I would love to have higher voltage pulses. the pulse transformer, will need less windings on the primary. When pulsed with the IGBT, this leads to the IGBT getting hot.

Another way to get the pulses, is to use a joule thief, with a bifilar coil. joule thief, pulses a darlington transistor open and close. this darlington is then hooked up from the collector, to the pulse transformer primary, and the voltage supply positive.
The pulse transformer, is then pulsed, when the collector flows to emitter to ground. In a joule thief circuit, this happens really fast(40khz). But relatively low compared to the resonant frequency of the coils (550khz).

I need to develop this, So I have a pulse thats high voltage.

the resonant frequency of the bifilar coils can then be tuned down, to match the joule thief frequency, by adding capacitance to the coils

the trick is, the secondary needs to have a proper impendance match to the bifilar coils.
In the end i will use 2 coils to pulse the system. parallel(low impedance) or series(high impedance doesnt really matter.

The whole reason, to make high voltage pulses, is that in my experiment, with 12V i produce 730Vdc in 2 caps of 6.3uF in around 1,5 seconds.
The energy stored in the caps, E=1/2 CV*V=1/2 * 0.00000315 *(730*730)=0,84 Joule. Thats not much. but it is dependent on the V*V (not linear).

12v to 730V is around 60 times.
With a 19V pulse it would give 1156V, with the same caps this would give 2,10 joules.
50V gives 3000V would store 14,18 joule
100V gives 6000V  into 56,7 joule
200V .... 12000V.... 226,8 joules
400V ... 24000V ....907,2 joules 
At this time, a differnt capacitor would be needed, but we wouldnt have to charge the caps completely full...

So, producing a high voltage pulse is stepping up the energy big time, due to the square voltage factor

using a joule thief pulse, to pulse a step up transformer(discharge a capacitor in it), giving high voltage pulses, would be the way to go.

And also... I could use even more coils.

If I use 1 pulsed bifilar coil, it has a impedance of around 0,5mH, 2 bifilar coils in parallel are 0,38mH

the secondary of the step up pulse transformer has to be 0,5mH to transfer the power. (impedance matching).
use radio frequnecy ferrite for the pulse transformer, so it can handle the fast spikes. (or find a proper transformer, with the right impedance and ratio)

the secondary should have, a lot more turns then the primary(to step up the voltage of the pulses).
 So a single wire wound coil (not bifilar) would be better for the secondary(needs more turns to get the impedance).

The primary, should have a few turns(but enough). this coil could be bifilar, also better for the b emf of the primary system.
The current through the bifilar primary, is triggerd by a darlington NPN. So it will get hot, make some cooling.

The joule thief, needs its own bifilar coil, to trigger the gate of the darlington. the darlington and the bifilar coil, set the speed of the switching/pulsing of the step up transformer. the highest speed should be the goal. (the resonant coils will need to tune down to this frequency)

Maybe, the step up pulse transformer, could have its own power supply (from the system? getting higher voltage each time?) and the joulethief circuit have its own power supply (already works with 1,5V, it only serves to open and close the darlington fast)

to get better results, a IGBT based joule thief could be better

I made a inie minie tiny bifilar coil today. to see If I can get the frequency up on the joulethief.

Finaly the sun was out, so spent some time with friends outdoors.
Sweet :cool:

i want to use a hollow ferrite core. to allow the spinning. There was a AM radio guy,  who showed way better reception.  his work was based on spinning fields.

I cant get the voltage high enough up, with the joule thief circuit, to get high enough pulses, from the step up transformer.
So I wil have to build Nelson Rocha's circuit, or switch to capacitor discharges.

Nelson Rocha's schematic, using a stepup tranformer (a 230V to 5V in reverse)

(copied from matt watts)

I must say, this is frustrating.
could it be so hard to make high frequency high voltage hairpin pulses?

If I turn off a tranformator, it gives a pulse from the b emf. this is exact why i need. but in a high frequency rate.
A transformer based circuit like nelson rocha's might work.

Tesla's hair pin might work, but I like to stay away from spark gaps.

A discharge of a capacitor also gives the perfect energetic pulse. but I need a highfrequency one. so the cap needs the charge fast, in between discharges.

Maybe a large capacitor that doesnt discharge completely, charging a smaller cap that does discharge.
Is that what C1 C5 and D1 is about? discharging into T2

I wonder why C3 and C4 are reversed in polarity?

Decided to build the nelson rocha circuit. bought me a transformer, 230V-6V I can take out the laminated core if needed.

Build and tested it.
output of transformer shows sinewaves instead of pulses. but high voltage is there.
Suspect iron core of transformer to be to slow. will remove it, and try without core, to see if sine wave transformes into pulses.


Spikes on bifilar seems clipped. (could be due to diodes). will also try another bifilar.

used a darlington ESM 2030DV

used a 4,7nF 100V rating capacitor instead of 300V.

Now works around 4,5Volts DC input. Higher voltages spike the current, and the output is fuzzy. a high frequency noise is heard. (as in some of nelson Rocha' s videos.

OK, made a new transformer, didnt have a toroid, so made it from a ferrite rod, with 2x 39 turns on it. the pulses at the collector look sharp now, undistorted.

Also removed the iron core, and placed a ferrite rod in there. I can re-use the iron core

upped the voltage to 5,47V 0,36A

I now see a 500V sine. still no pulses, like a hairpin

oops my 2200uF cap wasn't connected....

The base pulses get chopped up clipped above 25V, they want to go higher, but cant, and then the energy is prolonged. (widened).
The caps used at the base are 25V and 35V rating, so I will try higher voltage rating, so the pulses stay spiking

ran out of luck with the capacitors. no more high voltage 100uF will order some higher voltage ratings

If the pulses at the base aren't spiked, the collector- emitter stays connected to long

at the collector, with 7vdc the spikes are 250v (so the caps at the base would be to)

or....  should i try with a toroid?  havent got one. will order both

Order came in, worked all day. Made 2 toroids, last one with more windings, thinner wire.

I got the spikes, including a high pitich sound. above a certain voltage (input) the noise became erratic. the signal, then also became erratic..
My conclusion, the transformer, is not handling the high voltage, and sparks over. It also might be to big.
Around 5 V 0,3A (i remember) the spikes where pretty good, but at the cross over point there also was a lower voltgage sine(transformer capacitance to big?). the spikes also went both ways, So maybe 2 extra ulfra fast diode, to rectify.

I made a lot of mistakes, but corrected them. and with result. Not there yet, but it starts looking good.
I know a transformer that can handle 5kV, completely potted. so, ill buy that one, and swap it in.

aaaarrgg frustration!!!
or in other words, it didnt work with the second tranformer rated at 5kV.
But Matt Watts seems to have succes, so I will look at his tech to achieve hairpin pulses.

Refusing to give up.
Checked resistance of the trimpot. It was toast.  replaced it with a new 10K, with a 5K parallel over it, and a 100 ohm in series, so it would never be 0 ohm.

Tested with the old transformer, and It produced around 600V sine wave on the output, without the rectifier. connected a neon light, with one leg and it lid up.

connected the center tap of the output to earth, and the core.

3,5Vdc input and around 0,2V A

again the high pitch noise.

I made this circuit messy as hell. and since its higgh frequency, the design should matter (rebuild it properly).
I also use thin wire, might also be a factor.

If the effect occurs at 7Vdc 0.3A thats still 2,3W with high voltage pulses.

Aslo C4 seems to be wrong in polarity. It got hot, and I reversed it. It processes pulses to the gate.

Never give up. never. reach for the top
 :yodel:

the squealing sound, might be prevented if I pot the toroid and the transformer. I just cant hear where its coming from.

The hairpin circuit is still a option.
replace the lamps, with a bifilar coil, inbetween the capacitors. maybe without the U bar.
what worries me, with this concept, is the sparkgap. I dont like sparkgaps. I would like to close it fast and open it fast. with no connection to ground or other points. So the caps will discharge through the bifilar coil.

my igbt can switch 600V

the rocha circuit can produce high voltage dc when the transformer is rectified. I could use this as a high voltagesource to charge 2 capacitor with a series bifilar coil in between.

then discharge the capacitors with the igbt.
the frequency is high. so the caps should be small.
at the same time it should have a relation to the capacitance of the bifilar coil.

I guess its about trying differnt capacitors. and see the results (at low voltage) when there is a match, i can switch to highvoltage.

the problem might be that the caps discharge to ground with the igbt. still the charge alao needs to move between the capacitors where the coil is. but will it be a one way direction with a magnetic field or not?

i have seen the pulses at the bas and collector of a joulethief.

could i also use a pnp together with the npn somehow to switch both pulses?

rough sketch
made with my fingers on my phone

Posted: May 13th, 2017, 06:42 AM

positive and minus connected to joulethief.
and 2 more capacitors needed from ground (between npn and pnp) and positive.