Tesla coil circuit draws lots of current but the spark size is small.

[SalinsLV]

Hello!

I have encountered a weird problem. I have 2 almost identical Tesla coils. Primary coil is 5 turns for both coils.
1st secondary coil:
24cm length
7,5cm width/diameter
0,22mm wire diameter
~1090 turns
Measured Fres: ~433,8 kHz

2st secondary coil:
24cm length
7,5cm width/diameter
0,2mm wire diameter
~1200 turns
Measured Fres: ~337,8 kHz

The 1st Tesla coil has a current draw of about 800mA at 200v while the 2nd coil draws 1,6A of current at 200v. At first i though maybe there are constant flashovers between windings on the 2nd coil, but when testing i didn't see any sparks between the secondary coil.
What i don't understand is why does the 2nd coil consume more power while having smaller sparks. NOTE: the circuit remains unchanged, i only swapped the coils.

I added a picture below. The coil on the left side is the one that draws a lot of current but produces small sparks, while on the right side is the coil that functions properly.

[klugesmith]

Maybe it's a frequency tuning thing. 
You have a good example of coils with identical dimensions, differing only in number of turns (and wire thickness).
The one with more turns has more inductance (by square of the turns ratio) and lower resonant frequency.

SSTC driver circuit might work better at one frequency, and I'd give particular attention to phase control details.
But am not a Tesla coiler, nor pretended to be one.

[SalinsLV]

I added resonant frequency measurements. The working coil has a Fres of 433,8 kHz and the one that draws more current has Fres of 337,8 kHz.

[klugesmith]

What would happen if you put a capacitive load on top of the higher frequency coil,
enough to bring the resonant frequency down to match the lower frequency coil?

That would not make the secondary circuits electrically identical.
Modified one would have lower L, higher C.  Same resonant frequency, lower magnitude of impedance.

Maybe both coils could benefit from top loads & lower frequency operation.

[davekni]

SSTCs are usually designed with a sufficiently-large DC blocking capacitor that the primary resonant frequency is far below normal operation at the secondary resonant frequency.  However, I've seen two cases already where SSTC primary capacitance was too low, effectively making an unintentional DRSSTC.  That could be your case, where one secondary is getting close to primary resonance.  The test that klugesmith just suggested would help determine if frequency is the key parameter.  Or, share your primary circuit details.

Another possibility is that one coil has a shorted turn somewhere.  Not too common, though.  Or some difference in construction that includes some lossy ferrous or conductive material.

[Zipdox]

Have you calculated the peaj primary current? http://kaizerpowerelectronics.dk/tesla-coils/sstc-design-guide/

Might also be that your feedback is out of phase.

[SalinsLV]

When i add top load to the working coil, the spark length decreases but the current draw increases.
The only feedback i have is the secondary coil bottom wire connected to the gate of the transistor.

[Mads Barnkob]

When i add top load to the working coil, the spark length decreases but the current draw increases.
The only feedback i have is the secondary coil bottom wire connected to the gate of the transistor.

That sounds incredible crude... As in the grounded secondary wire connected directly to the gate?

Could you please post a schematic of your electronics?

[Zipdox]

I fear he has built a slayer exciter...

[SalinsLV]

I added the circuit in attachments. The idea is that the Left side of the circuit is like class D audio amplifier. It modulates the power to the driver. The right side is "kacher brovina" or slayer exciter. This circuit works just fine. 10-15cm sparks that play music. This circuit works even at 250V. At 300V something shorts that causes Q1 to blow up. The max current before destruction is about 2,5A at 250V. I know this circuit is trash but i'm currently stuck. I want to implement Gate drive transformers to isolate the mosfets because when one of them shorts out, half of the circuit blows up(PCB traces vaporise, IR2113 gate signals stop working and usually sn74hc also dies). I previously had connected 1.5ke400ca TVS diodes parallel to every IRFP460 mosfet, but if the diode shorts, the transistor shorts and the chain of destruction starts once again.I also had a resistor parallel to primary coil but that only reduced spark size and increased current draw so i removed it. The optic receiver does receive signal but i don't use it because the spark with optic fibre is 2mm at maximum. If i connect my phone directly to LM393 3rd pin, then i can get maxiumum 15cm sparks.

I really like "kacher brovina" (that little driver on the right) because it's so simple and i have achieved +45cm sparks only with the driver(only power modulation was a triac, nothing special).

Getting back to GDTs. I would like to implement them in this circuit but i don't know how to do that. I can't just connect GDT to IR2113, it would act as a short circuit.

[SalinsLV]

Some more pictures of my experience. Blown mosfets and ICs

[davekni]

I'm amazed that Q3 can survive at all, especially at 250V.  It's likely running into avalanche breakdown on every cycle, with peak voltage way above its 500Vds rating.  Also, the average gate drive voltage is increasing with supply voltage, rather than remaining around threshold.

What's the current rating of L2?  10uH is small for ~300kHz and 250V supply.  I think ripple current will be about +-10A.  C19 may have trouble handling that much ripple current too, even if L2 doesn't saturate by 10A.  Have you checked to see if C19 is still 1uF?

Do you have an oscilloscope?  This is a complex project to build blind (without scope traces).  Many of the issues would show up in scope traces before voltage is high enough to fry parts.  At some point all the fried parts will cost more than a basic scope.  I'd also suggest analog simulation.  LTSpice is my choice, a nice free tool, and one of the better ones even if paid options were included IMHO.  Simulation makes it easy to see L2 current etc. and try modifications without the work of soldering.  It won't catch inductor core saturation or other more subtle issues.

You could keep your class-D amplifier, with perhaps a fix for L2 and C19 currents (higher L2 value with sufficient current capability).  Then use the output to power a normal half-bridge (or full-bridge) SSTC circuit on a separate ECB.  Or, you could modulate duty cycle of SSTC bridge drive to avoid extra power devices of the class-D amplifier.

[SalinsLV]

Yes, Q3 is working on avalanche breakdown principle. I have an oscilloscope but because half of the main circuit is missing(because i don't have enough parts and i often take them out to use in other projects), i will need to put it back to together in order to take the measurements. About the inductor, I don't know the current rating because i simply took the biggest one i had at that time. It's wrong to assume these numbers,but i think that it's saturation point could be at 5A or maybe even lower. I will definitely replace it as You suggested, because that inductor was intended to act as a filter for frequencies above 20kHz but in the new version i won't need 20kHz because that is too high.

I have added in attachments a picture from this website: https://www.loneoceans.com/labs/qcw/#Design
I would want to recreate inductor with similar specifications as in the website, because it can withstand high currents before saturation. I think i found the core on aliexpress but i'm not sure: https://www.aliexpress.com/i/910690133.html

[davekni]

Transistor avalanche breakdown ratings are for the occasional transient event such as power-up.  Running continuously into breakdown wastes lots of power and overheats the transistor.

If you reduce the frequency of your class-D amplifier below 300kHz, then L2 needs to get even larger.  Higher frequency allows use of smaller inductors.  Yes, that core looks good, probably more than you need depending on your final class-D amplifier frequency choice.

[SalinsLV]

I did some measurements with the biggest inductors i currently have. The LC meter might be giving wrong values, but i have to work with what i have. So far i have these measurements from the coil i used in the Tesla coil circuit(the one with red wires and yellow core in the added pictures)
Inductance: 0,49 mH
Wire diameter: 1 mm
Outer core diameter: 4,2 cm
Inner core diameter: 2 cm
Height: 1,6 cm
Turns: 67

[davekni]

That yellow-core inductor you are using should be fine.  490uH is much better than 10uH, and looks reasonable for that coil.  Presuming the core is some form of powdered iron, it can likely handle 1T field, or about 20A for your 67-turn coil before saturation.  (Wire heating limits RMS current to a lower level.)  You could drop your class-D frequency from 300kHz to 50kHz to reduce switching losses and likely still be fine with that inductor.  Not sure if the 1uF C19 could handle the lower frequency (higher ripple current).  Depends on the type and rating of capacitor.