mini tesla (hopefully not a fail)

[bogdan]

Hello everyone, after a fail with my attempt to make a big tesla coil, i decided to start smaller, so i built a mini version. However it bothers me that the secondary coil resonates at @835kHz, that high number scares me a little so i stopped to ask a few questions. Is that frequency a bit high ? can i reduce it ? and what driver do you recommend for the primary to deal with such high frequency ?

[Zipdox]

No that resonance frequency isn't that high at all for a small coil.

[bogdan]

So what driver should i use ? i tried it with a 555, but it's not acurate at all at that frequency, i even tried it with an arduino, but the square wave...is not a square wave starting from 500kHz.

[Zipdox]

So what driver should i use ? i tried it with a 555, but it's not acurate at all at that frequency, i even tried it with an arduino, but the square wave...is not a square wave starting from 500kHz.

Using a fixed frequency oscillator is not a good idea. You should be using a driver that takes some kind of feedback.

[bogdan]

Well, i tried the  half-bridge steve ward, but the mosfet drivers just blew up in my face, and after a lot of troubleshooting i could not find anything so i abandoned the project...

[Zipdox]

Well, i tried the  half-bridge steve ward, but the mosfet drivers just blew up in my face, and after a lot of troubleshooting i could not find anything so i abandoned the project...

Schematic of what you built?

[bogdan]

here it is http://kaizerpowerelectronics.dk/tesla-coils/kaizer-sstc-i/

[davekni]

At 835kHz, gate driver chip failure seems likely.  The gate charge of a pair of IRFP460 FETs repeated at 835kHz likely translates to more power dissipation than the gate driver chips can dissipate.

SiC FETs typically have lower gate charge and switch quickly.  That's one option.  Another (or in combination) is to add discrete FET buffers to the gate driver outputs as in many DRSSTC designs, with heat sinks on the buffer FETs.

[bogdan]

Uhm...and in English please ? what does that mean for me ?

[davekni]

Sorry, I hadn't paid attention to this topic being in the beginners section.

As you have measured, small designs generally resonate at high frequencies, 835kHz in your case.  Drive electronics are more difficult at high frequencies, especially at high power.  Perhaps a Slayer oscillator running from 24Vdc would be a good starting point for a tiny coil.  I don't have personal experience with Slayer oscillators.  There are some examples on this forum.  Slayer oscillators use a single power transistor and use the secondary ground-return to drive the gate.

Another option for a small coil is to wind it with MANY turns of fine wire.  That's not easy either, but does reduce frequency.

Good luck!

[bogdan]

but a slayer will not give me any sparks, or very small ones, i want to have at least o couple of centimeters with this coil. If i want to make more turns, that means thinner wire, is it a bad thing ?

[davekni]

People do get sparks from Slayer driven Tesla coils.  I'd guess that 2cm is achievable, but don't have personal experience.  High frequency can make sparks at lower voltages.

Yes, fine wire does have a down-side.  The series resistance is higher, so secondary Q lower.  However, if you can get the frequency low enough to drive with a normal half-bridge, that may give better performance.

For half-bridge drive, adding a 555 to enable it for short times with longer gaps (low duty cycle) will allow gate-driver chips to survive high frequencies.  That might work even up to 835kHz of your existing coil, but will help even if you rewind to get down to 400kHz or 500kHz.  KaizerPowerElectronics web access isn't working for me at the moment, so I can't double check if enable modulation was already part of the schematic you tried or if there is a similar schematic including such.

[bogdan]

ok, i will look into the thing called slayer on steroids and see what results will i get, i will post them later on, hopefully i get enough time in this weekend. Thank you.

[davekni]

The "slayer on steroids" I've seen uses a gate-driver IC, so may have the same overheating issue.

Is a larger secondary diameter possible within your definition of a small coil?  Doubling the diameter will almost halve the frequency using the same wire gauge.

[bogdan]

i can test the circuit on my bigger coil, the frequency of that one is @141kHz, and if everything is ok, i can test the small one.

[bogdan]

So, i built this slayer on steroids circuit, and made a few tests, but, i don't seem to be able to tune it properly, i am posting some waveforms with oscilloscope directly on the coil, and with the power source set at 5V. The current draw is about 2A for all capacitors, except for the test run with no capacitor. Each photo has the capacitor near it. any idea is welcomed.

[davekni]

To be helpful, we need the specific schematic you implemented (schematic image or web link to the schematic) and an image or two of your setup showing where the antenna is placed and wired to the slayer oscillator.

[bogdan]

I hope you can tell from these

[davekni]

Have you tried reversing the primary coil leads?  Usually one polarity works better, and the opposite polarity may not work at all.

What value is R1?  I can't quite read the color bands in the image.

Oscillation without a capacitor was 213kHz, above your expected 141kHz secondary resonance frequency.  With the smallest capacitor you tried, 0.22uF, frequency was too low, 96kHz.  That implies a smaller capacitor would get closer to 141kHz, around 0.1uF or perhaps a bit lower.

Are the four IRFP140 FETs wired in parallel?  That adds up to more load than TC4426 is designed to drive.  May still work at 141kHz, but definitely won't at 835kHz.  A scope image of the gate waveform would be helpful, together with a drain waveform with the same configuration.

[bogdan]

"Have you tried reversing the primary coil leads?  Usually one polarity works better, and the opposite polarity may not work at all."
yes, i seem to get the best results this way

"What value is R1?"
8.9k, it was the one nearest to my hand

"Oscillation without a capacitor was 213kHz, above your expected 141kHz secondary resonance frequency.  With the smallest capacitor you tried, 0.22uF, frequency was too low, 96kHz.  That implies a smaller capacitor would get closer to 141kHz, around 0.1uF or perhaps a bit lower."
now this is strange to me as well, the 141kHz i got with a small capacitor, which sadly is gone. Could that be the reason? and the actual frequency of my coil is actually @210kHz ?? the test i made with the following capacitor shows different results, the capacitor is new, and also the tests are being made with a 12v power supply, also i get around 1cm of starting spark, and i can stretch it at @3cm. If i go lower than 100nF, either my caps are busted or something is wrong because they get really really hot, i will try to find some new ones.

"Are the four IRFP140 FETs wired in parallel?"
yes they are.

[davekni]

Yes, many capacitor types overheat with high-frequency AC use.  The polypropylene MKP10 0.22uF part should be good.  Do you have 4 or 5 of them?  Soldering a string of capacitors in series reduces capacitance.  Four 0.22uF caps in series will be 0.22uF / 4 = 0.055uF.

Thank you for sharing gate and drain scope traces.  Gate signal looks clean.  The TC4426 is managing to drive all four FETs.  I expect it is getting hot, or will get hot once frequency gets to 141kHz.  Temperature can be reduced some by gluing a small piece of aluminum or copper to the top of the IC as a heat-sink.

These latest scope images show oscillation at about 48kHz (21us period), even though the scope is reporting 147kHz.  The scope must be counting some of the ringing in its frequency measurement.  I suspect primary capacitance needs to be significantly lower, probably less than the 0.1uF I'd suggested previously.  Polypropylene capacitors are best for this use.  Mica is a possible alternative, but will run hotter.

Do you have design parameters for this larger coil?  If so, enter them in the "JavaTC" online program:
http://www.classictesla.com/java/javatc/javatc.html
It will output frequency and lots of other useful information.

[bogdan]

So, i made some tests with 4 and 6 caps, the IC is barely warm to touch, so that is not a problem...yet, however i will look for a tiny heatsink, also i have thermal double sided tape so that should be good.

"These latest scope images show oscillation at about 48kHz (21us period), even though the scope is reporting 147kHz.  The scope must be counting some of the ringing in its frequency measurement."
That can be because of the Chinese oscilloscope, i can't afford a better one unfortunately...
 I tried using javatc, but they ask for a ton of parameters and i get lost within the numbers and i quit very quick.

The following tests are made with a 12v PS, and the arcs are gone from 3 caps up.

[davekni]

Even fancy scopes can't magically differentiate between intended cycles and noise spikes when measuring frequency.  They may have a low-pass-filter option to help.  The cleaner gate waveforms appear to report the correct frequency.

37nF (0.22uF / 6) looks like it's hitting your expected frequency well, and the upper halves of the primary voltage appear like inverted sine-waves.  Thus I think that's a good capacitance.  What is the current draw from 12V supply for this 6-series-caps case?  What current draw did 1 and 2-cap cases have?  If the current is low with 37nF (6 caps in series), it may just need more voltage.  The most efficient case may have higher impedance, so need more voltage at less current for the same power.

I also suggest trying phase reversal (switching primary coil leads) again with 37nF.  Now that it's in tune, the optimum may not be the same as it was running at a harmonic.  (With the previous larger caps, the secondary was likely resonating at the 3rd harmonic of primary drive.)

Can you share what information you have for your coils, primary and secondary?  Diameters, heights, number of turns, wire gauge, top load diameter?

[bogdan]

The test at 12V with 1 cap draws @5A, and with 6 caps draws @0.72A. And after i switched the primary it draw @1A, but also the IC gets warm for a few seconds into the test, i am assuming it will get toasty within a minute, but other interesting stuff happened after i flipped the primary, like the frequency shoot up to almost 500kHz. As for the coil, it is almost time to go to work, but i will disassemble the coil tomorrow and measure everything. Also since i had laying around 4 more capacitors, i put those to test as well.

[davekni]

Looks like your primary coil was the preferred polarity.  500kHz with the leads reversed isn't useful on this coil.

0.72A at 12V and ~140kHz looks nicely tuned.  The primary is a parallel L/C tank circuit, so will draw minimum current when tuned to the drive frequency.  I recommend leaving 6 0.22uF caps in series and increasing voltage above 12V.  However, the 100V IRFP140 FETs limit input voltage to ~30V.  24 or 30V may be sufficient to get some sparks.  Efficiency will be better this way than 12V and a single 0.22uF cap.

As you can see on the 500kHz gate waveform, the TC4426 is struggling to drive 4 paralleled IRFP140 gates.  At 835kHz it will be worse and get hotter.  It might work at 835kHz driving a single FET.  (If I understand your intent correctly, you are building this slayer circuit as a test for eventual use with your tiny coil.)

To get longer sparks, a Tesla coil needs to be driven for short periods with gaps between each period.  This is what interrupters do.  During each enable period, the power can be higher, as the gaps keep the average power reasonable.  Simple circuits use a 555 for interrupting, pulsing the driver chip's enable pin high for short times periodically.

[bogdan]

Ok, i will try to raise the voltage tomorrow, do i need to protect the input of the IC since it is connected directly with the other end of the secondary ? A zener at least ? Unfortunately this IC does not have an enable pin...i will look for alternatives, and does it has to be inverted ?? i don't quite understand why. and you have the measurements of my coil in the bottom, one thing i forgot to take a photo is the secondary diameter, which is 110mm, same as the inside diameter of my top load.
 

[davekni]

Do you have a separate 12V supply for the driver chip as shown in the PowerMax slayer circuit you copied?  That is necessary once higher voltage is used for the primary coil.  D1 and D2 are there (one to 12V and one go ground) to protect the driver input.

A non-inverting driver will work fine with reversal of the primary coil leads.

Another option is to keep your existing TC4426 driver and add a PFET between the 555 interrupter and driver input.  I'll sketch a schematic of that option if you are interested.  The PFET would pull the driver input high when disabled, forcing the IRFP140 gates low (off).

Does the foil tape at the bottom of the secondary coil make a full turn?  If so, that shorted turn will load the coil and hurt efficiency.  Similarly, most designs have the top-load spaced above the top of the secondary winding.  Otherwise the top-load also acts as a shorted turn.  That is less of an issue being at the top of the secondary.  A shorted turn at the bottom is more of a problem.  (Is the ring that holds the bottom of the secondary made of plastic or other non-conducting material?  Looks OK, but I can't tell material for sure from images.)

I guessed 72cm as the length of your secondary tube.  Winding appears to start about 2.5cm up, so 69.5cm of winding.  Using that, here's the JavaTC results for your coil.  I adjusted secondary turn count to get resonant frequency close to your 141kHz result, ending up with 2100 turns.  The images suggest finer wire and more turns, but it's hard to tell.  If the bottom tape is making a shorted turn, that would explain some of the difference.

For this 141kHz coil, the standard SSTC 1 circuit you started with should work fine.  So, another option is to go back to that circuit and figure out what was causing issues.  The only reason I suggested slayer was for your 835kHz tiny coil.


J A V A T C version 13.5 - CONSOLIDATED OUTPUT
11/18/2020, 10:45:04 AM

Units = Centimeters
Ambient Temp = 68ºF

----------------------------------------------------
Surrounding Inputs:
----------------------------------------------------
999 = Ground Plane Radius
222 = Wall Radius
255 = Ceiling Height

----------------------------------------------------
Secondary Coil Inputs:
----------------------------------------------------
Current Profile = G.PROFILE_LOADED
5.5 = Radius 1
5.5 = Radius 2
2.5 = Height 1
72 = Height 2
2100 = Turns
0.025 = Wire Diameter

----------------------------------------------------
Primary Coil Inputs:
----------------------------------------------------
Round Primary Conductor
7.6 = Radius 1
7.6 = Radius 2
1 = Height 1
13.6 = Height 2
14 = Turns
0.25 = Wire Diameter
0 = Ribbon Width
0 = Ribbon Thickness
0.037 = Primary Cap (uF)
80 = Total Lead Length
0.2 = Lead Diameter

----------------------------------------------------
Secondary Coil Outputs:
----------------------------------------------------
140.37 [kHz] = Secondary Resonant Frequency
90 [deg °] = Angle of Secondary
69.5 [cm] = Length of Winding
30.22 [cm] = Turns Per Unit
0.08095 [mm] = Space Between Turns (edge to edge)
725.71 [m] = Length of Wire
6.32 [:1] = H/D Aspect Ratio
252.8064 [Ohms] = DC Resistance
52168 [Ohms] = Reactance at Resonance
0.317 [ kg] = Weight of Wire
59.15 [mH] = Les-Effective Series Inductance
70.635 [mH] = Lee-Equivalent Energy Inductance
71.54 [mH] = Ldc-Low Frequency Inductance
21.734 [pF] = Ces-Effective Shunt Capacitance
18.2 [pF] = Cee-Equivalent Energy Capacitance
50.389 [pF] = Cdc-Low Frequency Capacitance
0.2156 [mm] = Skin Depth
14.097 [pF] = Topload Effective Capacitance
309.9564 [Ohms] = Effective AC Resistance
168 [Q] = Quality Factor

----------------------------------------------------
Primary Coil Outputs:
----------------------------------------------------
169.41 [kHz] = Primary Resonant Frequency
17.15 [% low] = Percent Detuned
90 [deg °] = Angle of Primary
668.53 [cm] = Length of Wire
23.48 [mOhms] = DC Resistance
0.65 [cm] = Average spacing between turns (edge to edge)
1.962 [ cm] = Proximity between coils
0 [cm] = Recommended minimum proximity between coils
22.792 [µH] = Ldc-Low Frequency Inductance
0.0539 [µF] = Cap size needed with Primary L (reference)
1.061 [µH] = Lead Length Inductance
373.356 [µH] = Lm-Mutual Inductance
0.292 [k] = Coupling Coefficient
0.13 [k] = Recommended Coupling Coefficient
3.42 [half cycles] = Number of half cycles for energy transfer at K
9.56 [µs] = Time for total energy transfer

----------------------------------------------------
Top Load Inputs:
----------------------------------------------------
Toroid #1: minor=12, major=34, height=72, topload

[bogdan]

Yes i have a separate power supply for the IC. Yes the foil at the bottom also on the top makes a full turn and is shorted. The wire i used is 0.1mm i removed the tape at the bottom and i made some more solid connections for the capacitors, and i had a big surprise, my frequency on the gate is 44kHz...sometimes i feel i am followed by a curse or something. Can the foil at the bottom have an effect that big ? do i need to start over and find the frequency of the secondary and go trough all the process again ?
If i want to go back to previous circuit for this coil i need to order new IC from China, that was the only place i found them, and i don't fully understand the gate drive transformer, i am walking into 2 unknown territories.

[costas_p]

Hello bogdan,
I am not trying to seem smart or something, but I can say , that I have successfully replicated the 'SSTC on steroids' , with DS0026 (from ebay) and a 240 to 24v transformer, utilizing a voltage doubler for around 71V, also I made the full circuit (with the 555 interrupter) and produces around 5cm sparks at 120W

I am reading and replying from my mobile since I am out of house for the weekend, at Monday I can help you as much as you like about this circuit, and make it work.

[plasma]

I think 6 caps at 0.72A is good for 12V ,the resistance would be low, like said above more source voltage will mean more circulated current, I'm guessing a Q of 20 for the large inductor minus resistance, you should have 20A if you measure the current between the secondary and top load, measuring the voltage across a 1 ohm resistors should give a value to workout current, the bandwidth should change though, so more source current will be supplied.

[davekni]

Removing the foil tape will lower frequency, but not from 141kHz to 44kHz.  Perhaps to 120kHz.  Did you remove tape at the top too?  If not yet, I recommend removing top tape and raising the top-load a few cm with some kind of non-conductive spacer.

Was my 720mm secondary height close?  Can you measure roughly how many turns per mm your secondary coil has?  With 0.1mm wire, 7 turns per mm would be reasonable.  That would be about 5000 turns, which might get you close to 44kHz.

I agree that it makes sense to get this slayer circuit working first.  Then you can consider upgrading to the half-bridge version once other issues are resolved.

[bogdan]

Any help is welcomed, i want to understand as much as possible about this resonant frequency and how to troubleshoot the coil. only at the bottom, t the top hopefully i will do in next week, it's a bit more complicated. Measured with a digital caliper and a magnifying glass, your 7 turns/mm was spot on , with 74cm total height of windings, that would make for 5180 turns...+- a few. So the plan for next week is to remove the top load and put it higher than the last turn, and build a simple slayer to check for new frequency. Am i on the right track ? If the wire is too thin, i have 2 spools of wire, 0.25mm and 0.4mm if you say it will be better i can rewind the secondary with no problem.

[davekni]

With secondary of 5180 turns and 74cm long, JavaTC estimates 58kHz.  That should be close to accurate once the upper shorted-turn is removed.

Yes, eventually larger wire with fewer turns will be more optimum for that size coil.  That will result in a higher resonant frequency closer to your original 141kHz result.  However, I'd suggest experimenting with this winding before changing.  The lower 58kHz makes debugging drive circuitry a bit easier.

Do you have any sort of signal generator, even a home-made 555 circuit with potentiometer for adjusting frequency?  That's the best way to find secondary resonant frequency.  Connect the bottom of the secondary winding to the signal generator and adjust frequency slowly until a peak in signal amplitude shows up on a scope probe hanging in the air near the top-load.

[bogdan]

Yes, i have a signal generator, up to 2MHz...or so they say, but for my range of frequency should be enough. Ok i will use this method as well to check for the resonant frequency once i lift up my toroid and remove the shorted tape at the top as well. Question, can i use a conductor, like a threaded rod to lift up the toroid, or use some wood or something isolating for the support as well ?

[davekni]

I've seen a single threaded rod in the middle used before.  Multiple rods could work if they are insulated at one end or the other to avoid making any shorted loops.

[bogdan]

So, my mom took care of raising the topload... when she bumped into the coil and it fall down. That reduces the windings to 60cm. the lower foil is gone, the top foil is gone, i now have a bit of pvc pipe between the last winding and the toroid, but i had to turn the coil, since the damage was on the lower part. With the signal generator method i get 3 significant resonances, a small one, at @44kHz, a big one at 138kHz and another one exactly the same at 277kHz. Which one do i choose ?
Since the 138kHz was close to my original 140kHz i said i would give it a try with 12V, i also flipped the primary since my secondary was also flipped after the accident, but it did not liked that, it stayed at 500 and something kHz and did not moved no matter what capacitor i used. So i went with the original orientation of the primary and 5 caps, the results were in 140kHz on the coil, @50kHz on the gate and a current draw of 4A, also a bit of corona forming on the topload. Why is this difference happening, and why is the curent draw so high compared the the old secondary?
Tomorrow hopefully i can get some pictures of the coil and gate frequencies.

[davekni]

Removing turns is one way to get space to the top-load:)

Let's figure out the secondary resonant frequency before trying to run the coil.  For that test, leave the primary unconnected.  Otherwise primary resonant frequency will complicate the results.  Can you post scope images of your peaks at 44, 138, and 277kHz?  Some of those may be harmonics from the signal generator.  Can the signal generator be adjusted in steps finer than 1kHz?  The peak at 44kHz might actually be much stronger at 44.5kHz or some other slight change from 44kHz.  The top-load signal picked up by the scope-probe near by should look like a sine wave even if the signal generator output is a square or other wave shape.

[bogdan]

It can do steps of 1Hz so that is not a problem, and can do pure sine, square or triangle. Which one should i use for the test ?

[bogdan]

I took the coil out and put it on a flat surface (i could not imagine it makes such a big difference) with nothing around it except for the oscilloscope probe, and i started with steps of 1kHz, and when i saw some waves i started with steps of 100Hz, and i got this result:
One strange thing that happened, i did not included in the photos, is that on my generator was one thing, @ 58kHz, and on the oscilloscope was 100 and something. Is this normal ? a harmonic or something ?
 

[davekni]

Yes, for the case of generator at 58kHz, the coil is responding to the 3rd harmonic of the generator waveform (3 * 58kHz).  The 178kHz on the scope is that coil frequency.  However, I suspect 48kHz of your first measurement is the fundamental resonance of your secondary coil and top-load.  I'm guessing that 178kHz is higher mode with multiple high-voltage nodes up the coil.  That is not where you want to run.

Was the top-load in place and connected for this test?  The top-load is necessary to find the secondary resonant frequency.

A picture of the test would be great.  I was a bit surprised at how low the amplitude is on scope traces, ~100mVpp.  That may be fine depending on how far away the probe is from the top-load and what amplitude the signal-generator generates.  (The scope probe needs to be reasonably far away to not add capacitance, say no closer than half of your secondary height, ~300mm or more from the top-load.)  I'd set generator output amplitude at its maximum.

If the top-load was connected, then don't bother re-running the test.  48kHz is a reasonable result for your coil, not too far from the 53kHz JavaTC shows presuming 5000 turns.  (Was the scope signal much larger at 48.39kHz than at other frequencies such as 40kHz or 60kHz?  It should be a big difference, say 5x larger.)

[bogdan]

Yes, the top-load was in it's place when i ran the test. 48.39kHz was the biggest until i reached 177.85kHz, in 100Hz steps the output goes away in about 5 steps so at @ 40 and 60 i have almost no output

[davekni]

Great!  48kHz is the fundamental secondary frequency.  The goal now is to get the Slayer oscillator running around that frequency.  I'd start with 0.33uF, but also try 0.22uF and 0.44uF (two 0.22uF caps in parallel).  JavaTC indicates 0.38uF for primary resonance matching.  I think Slayer oscillators tend to work best with capacitors a bit under the resonant value.

I'm guessing that the higher-mode 178kHz showed higher amplitude because a node part-way up the secondary coil was closer to the scope probe.  Not sure.  Either way, the fundamental frequency of 48kHz is where the coil will run best.  (The frequency will not be exactly 48kHz, as the primary coil will shift the secondary frequency slightly.)

[bogdan]

I have made some tests with a few capacitors, and i have managed to kill my IC...i don't know how... All i have now is the non inverting pair of this one, the TC4427, and i inverted the primary, all the tests are made with this configuration, the IC and MOSFETs are stone cold.

[davekni]

Without labels, I can't tell which signal each scope image is measuring.  If the square-wave traces are of gate-to-source (Vgs), the voltage is too low to turn on the FET completely.  Vgs and Vds are the most useful plots to see.  If the other traces are Vds (drain-to-source), then the hump in the middle of low times would fit with low Vgs.  Otherwise, 0.33uF appears to be best.

[bogdan]

Yes, the traces are what you said. I made the test with 6.5V because it was already drawing over 3A. Here is the test with the 33nF cap with 12V.

[davekni]

Could you please scope Vds directly at the FET leads?  That will be a bit easier to decipher than measuring at the coil.

Also, a picture showing wiring from FETs to 0.33uF cap and primary coil would help.  (BTW, I presume you mean 330nF, not 33nF.)

After looking again at your scope images and running a quick simulation, I think I was guessing wrong as to which way around the scope is connected on the coil.  (Another reason to scope at the FETs, ground to source and probe to drain.)  After reinterpreting the images, I think 0.22uF is probably better.

[bogdan]

Yes, i meant 330nF, my bad, sorry. When i connect the scope to the coil, i just put the ground and probe directly on the coil/cap connection, since my scope is on a battery, i don't have to deal with it's ground. Here are the tests and traces for gate and drain, the unlabeled picture is the gate trace of 330nF.

[davekni]

Thank you for the scope images.  Those measurements on the FETs are easier to interpret.

Both 220nF and 330nF look reasonable.  330nF has a bit lower Vds, so a bit more headroom for increasing bus voltage, to around 30V for your 100V FETs.

Are you interested in downloading and learning the free analog simulator LTSpice?  I made a simulation of your Slayer circuit:


Plots look reasonably similar to your scope traces.  Here's Vds (green) and Vgs (blue) with 12V Vbus supply:


These are image captures.  I'll zip the simulation schematic if you decide to use LTSpice.  (There are other free analog simulators too.)

The current draw from VBus looks about right too.  Most of the power is going into secondary winding resistance.  Per simulation, increasing VBus to 30V gets just over 55kV peak on the top-load, likely enough to get some arcs.  Current goes up to 14A.  Do you have a supply that can provide 30V at 14A?  Or, if you have a large capacitor, charge it to 30V and use it for momentary power.  Unless you add a heatsink to the FETs, momentary operation is all that they will handle at high power.

Most SSTC designs include a 555 timer or other interrupter source to enable the driver for short periods of time.  That would make the average current draw much less than 14A at 30V, and is one of the key reasons interrupters are used.

The only other option that comes to mind for reducing power is to rewind the secondary with thicker wire.  That will reduce resistance and increase resonant frequency.  Doubling wire diameter should roughly cut secondary losses in half (for a given secondary voltage).  So, power will still be fairly high.  That points back to adding an interrupter.

[bogdan]

I have used LTSpice before when i was playing with the induction heater circuit, i will try to find some higher voltage mosfets. Meanwhile i tried it with a 24V transformer, with the ground connected, and i touch only momentarily the positive lead of the transformer to the connection on the pcb...and my bridge rectifier is all over the room in tiny pieces. Fortunately the circuit is ok, but i did not hear any corona noises before it blew up.

[davekni]

Good to hear that you have used LTSpice before.  Here's the LTSpice simulation schematic for you to try, zipped to make it compatible with attachments for this forum:
slayer3.zip

Try running this circuit.  I believe the topology matches what you have implemented.  Once you are comfortable with exploring this, we can discuss adding interruption.  That will work better than momentarily connecting the power supply.

[bogdan]

Thank you, i will try and learn from it as much as i can

[bogdan]

So, i have been playing for some time with the simulation, thank you again for that, and with a bunch of tesla coils that i make(i rather built something that i will throw away than stay and do nothing), and i give you the best result i have so far with a new coil i built,
changes to the circuit:
-the fet driver is ds0026(i blew up the tc4427 and tc4426 that i had )
-the fets are 4 IRLZ34CN
 The TC is
-topload just some dishes put together so that they form a sphere, diameter @20cm
-secondary 0.15mm wire on a 50mm pipe for 45 cm
-primary almost 6 turns of 2mm wire
 The resonant frequency is 185kHz. i put 25v and it draws @12A. The fets stay pretty cold but the IC gets a little warm, so i put one of those small cooler for the stepper motor drivers, it is better than nothing
I am happy that i achieved some corona discharge without the need of a screwdriver near

[davekni]

Great to see progress!

That is a tall thin secondary.  A lower aspect ratio would improve coupling coefficient, which should help with SSTC performance.

[bogdan]

Now that i am starting to actually make some progress, i want to build a tesla after your recommendations, if you want to help me and have the time of course. I have available for secondary 50mm, 75mm, and 110mm pipe, and for wire i have 0.15mm and 0.22mm. However, should we still continue on this topic? it is getting off topic a bit.

[davekni]

A new topic wouldn't hurt.

For design recommendations, first look at guides from Mads web pages or Steve Wards or other examples.  Come up with a proposal of your own that looks something like successful examples you see, modified for the materials you have on hand.  Use the JavaTC web program to model your coil.  Then post the design you come up with for feedback here.  I'd love to be helpful, but don't want to replace the learning you will get from researching and designing yourself.

[bogdan]

Ok, so here goes my thinking:
“Secondary: 4.5"x18" winding of 30awg magnet wire.
Primary: 6 turns of 12awg on a 6.3"dia x 6" tall form.
Toroid: 8"x1.5".”
This is the design i choose since i have almost similar materials.
Mine will be:
Secondary 110X460mm winding with 0.22mm wire
Primary 6 turns of 2mm wire with 160 dia and 150mm tall
Toroid 330X110 mm
Some differences but quite similar. Now i have played a little with javatc, and i think i have a plausible result. So, what do you think? It is ok to copy, but I don’t understand the chosen materials and primary, why not flat? Or conical? Will it be better for coupling? I still don’t understand this coupling thing.

[davekni]

That coil design looks great.  Good coupling for driving with SSTC circuits.

A couple quirks with JavaTC:  Most round parameters are listed as diameters.  However, primary and secondary coils are listed as radius.  So you need to enter 5.5cm for the secondary radius and 8cm for the primary radius.  Also, I've found that the helpful little drawing JavaTC makes doesn't show up correctly if heights are entered as 0.  It works with 0.001.  The drawing is helpful to find bugs.  Your toroid center height is listed at 11cm, which is below the top of your primary.  It should be around 50-55cm, so above the top of the secondary.  Try JavaTC again with those fixes.

Concerning coupling factor, qualitatively, it is how much overlap there is between the magnetic fields of two coils.  If coils are at opposite ends of a room, there is very little coupling, so very little energy transfer.  If one coil is wound on the same form on top of another, they share much of their fields.  That is the way transformers are designed.

For coils of the same inductance, coupling is how much of that inductance is shared.  The left and right circuits in this schematic are equivalent:


For different inductance, the math is a bit more complex.  The idea is still the same.

[bogdan]

ok, i hope i made all the adjustments correctly, i made that one on my phone so the writing was a little small . The frequency seems a little high for me, or am i wrong ?

[davekni]

Getting closer.  I'd missed one other issue previously.  For the top-load, you need to click "Add" after entering the dimensions.  (JavaTC allows multiple top-loads.  You can enter multiple dimension sets, adding each one.)  Before adding the top load, choose an appropriate center height, something above the top of your secondary winding.  11cm height won't work.

The missing top-load is making the frequency higher.

And I still suggest making bottom coil heights 0.001 instead of 0.  At least on my browser, the coil sketch doesn't show up with exactly 0 for bottom heights.

[bogdan]

Ok, very excited, and learning something each step i haven't even noticed the sketch on the page

[davekni]

That little sketch is great for seeing typos in entered values.  One thing it doesn't show is scale.  Looks like this run accidentally has "inches" checked instead of "cm".  Some parameters remain roughly the same with scaling (coupling factor, impedance), but frequency will be higher once you check the "cm" box.

Have you noticed the button near the bottom for "Make a Load File"?  That allows saving parameters to your computer as a text file, which you can load again next time.  Saves a bunch of time when making small changes.

You will need to increase primary length slightly for 0.022mm wire - enough to accommodate insulation and for tiny bits of space due to imperfect winding of the coil.  Notice that now your "Space Between Turns" output is 0.  Something more like 0.2mm would be more realistic (0.02cm).

[bogdan]

Ok, i hope this time i got it right, i double checked, everything seems to be ok. Also my primary is done, and the secondary should follow this week, then we will see the real frequency and working of the coil.

[davekni]

Looks great!  Have fun with your ongoing construction.

[bogdan]

Yaayy, i finished...with a few issues, my spool of wire ended before i was able to finish, but only 6 cm are missing, i tested it out with the oscilloscope method, it gave me the highest reading at exactly 375kHz, i plugged in my last circuit, and put 4 of my 0.22uF caps in series, it seems to do the best with 4, powered with 12v, it draws 10A and i got a photo of the gate signal.
However, why is there such a difference between java and real coil? java says something like 136kHz without those last 6 cm that i am missing, and in reality i got 375. Can it be a harmonic ?

[davekni]

Yes, it could be a harmonic.  JavaTC isn't going to be far off unless some parameter doesn't match your real coil.

Do you have the top-load in place and connected to the top end of the secondary wire?  Is the bottom end of the secondary connected to ground?  Or, if you are testing by driving the bottom of the secondary with a signal generator, the return (ground) of the signal generator needs to be tied to earth ground.

Besides ground through a line cord, it is best to have a counterpoise - some conductive sheet on the floor or surface under your coil.  Should be far enough under to avoid acting as a shorted turn for your primary, say at least as far below as the primary outer radius.  Aluminum foil is easy and cheap.

[bogdan]

This is what i got, 116.4kHz with the signal generator and oscilloscope method. With the circuit connected i seem to get the best result with a 0.33uF cap, at 12V it draws 17A. you have the waveforms of the gate drive, and directly on the coil.

[davekni]

Looks like the secondary resonance is around the 103kHz you measured.  Look closely at the second scope capture.  Besides the higher-frequency harmonics, there is a fundamental that is just over 100kHz (just under 10us per cycle).  For example, count just the most negative peaks across the screen - just over 10 in 100us.  The harmonics are likely resonances in the drive circuitry rather than in the secondary coil.  (The secondary will have higher-frequency modes where peak voltage is in the middle of the coil rather than at the top.  I'm guessing that resonances in the primary circuitry are more significant here.)

[bogdan]

Yes, i saw it. But why so much current draw compared to my previous versions?

[davekni]

My guess is that the high-frequency part is dissipating a lot of power in the FET.  From what I read here, Slayer oscillator circuits require tuning and luck to achieve success driving Tesla coils.

[bogdan]

So it is time to find a better driver then. Do you have any recommendations?

[davekni]

I'd start with one of the basic SSTC designs such as Steve Ward's SSTC-5 (Mini-SSTC) or Mads (Kaiser Power Electronics) SSTC-III.  These circuits can be used with secondary current feedback from the bottom of the secondary winding instead of antenna feedback as shown.  Current feedback does require either a current-transformer or a capacitor load to attenuate the secondary current signal enough to match the level of an antenna feedback.  Current feedback would be my personal preference.  However, I have not personally built these SSTC circuits.

For the half-bridge layout, I'd look at my example post for low inductance:
https://highvoltageforum.net/index.php?topic=1324.msg9795#msg9795

[bogdan]

I actually built the Kaizer SSTC III a while back, but never really got it to work, note that i am unfamiliar with GDTs, i never worked with them.

[davekni]

That SSTC-III build looks fine.  Your GDT is very well made and wired (presuming it is a ferrite core and not powdered iron).  Make sure unused HC14 inputs are tied to something.  If not already, bridging adjacent pins can work, pin 14-13, 12-11, 10-9, and 5-4.  That avoids any floating input pins.  (This presumes you are using the inverters of pins 1,2,3,4 for the intended circuit.)

At 100uF, the bulk capacitance is on the low side.  That shouldn't keep it from running.  Why not connect this to your latest coil?  Initially feed the antenna input from your signal generator with a ~5Vpp waveform and use a lower bus voltage.  Scope the half-bridge output to verify that GDT outputs are phased correctly.  Scope the gate signals too.  That will show if there is any issue with the GDT core being too low inductance.

[bogdan]

HC14 inputs are grounded, however, i managed to accidentally touch my signal generator's leads to 12V and now it is dead... i made some tests with my internal generator from my oscilloscope, but i can't move the ground. But i do have some questions, you will have test signals from
-555 timer output
-input of the ucc
-one of the GDT's primary terminal
What i don't understand is, it should be on for as long as the 555 is on, right? Here it looks like it is on all the time.
555 timer

Input of ucc

GDT

[bogdan]

Sorry to replay, but in edit mode i can't add any attachments. So i added 2 LED modules in antiparalel where the primary should be, just to have a load on the MOSFETs, i scoped the gate of one of the MOSFET, and the result is this. Take note that both grounds are connected together.

[davekni]

Your first scope capture of UCC input and GDT waveforms shows ~33Hz frequency.  That is way to low for testing the UCC chips and GDT.  It would be good to repeat those scope captures at 100kHz.  (The 555 output should be low frequency as shown, ~7Hz.)

There have been a lot of problems with counterfeit UCC driver chips - cheap Chinese knock-off parts claimed and labeled as real UCC chips.  Most of the problems reported here are with non-functioning enable inputs.  I'd guess that is what you have.  If your budget allows, purchasing parts from standard electronic part distributors is more reliable.  Try searching for part numbers on
      http://www.oemstrade.com/
That will list what distributors carry the part and pricing.

LED modules aren't a good load simulation.  They are designed for constant-current drive and often have considerable capacitance.  For low-power testing, a resistor is fine, or a coil of wire with at least as much inductance as the final primary will have.

[bogdan]

That may have happened because i did not changed the time on oscilloscope, i just wanted to see the waveform, here are the tests with proper time. I ordered them from China because i did not find them on my sites where i usually buy stuff. I did find them now on a big site and they say it's from Texas instruments, so that should be ok. I will wait for them to arrive, probably on Tuesday - Wednesday and rerun all the tests, also ordered 4 irfp460, do you think it will be ok if i make a mains full bridge (i have 230V)? If not, can you recommend some transistors that you think could do the job. Meanwhile, is there any other test i could do with the fake Chinese to prepare for the arrival of proper ones? Also, if this will be a succes i will build a full bridge with indications from your topic on how to build a half bridge with low inductance.
input of the UCC

one of the GDT primary

one of the Gate of the MOSFET

[davekni]

Yes, IRFP460 should be fine for a full bridge on 320Vdc (peak of 220Vac).  IRFP460 Vds rating is 500V, leaving 180V margin for switching spikes.  When driving a full-bridge, C4 (the capacitor in series with GDT primary) will need to be larger.  0.1uF is on the small side for driving two IRFP460 parts.  I'd increase it to 0.47 or 1.0uF to drive a full-bridge.

The GDT input waveform is more rounded than I'd expect.  Can you scope the other GDT input as well?  Can you also scope both GDT inputs with the signal generator set to 50kHz?  That would help in determining what may be causing the rounding, even though your coil will not run at 50kHz.  Increasing the value of C4 might help with the GDT waveform rounding too.

Do you have any information on the core you used for your nicely-constructed GDT?  It looks like ferrite in the image, but pictures aren't definitive in determining core type.  If core inductance is low, that could also explain the GDT input waveform rounding.

[bogdan]

I have no idea about the specs of that core, i am sorry. Herer are the tests, however, my closest option is 10kHz, i don't have a 50kHz option output on my cheap scope...sorry that i killed my function generator... When i probe sometimes the whole thing just turns off, so i had to use a 10-1 probe on the scope.

first cap



1uF cap

[davekni]

The GDT appears to handle 10kHz fine, so the core is good.  Waveforms look fine with 1uF.  I suggest leaving 1uF connected.

With 1uF for C4, you could scope low-side-FET gate waveform again and the half-bridge output with some low VBus voltage (10-20V) to verify that gate drive phasing is correct (one FET on while the other is off, not both on together).

[bogdan]

sure

gates first 2, and i replaced the led modules with a coil, 3rd and 4th picture

[davekni]

I'm assuming the actual scope scale is 20V/div, not 2V.  If so, the waveforms look acceptable.  GDT appears to be wired correctly.

For looking at waveform shape details, it is easier if one cycle fills more of the screen.  For example, if your scope can do this, 2us/division horizontal and 5V/division vertical.  In cases where 5V/division causes waveforms to extend off-screen, then change to 10V/div.

The spikes on half-bridge output (third image) are likely due to VBus wiring inductance.  They cause no issues at low voltage, but need to be checked at higher voltage before going all the way to 320V, making sure spikes don't get past the FET's 500V rating.  If you change to the parallel-plane layout, those spikes should be much lower.

Good luck with the new UCC chips when they come!

[bogdan]

Yes, the probe is set to 10-1, I am glad to hear everything is on the right track. I will post new tests when the new UCC arrive.

[bogdan]

My new UCC arrived, they are ok, the enable pin works, also i put the coil in the setup, powered with 30v. However, the antenna has to be really close to the ground part of the secondary in order for the coil to oscilate, and the ground leeks a lot of corona, so much that on the top i can't draw any arcs

[davekni]

If you are getting corona at the ground end of your secondary, then there is a problem with the connection of the bottom of the secondary to ground.  There should be almost no voltage at the bottom end of the secondary.

Make sure you test both polarities.  Swap wires on either the GDT primary or Tesla coil primary to invert phase.  One phasing will work much better than the other.  Often one phasing does not oscillate at all.