mini tesla (hopefully not a fail)

[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