ZVS Mazzilli circ with active gate driver - what causes ringing?

[82026]

Hi all,
Trying to make ZVS mazzilli circuit with active gate driver and getting bad ringing at switch over. While trying to debug it, found most noisy point to be on collector of Q401(Q402).
Here is my circuit attached with scope traces of that point and of ouputs. Could somebody experienced give a hint what causes this ringing and how to fight it?

[flyrod]

What does the signal on the other end of R408 look like?

What does your circuit layout look like?  For example, how long is the wire between C422 and Q406?

Also, out of curiosity, how did you come up with that circuit design?

This site will only let me post about once every other day.  Maybe try asking over at 4hv.org

[82026]

here is layout.  Traces of R408 are short  as well as C422 to Q406.
nearly all bottom layer is poured with GND. Could it be a problem? 
The circuit is from:
https://www.elektormagazine.com/magazine/elektor-201611/39970
(click download article and see Figure 5)
This ciruit seems to be tested to work and it does work for me but the ringing creates noise which I want to avoid. 

Also I'm reposting this topic at 4hv.org but they do no allow me to post pictures (circuit and and layout)

[T3sl4co1l]

Why do people make this circuit with all the extra junk?

It's an analog circuit.  It only works because it it has real linear gain at the resonant frequency.  If you try to sharpen up the gate drive edges, or, I'm not even sure what's being done here, some biasing/level shifting or local feedback thing, you're just making it more liable to oscillate at other frequencies, potentially very high frequencies that make it burn up!

The underlying circuit by the way is the Baxandall oscillator (or more accurately, one of the equivalent topologies presented in his titular paper); it's also commonly misnamed "Royer" but that actually refers to the voltage-fed, saturation-commutated version, with an otherwise very similar circuit.

Here's an example of mine which is unfortunately a terrible example for discussing the circuit,



it has basically the same circuit except Q401, Q405, D403 and R416 are deleted (Q401 shorted through C-E), and R408 = 1k.  And equivalent for the other side of course.  There's also a 0.47uF for the VCC bypass.  Runs at ~600kHz, a pair of FDP33N25 I think.

Tim

[82026]

Hi, could you post a sketch of your circuit - to better understand how it looks/works.
I found identical gate drive circ described on page 53 of this paper:
https://pdfs.semanticscholar.org/1265/19da946b5df9b8941a521d5a7019720bcd0d.pdf
However not fully studied it yet...



 

[82026]

ok, made modifications as T3sl4co1l suggested. The ringing is gone and efficiency (used current) stays the the same. issue seems to be solved.

[T3sl4co1l]

Hi, could you post a sketch of your circuit - to better understand how it looks/works.
I found identical gate drive circ described on page 53 of this paper:
https://pdfs.semanticscholar.org/1265/19da946b5df9b8941a521d5a7019720bcd0d.pdf
However not fully studied it yet...

Looks to be a cromulent paper as such, but seems to lack motivation, innovation and results (the poor efficiency at the end is particularly unimpressive).

That is to say, it looks to have all the trappings of an academic paper; but doesn't seem to have much analytical or engineering value.

Tim

[flyrod]

Why do people make this circuit with all the extra junk?

So the circuit you're describing is basically the one posted here:?

https://highvoltageforum.net/index.php?topic=530.160

This was reverse engineered from an ebay board.

82026:  Did you add a bypass capacitor in addition to removing some parts?