[SOLVED] Mosfet ringing at switch ON step

[Monolith]

When I switch the IRFP640 N Mosfet ON, I get some ringing. My guess is that this is due to the Miller effect.

How can I avoid it?

 

[treez]

Switch it on slower by rc filtering the gate ....but maybe the ringing does not matter, do you have scope shot? schem?

 

[KlausST]

Hi,

Maybe a impedance problem.

The loop: FET source - driver GND - driver Vcc capacitor - driver output - to FET gate. And bootstrap circuit.
Needs to be very short, wide traces, low inductive.

A low inductive resistor in the gate line could improve stability.

Klaus

 

[treez]

also beware of scope noise...esp when you have long dangling ground hook.

 

[Monolith]

Scope picture shows Mosfet driver output, yellow, Mosfet gate red trace. ON voltage about 13V.

- - - Updated - - -

Yellow trace is Flyback, red trace is gate voltage.

Trying to switch at 0 Volt, ZVS

 

[FvM]

Firstly. A certain impact of miller effect is normal operation. What's your specific problem with it?

Secondly. We need to see the actual gate driver circuit to understand how far "driver ouput" can be different from "Mosfet gate" voltage. Ringing suggests existence of wiring inductance, which can be generally considered "avoidable". But may be not avoidable for your design.

- - - Updated - - -

The new waveform needs additional explanation respectively circuit information.

If this is actually drain-to-gate crosstalk (could be, because the capacitance is very nonlinear and considerably increasing near Vds = 0), there's apparently a problem of too high gate driver impedance.

 

[Monolith]

This is the simplified LTSpice simulation circuit. The actual circuit uses a Mosfet driver MIC4452

- - - Updated - - -

Firstly. A certain impact of miller effect is normal operation. What's your specific problem with it?

Secondly. We need to see the actual gate driver circuit to understand how far "driver ouput" can be different from "Mosfet gate" voltage. Ringing suggests existence of wiring inductance, which can be generally considered "avoidable". But may be not avoidable for your design.

- - - Updated - - -

The new waveform needs additional explanation respectively circuit information.

If this is actually drain-to-gate crosstalk (could be, because the capacitance is very nonlinear and considerably increasing near Vds = 0), there's apparently a problem of too high gate driver impedance.

Thanks for the replies.
Yes, it is a crosstalk problem. The gate driver is supposed to give 12A peak current at 18V, (but running at 13V now) however, adding 10 Ohm in series seems to give less ringing.

 

[FvM]

adding 10 Ohm in series seems to give less ringing.

Then it's primarly a wiring inductance problem, I think.

 

[Monolith]

Then it's primarly a wiring inductance problem, I think.

Thanks for the help.
This problem occurs on the prototype perforated circuit board. Distance of the Mosfet driver to the gate is about 40mm. Resistor length of about 20mm is included in that. Now drawing a PCB, SMD, with minimum distance between driver and gate. Trying to avoid the problem.

 

[KlausST]

Hi,

Resistor length of about 20mm is included in that.

What resistor? Could it be a high impedance wire wound resistor?

Klaus

 

[Monolith]

Hi,


What resistor? Could it be a high impedance wire wound resistor?

Klaus

I had a wirewound resistor of 1.8 Ohm and got oscillations. Changed to what I think is a carbon 3W, 10 Ohm resistor which gives what you see on the pictures. Tried several different kinds of Mosfets. Each kind behaves a little different.

 

[KlausST]

Hi,

don´t use wirewound.

Klaus

 

[FvM]

Im unable to locate a resonance frequency of about 2 MHz in the gate circuit, would require several µH wiring inductance which isn't feasible.

I think, we see something completely different, a resonance of the 40 nF capacitor with a 150 nH capacitor ESL + ground loop inductance, series resonant circuit closed through the MOSFET. It's coupling to the gate circuit, or at least your gate voltage probe.

We should look at the topic again when you have finished your final design with regular PCB wiring.

 

[D.A.(Tony)Stewart]

Impedance control gives best control of ringing due to transient CISS during transition.

THe ratio of source to switch to load impedance has a penalty of rising CISS with lower RdsOn.
I prefer to choose a ratio of 50:1 max from driver to switch impedance. Then asymetry is another choice due to shoot thru, so lower impedance turn off and higher turn on is preferred to assist in dead band of complementary drivers or half bridge.

Then use very short leads or none with coaxial pin and ring of 10:1 probe.

Also make sure C4 is very low ESR/low ESL path. ( <<10mΩ) such as film.

 

[Monolith]

Thanks for all the good advice.

Smooth, noiseless switching is very important for my application.

Some of the new SiC Mosfets apparently have much less Miller effect. Is there anybody with experience in using these?

What about using 2 Mosfets in Cascode?