advancing the Solid state Tesla hairpin circuit


evostars

gate driver circuit
« Reply #27, on May 24th, 2019, 03:15 PM »Last edited on May 25th, 2019, 08:25 AM
First time I worked with EasyEDA, so did my best...
here is the schematic : https://easyeda.com/MasterIvo/high-side-gate-driver-for-n-mosfet

the 12V buck converter, is powered from the VCC of the mosfet:

VCC mosfet is the main power input min 13,5V - max 36V
into  buck converter (2A) making 12V
12V into 78L05 (0.1A) for 5V VCC1
12V into isolated boost converter (2W altough 3W would be better) for VCC2

12V is also used for the cooling fan of the mosfet radiator

edit: did several corrections to the schematic

 Schematic.png - 56.56 kB, 999x431, viewed 139 times.


evostars

Oil bath
« Reply #28, on May 25th, 2019, 04:40 AM »
since I work with dielectric fields, and air isn't the best dielectric medium, I want to use oil.

I tried Latex as a dielectric but it was much to hard to work with, and it didn't stick to the coils.

I'm thinking of using a bin where I can put the coils in, fill it up with oil and put a lid on it. the Lid could hold the circuit (although it could also be submerged, I preffer not to).

But maybe its to soon...
baby steps...
first finish the circuits.

I'll probably need to make new (equal) coils. Those would need to fit the basket.

evostars

Re: advancing the Solid state Tesla hairpin circuit
« Reply #29, on May 25th, 2019, 08:06 AM »
made an interesting mistake.
I felt a bit weak, So I hooked up the coils without a load. I decided to let it run for a while (feel fine already).
when I turned off the system, I saw the pulse generator was powered by a wall socket, instead of the high side battery.
It worked fine! the wall socket power supply must have some how isolated it.




evostars

first test successful
« Reply #33, on May 26th, 2019, 05:32 AM »
Finished the first circuit and tested it.
easily could produce 800V impulses, and off scale sine wave in l2.

super cool.

next up. building the second circuit

 20190526_141712.jpg - 1989.99 kB, 3264x1836, viewed 89 times.


evostars

second circuit tested and working
« Reply #34, on May 26th, 2019, 10:57 AM »
Fully assembled second circuit, and tested it.
First something shorted out, so I had dinner, checked everything, and nothing found. hooked it up, and it works. funny how these things work.

So, now I have 2 equal working circuits. Time to hook up the coils... need at least 4 coils lets see what I have got...

evostars

double system working together
« Reply #35, on May 26th, 2019, 11:33 AM »
Found 3 equal coils and one Phi ratio, that I used for L1 (higher inductance, bigger back emf)

L2 coils work in unison.  out of phase, as I'm driving them from my cmos ttl output from the square wave generator, that are out of phase.

impulses alternating. one each half period.

Loose coupled both the L2 coils (L1's separated) stacked them with 4.5cm distance.

When the l2's are counter rotated, they draw much more amps, and produce much bigger resonant voltages.
When the L2's are not mirrored/counter rotated  voltages dont really change.

Need to check connections in coils.

BUT ITS WORKING!!!  :clap: :clap2: :bliss:

Time for a break.

evostars

Re: advancing the Solid state Tesla hairpin circuit
« Reply #36, on May 27th, 2019, 08:30 AM »
just did another quick test. Really interesting.    when the 2 L2 coils with their alternating impulses are loose coupled, the resonant voltage goes way up in both coils, as expected.

But also, they change phase! very peculiar. as if the pressure is to much. and pushes the fields the other way... dont know precisely whats happening yet.  but very interesting.

Lynx

Re: advancing the Solid state Tesla hairpin circuit
« Reply #37, on May 27th, 2019, 10:32 AM »
Quote from evostars on May 27th, 2019, 08:30 AM
But also, they change phase! very peculiar. as if the pressure is to much. and pushes the fields the other way... dont know precisely whats happening yet.  but very interesting.
Break on through to the other side :thumbsup2:
Find out what gives, then show the rest of us how to replicate :-D


evostars

Re: advancing the Solid state Tesla hairpin circuit
« Reply #39, on May 27th, 2019, 03:48 PM »
using the 2 out of phase cmos ttl outouts for the 2 circuits left me without a switch.

made a nice board with bnc connectors and a dpdt switch.

also got zapped AGAIN across both hands. really need protection. If I'm going to work at 1000V impulses, the zap is 1 joule, within a fraction of a second.
 :emperor:
bad. very bad.

 will install bleeder resistors or some sort of safety discharge thing, to stay healthy.

 20190528_004107.jpg - 1641.78 kB, 3264x1836, viewed 81 times.


evostars

Re: advancing the Solid state Tesla hairpin circuit
« Reply #40, on May 28th, 2019, 06:58 AM »
If I put a 2W 1Mohm resistor over the 2.04 uF cap, it would burn 1.3W or something like that. Thats not acceptable.

Instead to make it safe, I might need to make a discharge switch, so I can discharge the caps over a resistor. Something that will always work, and with some kind of back up plan, so I definitely know I'm safe.

or... as I always have done... stay at (relative) low voltage. 500V might be enough

evostars

Re: advancing the Solid state Tesla hairpin circuit
« Reply #41, on May 28th, 2019, 10:08 AM »
did some more testing with a single circuit, installed a discharge point for my power resistors, to discharge the caps. didnt use bleed resistors (yet).

installed a IRFP460A mosfet, without heatsink. Bad Idea It overheated soon, so installed a new one on the heatsink again. it really needs to be cooled (when the circuit draws amps).

Checked the power supply voltage, and it had a big ripple, so I connected the negative to ground, and ripple is gone. So it really needs a ground (I now use 1000uF for the input voltage, could be higher).

Also noticed, the resonant sine of L2 has changed phase, due to the new (reversed) functions of its adjoining caps. This makes sense. The resonance now takes place at the V+ side.
This also means, the back EMF, does not need to be delayed anymore (as with the previous version). It can be directly when the L1 coil is shut off (not just before turn on). So the diode function off the coil is still available, but not used.

Also added a 4cm distanced L3 coil (not tuned) on top of L2,  and connected a 5W 12V lamp to it. when L2 was tuned proper, it did light up. So thats working ok again.

I now use 60nF to tune L2.
Since I want maximum current, I want to play with the 60nF and see If I can get to the frequency that gives the highest current in L2 and L3. I need to load it, as my powersupply is now minimal 13V due to the minimum voltage needed for the boost converter.

The Idea is, the resonant L2 gets its power not only by the switching power supply, but also by the back emf impulse. and both should be matched for best results. Not completely sure about this, and that' s why I want to test it.

another thing I added, is 500pF across L1 so the impulse is a little bit slower, and more controlable (if its to fast, the circuit cant handle it).

standby power is now around 0.17A 13.5V=2.3W  This is due to the boost converters, and the cooling fan.


evostars

crazy amounts of energy
« Reply #42, on May 28th, 2019, 01:59 PM »Last edited on May 28th, 2019, 02:50 PM
Just to give an Idea of what this circuit produces:
I hooked up L1 and L2  no L3 coil.
L1 and L2 are not coupled.

I am measuring the series resonant L2.  voltage, current, and impulse.
The series resonant tuning capacitor is 60nF,
Fres=45.46kHz
DC power input: 1.8A 13.0V =23.4W

The peak to peak voltage in L2 (resonant side) is 972V (yellow 10:1 curve newfile1)
the peak to peak current is 18A (blue curve 10mV/A) EDIT:probed wrong, current is somewhat lower (voltage should be leading not current)
the impulse is -500V (yellow newfile3, blue is still current at 10mV/A)

The impulse is on the other side of the coil, so the total voltage difference(during the impulse, 1uS) over the coil is:
1/2 972=486+ 500=986V

These are impressive numbers to me, but, this is series resonant energy. and 21W is pretty much.
The real cool stuff happens, when L3 is loose coupled to L2 and is tuned to parallel resonate.

It would make sense that the more capacitance is series resonating with L2, the more current is needed to charge it up, and the more energy is resonating.
The question is,  how is it balanced with the impulse presented at the other end of the coil. My best guess is the 60nF is to big right now, it needs to be smaller, to become a better match for the -500V impulse.
But not to small, because when L3 is added, the frequency jumps higher. and I dont want it to high.

So what Am I looking for? How does this balance between impulse and series resonance manifest?
Lowest DC current draw, vs highest resonant voltage/current. This is do able. I'll just make a list with different values, and see if there is a logic to it.

 Newfile1.jpeg - 99.17 kB, 800x480, viewed 71 times.

 Newfile3.jpeg - 94.37 kB, 800x480, viewed 70 times.


evostars

Re: advancing the Solid state Tesla hairpin circuit
« Reply #43, on May 28th, 2019, 03:01 PM »
changed the series resonant capacitor from 60nF to 30nF and correctly probed the current.
I expected lower DC input current, and lower resonant voltage. but, both went up. while the impulse went down in voltage.
Is this larger voltage (offscale guessing around 1400Vpp) in L2 due to the better balance between the impulse and resonance?

C2         Adc    Vdc     Fres     Vimp   Vppl2     Appl2
60nF     1.81    13.0    45.44    -510    984       18.2
30nF     2.13    13.0    63.54    -410   1400?      16.2

need to test more tomorrow, and see If I can make a curve. Probably will need to load L2 with L3 with a lamp. To keep the voltages in line.

Note to self:
Quote
Arrow on current probe should point in direction of ground
, away from positive.

evostars

power supply ripple
« Reply #44, on May 29th, 2019, 05:15 AM »
My power supply on the board now has a ripple of 8Vpp thats way to much, going to increase capacitance. with a poly propylene cap.

evostars

Re: advancing the Solid state Tesla hairpin circuit
« Reply #45, on May 29th, 2019, 06:00 AM »
added a 1uF mpk10 but didn't do it still 8Vpp. Must be due to the boost converters.
Really need to increase capacitance a lot. find some room on the board. Since its so high frequency elco's dont work.


evostars

dipping below zero
« Reply #47, on May 29th, 2019, 06:46 AM »
interesting to see the L2 resonant sine dips below the zero voltage, when the resonant voltage really goes high
The DC offset created by the back emf isn't enough.
no implications as far as I see right now

 latest.png - 38.53 kB, 800x480, viewed 61 times.


patrick1

Re: advancing the Solid state Tesla hairpin circuit
« Reply #48, on May 29th, 2019, 09:29 AM »
Quote from evostars on May 25th, 2019, 08:06 AM
made an interesting mistake.
I felt a bit weak, So I hooked up the coils without a load. I decided to let it run for a while (feel fine already).
when I turned off the system, I saw the pulse generator was powered by a wall socket, instead of the high side battery.
It worked fine! the wall socket power supply must have some how isolated it.
wow facinating, do you mean your bifilar pancake coils ?

evostars

checking the ripple
« Reply #49, on May 29th, 2019, 10:55 AM »Last edited on May 29th, 2019, 11:11 AM
checking the ripple:
removed L2. removed Fan (from 12V)

standby power:
 60mA at 15V
no ripple

turned on switching L1 at 63.5kHz:
83mA at 15V  .
ripple is 1,5V but with a 4V initial spike.  (probably from gate charging?)

now I add the 12V fan:
standby:
192mA @ 15V
no ripple

turned on switching L1 at 63.5kHz:
211mA at 15V
1.5V ripple  4V spike (see attached scope 1:1 probed at V+)

EDIT:
Spike is from back emf at ground. Is the ground good enough?
Edit2:
YEP ground. when directly connected to ground at L1 ripple is: 200mV Big difference. Spike keeps being there at around 3,7V as expected.


checked with L2, and works also. the circuit just behaves different with the c1 and c2 reversed in functionality



 Newfile4.jpeg - 100.48 kB, 800x480, viewed 60 times.

 schematic2.jpg - 919.91 kB, 3264x1836, viewed 86 times.