... hmm... build another one? this time,... with socket
 :-/

Slow and careful Evo.  Don't rush anything, start with low power and gently work up as long as everything checks out along the way.

Hopefully you didn't smoke the chip.

Quote from Matt Watts on May 22nd, 2019, 05:17 AM

Slow and careful Evo.  Don't rush anything, start with low power and gently work up as long as everything checks out along the way.

Hopefully you didn't smoke the chip.

I hope so. Made another one, this time on a socket. voltages check out. connections are good.
I do see some kind of signal on the output side, but it is as if its  turned off, realy fast. As if its in protect mode.
These chips do have a protection build in, so I'll need to read that part again in the specs.

I looked into the specs (attached) of the 1EDI06I12AF once more, to see why it isn't working (yet).

first, Vcc1 for me is 10V right now.
and Vcc2=14,5V. those values are good.

from table 3 operating values:
Vcc1= 3.1V to 17V (10V now)
Vcc2= 13V to 35V  (14.5V no load, so could drop and trigger under voltage lock out, UVLO Vcc2=12.7V minimum?)
Logic input voltages (IN+,IN-) VLogicIN = -0.3 17 V

The input signal (table 5) needs to be
Vin low= maximum 1.5V and 30% of Vcc1 (3V)  (0V)
Vin high=minimal 3.5V and 70% of Vcc1 (7V)

From this... I conclude, I cant just use the logic level inputs. they need to be a minimum of 3.5V (need to check this)
But also... the Vcc1 needs to be much lower as Vin high needs to be at least 70% of Vcc1.
if 3.5V =70% then Vcc max=5V.
A higher voltage vcc wont work .

Solution is, not use the cmos outputs at 3.5V.  keep running Vcc1 at 10V.  and use the inverted inputs, to get the signals 180 degrees out of phase.
much easier also...

Vin has a max of 17V. need to check that. but also a min of -0.3V  need to check that also. I have seen the pulse gen signal mirrored on the negative side. so, I will limit this (zener, or straight diode to Vcc-)

Enough for now...

my square wave generator gives a +5V and -5V output relative to ground. together 10V... but.... relative to ground.
I think I can fix it, by not using the ground of the pulse gen, and using ground from Vcc1

this creates a problem, with the inverted signal, that needs to be referenced to Vcc1 high. cant have both. as Vcc is the same.

maybe a solution is to use the cmos out, and use a transistor to raise the voltage to Vin.

the generator has 2 cmos outputs, both 3.5V (ttl) referance to ground (positive) To use these, I would need a switch, for turning the system of (Or i need to turn the whole pulse generator of, and it takes time to start up.
But more important, I need to change Vcc1 to 5V. 

Vcc1 is now determined by the buckconverter 10V output, that is set to 10v so the isolated boost converter gives 14.5V unloaded.
So this 10V (that might be raised to 12V when the load is on) needs to be converted to 5V....
First tought is another buck converter, giving 5V. But since there is almost no current of Vcc, I might do a resistor voltage divider.
Power dissipation (Input side) PD, IN = 25 mW  (table 2 absolute maximums)

2x 180 Ohm with cap over it.
both 5V=0.027A and 0.139W per resistor

with 12V its 2x 6V
6v=0.0333A  and 0.2W 

add a 5V zener maybe, to deal with the extra 1V?

or a lm317

So why not use a good old LM7805 to get 5VDC?
https://stak.com/7800_Series_Linear_Voltage_Regulators__1.5A_Voltage_Regulator__Various_Pack_Sizes_Available_to_save_you_money_

Quote from Lynx on May 23rd, 2019, 08:00 AM

So why not use a good old LM7805 to get 5VDC?
https://stak.com/7800_Series_Linear_Voltage_Regulators__1.5A_Voltage_Regulator__Various_Pack_Sizes_Available_to_save_you_money_

yes! confused the lm317 with it.
even easier

thanks :thumbsup2:

Placed the 78L05 in with a 0.1uF over the 5V output, and voltages check out. now for the signal to the input... hope I dont lose to much voltage, with the bnc signal wire I bought.
And enough power left on the 100mA 78L05, for a low current red led, to see its working.

YES! measured a nice 15V square wave on the gate terminal. both work, also with the bnc!
Feels good. reached another goal. wanted to build this for years.
First time I tought of this, is when I wanted to make 2 similar tesla coils, and make them out of phase.
Now I can.

Now put a mosfet in and see how switches, then with a resistive load, then with a coil.
but first a well deserved brake.

as expected I needed to raise the voltage when the gate driver was connected to the mosfet.
tuned it to vcc1= 11.61V giving vcc2=14.92V just below the 15v zener threshold.

signals look good. nothing getting hot.
isolated boost converter is slightly warm.

so far so good.

Can easily produce a -900V impulse with a Sic mosfet, with very low power.
Next step, Incorporate it in the new radiant circuit.
I will set it up so it can handle the higher voltages (series diodes better caps).
I also got some silver plated copper wire 1 mm2 to hook it up.
extra low resistance for the impulses.

This is working out rather good! Faster than I expected. but not there yet. fingers crossed

Beautiful, 2 working gate drivers. I like where this is going.