mini tesla (hopefully not a fail)

[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).