Transient Troubles

[Acid Byte]

Hey guys and girls,

previously posted halfbridge is encountering some transient trouble.
i am working on cutting every wire as short as possible (so total redo on the layout)
but i think the problem could be caused by a different source.

the problem always causes the mosfet protection diode to short out. (MUR860)
they dont  go both at the same time. always one or the other.
still this is no expensive fix, the coil is just not reliable enough.
sometimes it runs for days without destruction.
sometimes just minutes..

here in the netherlands there is alot of 50hz EMI,
to be able to get rid of this i used a simple RC highpass filter wich got rid of the 50hz (so far it works)
now i know the RC highpass introduces a shift in phase.
could this cause the problem as my sstc is not switching ZVS because of the phase shift.

is there a way to get rid of the 50hz EMI without introducing a phase shiift into the driver?
for this sstc im using this schematic


note that the antenne input has the RC filter inbetween antenne and 100nF cap.

greetings Jeroen

[Mads Barnkob]

I really doubt that 50Hz transients would disturb your driver circuitry, with stepdown transformer to 24VDC to supply it and enough filtering, there should be no problems there. The antenna feedback can be very unstable and if a sudden phase shift is introduced it could be hard switching more than it can handle and blow up.

Are you only blowing up the MUR860 diode or do you also kill the MOSFET?

Did you check the gate resistors after each failure? It could be that you are only running on one MOSFET and conducts it all through the other diode if its MOSFET is not driven, and that is why it fails.

[Acid Byte]

I really doubt that 50Hz transients would disturb your driver circuitry, with stepdown transformer to 24VDC to supply it and enough filtering, there should be no problems there. The antenna feedback can be very unstable and if a sudden phase shift is introduced it could be hard switching more than it can handle and blow up.

Are you only blowing up the MUR860 diode or do you also kill the MOSFET?

Did you check the gate resistors after each failure? It could be that you are only running on one MOSFET and conducts it all through the other diode if its MOSFET is not driven, and that is why it fails.

I was only shorting the mur860 mosfet always survived.
so cause was over voltage through the mur860.
i skipped using the mur860's and start using the p6ke 440a instead.

redid the full layout (still not perfect but way better)
was able to cut out almost 1 meter of wire
placed the snubber as close to the mur860 +rail and source of the -side mosfet (where the other mur860 ends up)
skipped the antenna and put current transformer in its place in series with a 1kohm resistor.

this fixed 2 things.
i got rid of the 50hz buzz cause of the better layout and better grounding.
so far no more shorted protection diodes.

because i use the ct for feedback now without a input filter i got rid of the phase shift and significantly decreased the transients.
just like you explained in your drsstc guide.

Updated layout.

[Mads Barnkob]

You are on the way and you are right, it is not perfect The path from DC bus capacitors to switches to primary coil should be as short as possible and with the lowest possible impedance at the frequency its switching at, this usually translates to flat copper bar to utilize it as much as possible due to skin effect.

Snubbers really has to sit right on the terminals of the MOSFETs/IGBT, they have little to no effect if they are mounted on a long piece of wire away from the switch.

Did you do any measurements before and after since you know that transients are gone/reduced? Or just a guess since its not blowing up?

[Acid Byte]

To be honest i only scoped it up after.
Though before transients were able to short out the protection diodes near 100vdc input.
Now on 100vdc input that peak in voltage is almost gone.
Instead there is much lower frequency back to take its place.
The peak isnt as high as it was now reaching 160vdc before falling.
This means there is still some transient but it isnt way as high as it was before.
The only weird thing now so far is the oscillation i see keeps on happening even when that particular mosfet is conducting.

[Mads Barnkob]

What you are witnessing here is a relatively unknown subject in the amateurs world of building inverters, for MOSFETs and IGBTs there is a Typical Capacitance vs. Drain-to-Source (or collector-emitter) Voltage that will tease you with really high switching spikes at low voltages, as the junction capacitances fall with higher voltage across the device.

So where you see destructive switching spikes at 100 VDC, those might be low and completely acceptable at 320VDC. See Fig 5 in this datasheet for a IRFP460 MOSFET ( https://www.vishay.com/docs/91237/91237.pdf ) and you can see that the output capacitance Coss is as good as gone at 300VDC compared to 100VDC.

[Acid Byte]

What you are witnessing here is a relatively unknown subject in the amateurs world of building inverters, for MOSFETs and IGBTs there is a Typical Capacitance vs. Drain-to-Source (or collector-emitter) Voltage that will tease you with really high switching spikes at low voltages, as the junction capacitances fall with higher voltage across the device.

So where you see destructive switching spikes at 100 VDC, those might be low and completely acceptable at 320VDC. See Fig 5 in this datasheet for a IRFP460 MOSFET ( https://www.vishay.com/docs/91237/91237.pdf ) and you can see that the output capacitance Coss is as good as gone at 300VDC compared to 100VDC.

so your saying it could pretty well be that by the time i get to 310vdc the spikes are almost gone and have nothing to worry about anymore?
would be great hehehe
i just dont dare to scope it when running on mains.
dont have the propper equipment to do so.

thank you for your response!