Guru
Joined: 21/06/2018
Location: AustraliaPosts: 1225
| Posted: 01:25am 02 Mar 2019 |
A great presentation and informative analysis of FET switching behaviour. In my experience individual chokes in series with each of the paralleled FET gates in the form of a ferrite bead or 5-10uH choke often helps quench HF oscillations - these can be in the MHz range.
In the second to last CRO picture, the yellow gate display appears to show that the gate was not fully off when it reversed direction into the oscillation area. If I am reading the CRO correctly the fall of the gate transition is ~ 500nS, relatively slow.
A parallel diode across the gate resistors (cathode facing away from the gate) should help speed up the fall time, a 1N4148 may be sufficient but I usually use a power ultrafast diode such as MUR160 or UF1004. A lower value resistor may also help. - but these both create a higher drive requirement of the Gate drive IC or buffer used. I have found that the slower the change of gate voltage is as it passes through the miller region can make it more prone to oscillation issues.
When there is a fast transient on the Drain, it is coupled back to the Gate. The Gate resistor forms a differentiator with the Drain - Gate capacitor and a combination of fast transient and insufficiently low Gate impedance (series r - even with the driver output low as you found)) can definitely drive the gate sufficiently to cause the FET to begin unintended conduction.
If at first you dont succeed, I suggest you avoid sky diving....
Cheers Mike