L1 L2 close coupled, L2 L3 loose coupled, by 25mm nylon plate.
L2 tuned by 50.3nF series capacitance
L3 untuned.
impulse= 722.4 nS , F1 resonance=692kC/s
F3=525.76
L1:L2=9,2 X (harmonic)
L2:L3=6,99 X (harmonic)
L1:L3=0.76 X

This is like a Tesla coil. The L3 is riding on the L2 wave, but also on the L1 impulse (compounded wave).

Picture of the scope is not of this setting, I could not dial in it again, due to heating up of the L2 coil (30A!)
L2:L3 where perfectly tuned. to the 7th harmonic. the sine wave of L3 alsmost didn't show any decrease in amplitude.
But I could not tune it bettwe, by makinf L1 impulse fit L3. This would need to be done, by changing coupling distance between L2 and L3, (using paper sheets).

most importantly, I now cannot dial in LMD resonance, only TEM.
For LMD, L2 needs to be tuned about the L3 first resonant frequency (around 525kC/s)
L2 at that frequency, needs insane high voltages to function properly.

Still, with TEM mode, it was impressive to see L3 resonance so high in voltage, and almost not decline.
There is a phase shift between L2 voltage and L3 voltage, I think that is due to the high frequency of L3.

yellow L2 voltage
green L2 current, tuned to maximum
orange, L3 voltage

L1 inductive spike Impulse is a resonant half wave (picture1)

L1 impulse half wave, is a quarter current wave. This is used to amplify current of the L2 series resonant coil. L2 current is much slower, and zero at the moment of the impulse. quarter wave current from impulse helps start up of magnetic field.

L1 impulse, does not transform into L1 current quarter wave.
But it transforms directly into L2 current wave.
L2 is much slower then L1.

L1 impulse superimposed on L2 voltage maximum gives L2 current a extra boost right at the start.

if impulse is relatively very short in duration it will not cary much energy to assist the L2 current.

L1 and L2 should have a harmonic relation.
if L3 is untuned it needs to be an odd harmonic. which one works best?
5th 7th 9th 11th?

and how do I compare energy levels?
dollard said 7th but why?

also, tem and lmd are combined. both should have some sort of balance right?
L1 is TEM
L3 is LMD
L2 is TEM and LMD?.

Im building 2 new isolated gate drivers. this time recom 20V 0V  - 5V 2W powered.
curious if it won't need batteries.
I'll test them in the half bridge.

If L1 impulse amplifies L2 current
and L2 amplified current is coupled back into L1, then L1 should have less inductance than L2.

if L1=L2 then resistance will eat away
Let's make L1 smaller than L2.
Maybe just a single wire coil.

If take my L1 that is equal size L2, bifilar.
lets hook up the windings of L1 together in parallel. making it a single winding. with low resistance low capacity and lower inductance.

Still able to close couple to L2.

new gate drivers almost finished

new halfbridge isolated gate driver board is almost ready for testing. very happy to not use batteries or pulse transformers anymore.

internal gate resistance of the c2m0080120D:
RG(int) Internal Gate Resistance 3.9 Ω (at  f = 1 MHz, VAC = 25 mV)

before I used a  8.2 ohm for turn on, and a 3.3 for turn off.
8.2  ohm turn on had barely any ringing. 3.3 ohm had a bit more ringing at turn off

now going to use:
4.7 ohm 1W metal oxide for turn off (pin 7) and
6.8 ohm 1W metal oxide for turn on (pin 6)
at the 1EDI60I12AF gate driver IC pin 6,7  to mosfet gate

It works!
some ringing on the gate,  need to increase the turn on resistors. minium is 12V, so mosfet stays on, but it rings above the 25V max gate source voltage.

turn off isn't visible due to the impulse.

tested with 20V over half bridge at 78kC/s resonance. 49% duty
0.24A 0.30A on the symmetrical powersupply, which is series connected dual channel.
the 0.24A is from the low side switch, somehow that one takes less amps. I have seen this before, but most off the time, the high side took less amps.

scope shows gate ringing. probed with 10:1  bit to much ringing for my taste. I'll stick to 8.2 ohm for turn on.

Turn off, I will need to see with a resistive load that doesnt give an inductive spike impulse. like a lamp.

But it works! :D

changed turn on gate resistor (pin 7) from 6.8 to 8.2 ohm 1W
now it's alright. just a small amount over 25V.
Measured 1:1 probe, on low side gate.
turn off ring also is seen here. also no problem there. (4,7 ohm 1W turn off resistor at pin 6)
measured with 20V over half bridge at 78kC/s

So 8. 2 Ohm for turn on and 4,7 ohm for (lower= faster) turn off. is the bare minimum. To be safe, 10 ohm for turn on should be used. (as usually is done)

wow  I expected to much from these stth1210 diodes... already gave out.
I had an impulse, but saw it jitter. was only 200V impulse.
than increased power to amplifiy impulse voltage, and it gave out.
tsssss  :-/

it looked so good on the spec sheet. they aren't even really hot. got a passive cool body and a fan on it.
specs said 12A 1000V Trr 48ns  softness factor 2

I see there also is a STTH3010 with softness factor 1 30A 42ns 1000v hmm. maybe. maybe not

OK, back to the MUR8100. these last ...

edit: softness factor, 1 or 2 is a no go. turn off should be very abrupt

ok got the mur8100E back in and everything is fine.
now I can continue testing, without batteries.
but first.
FOOD

Diodes can be placed in series, would that also work for blocking impulses?

I have 4 equal coils. in pairs of 2 close coupled.

I wanted to test L1 with smaller inductance, so Instead of series connecting its windings, I put them in parallel. Making it from a bifilar coil, into a unifilar coil (both windings become one, for low resisitance).

I kept it close coupled to L2, and tested.
To my suprise, the polarity of the impulse on the L2 voltage sine wave was now reversed some how... negative impuls on positive voltage maximum of L2.

I cant explain this.

I compared the winding direction, and coupling, and both set of coils are the same. very strange.

also the current draw of the parallel windings L1, close coupled to L2, was much higher. This can be explained, by L1 having lower impedance, so there is more current allowed to flow to ground. But it could also indicated, the fields of L1 and L2 are repulsing.
but I cant figure out why or whats going on...

Mercurius is still retrograde, would not suprise me if it somehow is gone if I test again...
MAgic... :wtf:

tested again, still reversed.

Then reverse connected L1, outside rim to half bridge, and inside rim to ground.
Now the impulse polarity was good, but amps of supply went cray high (5A current limited at 2x 6.2V)
conclusion, fields are not opposing.

Why then is the impulse with a bifilar coil positive on the positive voltage maximum L2.
And impulse of unifilar coil is negative on positive voltage maximum L2

what is going on here?!

The impulse is created, by resonant discharge of L1. bifilar or unifilar, should make no difference in polarity. Positive in is negative out, says Lenz law. unifilar is much higher in resonant frequency then bifilar, due to less capacity and inductance. (bifilar has mutual inductance between windings).

L2...  is L2 the cause? Hmm. I can decouple L1 and L2. see if something happens to polarity by coupling...

I fully decoupled L1 and L2. both grounded on outside rim, and inside rim connected to half bridge.

now indeed, polarity is correct.
frequency is lower, due to loss of mutual coupling inductance and capacitance. current draw is also little higher as can be expected.

Conclusion, close coupling L1 L2 with L1 unifilar, and L2 bifilar, changes the L2 resonance polarity!
But WHY?

I know the magnetic fields of the coils, can act as diodes. thats the only thing what could explain it. but L1 and L2 magnetic fields should be assisting ...
Is this frequency depended? ok close couple L1 L2 again, and use difference L2 capacity to tune to different frequencies

changing freqeuncy when close coupled didnt change anything.

I loose coupled L1 to L2. 2,5cm distance.

now, the impulses where shifted to zero volts! current draw was much lower.
This is mighty interesting!

the positive impulse is 400nS, the negative impulse is not fully formed, I suspect this is caused by a faulty mur8100E blocking diode on the lowside. I need to replace that diode.

Scope shows L2 current in green, and L2 voltage in yellow, with L1 impulses on the 0 volts! very cool, as now we have a voltage/ current in phase situation. Power!

need to think this over.

I tried the same as pt4, but now with the bifilar L1, 2,5cm distanced from L2.
now  I could NOT shift the impulse to the zero volt point.

OK, so with the unifilar L1, 2,5cm from L2 (loose coupled):
L1 still produces impulses at switch off (dead time)
But L2, is 90 degrees phase shifted.
instead of being impulsed at V max, it is now impulsed at Imax, Vzero.

series resonant L2 is pushed? out of phase by pulsed L1? it makes no sense.
Still it happens.
This is science. this is research. Now we are learning something new.

A clear difference between the bifilar and unifilar L1 coil.

another thought I have, is the difference in impulse speed. the bifilar coil with its lower resonant frequency (due to larger capacity and inductance) has a slower impulse.
around 800nS

the unifilar impulses is around 400nS

could this have an effect?
when close coupled, is completely phase shifts 180 degrees (L2 resonance)

when loose coupled, it is 90 degrees

when not coupled, no phase shift.

So what can cause this phase shift. The magnetic field? no, as L1 and L2 have attracting magnetic fields. already tested this. (low current on input power vs high when reversed coupled)

what else? the impulse?

Does it make sense, to have a negative impulse on the postive maximum and vice versa, when close coupled?

I tend to let this go, as I dont get it. but it is mighty interesting.

tomorrow my new Pcb's will arrive. these are the new radiant  halfbridge boards. made out of 1 piece for 2 isolated gate drivers mosfets and diodes.
after testing that I probably will have the final design. (with fuses)

the reason the low side diode acts up, is probably over voltage. they can take 1000 V
but I wasn't careful, and due to the low duration my input voltage quickly was to high, and that diode suffered.
it again was the negative impulse that knocked the diode out.

really need to take care of max voltage, as now the avalanche mode of the body diodes do not protect the added mur8100e diodes.

the mosfet is 1200V rated and the mur8100e diode is 1000V rated.
So... maybe I should use 900V SiC mosfets. so they protect the diodes.

its a hassle to change the diodes... already ripped a bit of the pcb copper lead.

good candidate


https://nl.mouser.com/ProductDetail/Wolfspeed-Cree/C3M0065090D?qs=%2Fha2pyFadujzsfT7raD8v7sJKQWPXTvN7IFrDEDiDpnHams%252BSBZJ8w%3D%3D

MOSFET G3 SiC MOSFET 900V, 65mOhm
C3M0065090D
Wolfspeed / Cree

datasheet

only its the C3 version. seems C2 are out.
difference is gate voltage. C2 was 20 - 5
C3 is 15 - 4 V
so I will also need to find a new isolated dcdc  converter
 pfff.
jist did a quick mouser search.... no 12v in - 4 15 V out. but... need to look better..
sleep

pfff
again...
I thought I finished designing.

ok

found an isolated dcdc converter with 12V in and - 5v 0v + 15V outputs.

MGJ1D121505MPC

https://nl.mouser.com/ProductDetail/Murata-Power-Solutions/MGJ1D121505MPC-R13?qs=YlHR7fkrjk93QqlmLCxZwg%3D%3D

https://nl.mouser.com/datasheet/2/281/kdc_mgj1-961560.pdf

no through hole version. only smt
but I believe it works with "pins" so maybe that will make it easier for assembly.

Never ending story. But I really need the mosfet lower than the diode voltage rating else I will keep plowing diodes.
I will get there.

And it is all for YOU
so you can use the final board also

and back to redesigning PCB for new isolated dcdc converter
Oh Wait, the sun is out!

still also thinking about that L1 L2 anomaly.
is that impulse of L1, that collapse of magnetic field energy, able to kick L2 180 degrees out of phase?
what will happen if I destroy the impulse via a fast diode to ground? will it not flip phase?

The old designed PCbs came in yesterday. almost finished building  one. but its still based on the C2 mosfets with the 20 - 5 volts.
I love the blue, and having a single pcb for the whole half bridge

I redesigned the (new)pcb for the murata 15 - 5 volt dc. I will also add a diode and fuse for the 12V side.
these will be ordered when checked.

love the new blue PCB.
testing showd some points were not grounded.
fixed for next design

diodes need an attachment thingy with the coolbody.

plugged the 12v fans in series again. diodes run at 28C very cool!

and again current amplification when L1 and l2 are close coupled, and tuned a bit higher than Fres.

at Fr amps are 7.4A pp
little above Fr amps are 11.2A pp
Nice!

next design hopefully will be definite design. but we'll see....

very happy with this pcb. tiny blue symmetric
I love it
 :kiss: