gdt high frequency ring

[davekni]

Thank you for scoping pictures that helps.  Even with a battery-operated scope, always connect the probe "ground" lead to a steady low-impedance node such as VBus- or VBus+ (the planes marked with red "+" and "-" in your picture, or FET source lead connected to VBus-), not to a gate.  Capacitance of the scope and probe lead will distort readings and can disrupt proper circuit operation.

Shifting horizontal position to the left is fine.  Vertical coupling looks to still be "AC".  It is hard to see at the image resolution, but it looks like the "~" symbol in front of the ch1 10V label at the lower left of the scope capture, which indicates AC-coupled mode.  Change the scope vertical channels to DC-coupled.

The "ringing" shown there looks like it may be multiple transitions from the driver.  Does it show up when scoping one GDT input (with scope "ground" to the driver circuit ground, not to the other GDT output)?

For gate waveform scoping, it can be useful to short VBus+ to VBus- with no power supply connected (or supply set to 0V).

[balazs]

I closed VBus + towards VBus. I provided the feedback signal with a signal generator. I took out the 74hc14 because when it was in the circuit, it didn't work. I connected the feedback signal directly to the ic pins. Probe directly to the GDT. The signal was perfect. However, when I put on the Primer and placed the 30v on it, the signal became bad again.

The oscilloscope was already in DC mode, this "~" signal is just an ornament.

[davekni]

Thank you for the DC clarification.  (I'm puzzled why the "~" shows up.  I've seen that used on scopes to indicate AC before.)

The probe connections are still reversed.  The alligator-clip is the scope "ground" connection.  It should be on VBus-.  The probe tip should be on the GDT output where it connects to the gate series resistor.

If the "ringing" still shows up with VBus powered and the scope correctly connected, then please probe one of the driver outputs (with scope ground on driver board ground).  If you have two probes, it would be ideal to see both simultaneously on ch1 and ch2.

The GDT output with no VBus power looks fine.  My guess is that half-bridge output is coupling to the antenna somehow.

[balazs]

I changed the probe leg so the ringing was over. Then I put the 74hc14 back on and tested the antenna. The sign was much prettier than it was.
Would this sign no longer ruin mosfets?

[davekni]

It still looks like there may be multiple transitions on the GDT input (driver output) with the HC14 connected.  I suggest scoping one of the GDT inputs.

The scope's display of DC coupling appears to be the trigger coupling, not the input channel coupling.  I think you are still in AC-coupled input mode.

[John123]

Would some kind of extra filtering or hysteresis help in this case around the 74hc14?

Would this sign no longer ruin mosfets?

Hard to tell, I'd remove the big 400v electrolytic bus filtering capacitor for now and current limit/put a power resistor in series with the supply for testing, that would prevent damaging current spikes in the event of a fault. Have you included the zener gate protection yet?

Also how much of that breadboard is being used here? Those wires jumping around could be coupling into your antenna. Have you got any solderable prototyping board? That would be better and cut out much of the parasitic inductance and capacitance associated with breadboard and jumper wires. I've always had problems using breadboards with this sort of circuit.

[balazs]

I bought zener diodes and resistors and built the protection. The signal became more beautiful again, and the ringing almost disappeared from it. The oscilloscope has been in DC mode so far, as shown in the first image.
I also measured the signal at the output of 74hc14, this is the second picture. The third image shows the output of the GDT.

[davekni]

Yes, the 74HC14 scope trace clearly proves the vertical channel is DC-coupled.  The "~" symbol is really strange given that.  Thank you for proving me wrong!

Signals are looking good.  Hope everything continues to go well!

[John123]

Much better! Are the zeners getting warm?

[balazs]

Without Toroid, it worked perfectly, worked for a long time, and had no problems. Then I made a box. The sign here wasn’t that nice, there were falls in it, I didn’t take photos, but it wasn’t big enough
to deal with it. When I first tried it without a totoid, it worked perfectly. Then after I put on the toroid and the interrupter and duty frekvenci, it burned down, the 2 mosfet. After I replaced it, the signal became more beautiful again.
 I don’t know if that could have caused the death of these mosses.

[balazs]

After further attempts, strange falls occurred in the signal. Why can this be?

[davekni]

Your half-bridge is nicely constructed for low parasitic inductance.  However, I expect you are now experiencing issues with noise coupling into driver circuitry.  Circuit boards with ground planes are ideal.  You may be able to get away with your existing construction if sufficiently shielded.  Does the gray box contain the driver circuitry (circuitry feeding GDT)?  Is the gray box plastic or metal?  If metal, is the metal connected to ground?

BTW, for white plug-board construction, here's an example of the best I've seen, with short wires flat against the board and bypass capacitors directly between chip power pins:
https://highvoltageforum.net/index.php?topic=1527.msg11899;topicseen#new

Boards with ground planes are still better then even the best-constructed plug-board implementation.

[balazs]

Thank you for your answers
I shielded the driver, I tied the shadows into the protective earth, the box was plastic. The falls have disappeared from the signal. I tried the coil and it was 1000 W. I tried for about 15 minutes. The strange thing was that the circuit breaker transformer was broken. After that, the coil was still working. Then one of the starts gave no spark and slammed the fuse. This was very strange because so far when the mosfets have burned, I have either started the coil or changed the frequency of the circuit breaker. I noticed a strangeness when measuring the signal. When it is
the spark gets into the grounded part of the secondary coil, the signal at the GDT output is sometimes distorted, which is pretty weird. Cannot be these faults because the outside of the bnc connector transmits the negative signal of the interruption and the protective earth? (positive sign of interrupt runs inside)
I do not tie the secondary to the protective soil.
I would have another question on how to measure the GDT output with 230V operation, because it may not be a good idea to change the oscilloscope to -320V.

[davekni]

If you are getting sparks down to ground with an SSTC, that is impressive performance.  Do you have a breakout point?  Perhaps aim that away from ground (opposite side of top-load).  Also, either add a ground ring as in DRSSTC designs or back off power a little and enjoy your coil.

[balazs]

I modify the primer and also add a grounding ring.
How far should I roll the primer from the secondary?
The coil worked pretty well, but I saw a little spark on top of the antenna. Because of this, I placed the annanta a little further away, from where the mosfets deteriorated after a while. How long do I have to place the annanta? What a problem if there is a small spark on the top of the antenna.
Can I connect a 10x 300 V oscilloscope probe to the GDT output and "VBus-" where "320v-" is?

[davekni]

Since you are not getting arcing from secondary to primary, I'd leave spacing as it is.  No magic answer for perfect primary-to-secondary spacing.

If it works better with the antenna in its original closer position, leave it there.  To reduce corona at the top, make the end rounded, such as forming the end into a loop or adding a small metal sphere to the end.  Just make sure the antenna isn't so close that an arc forms from the secondary top to the antenna.  An insulating sphere can work too, such as drilling a small hole half-way-through a solid rubber ball and sliding the ball over the end of the antenna.

If your scope is battery-powered and not connected to anything else (not to a computer and not touched by a person), then you can scope with the probe "ground" to VBus- (-320V) and tip to gate.  If you measure two signals, the other probe "ground" lead must be to the same VBus- node.