Hi AAC,

I am building a flyback DC/DC converter whose driver circuit is being fed from the same supply as the converter itself. I am not using the AUX winding option. See attached schematic and oscilloscope results. The scaling factors are correct.

Background: When I turn off on the flyback switch, there are switching transients at VDS and the supply due to parasitic leakage current being circulated around the circuit – since that parasitic resonant current has to go somewhere, obviously. This is a fact of DC/DC circuits I cannot change too much. But what is especially causing me a headache is the fact that these same turn-off transients are reverberating all over my control circuit as well. This causes problems in the gate drive. I’m running into problems of accidental FET turn on, FET avalanching, and transformer and FET overheating issues. I have attached a schematic and test results in this circuit to show for example how transients can reach my +5V LDO supply – which is the source of controller problems downstream. The green probe in my circuit is a differential probe and unfortunately I only have one. The brown overhead pictures show some places where these switching transients are popping up. My transformer operates heavily in DCM as you can tell. If I can make it operate closer to critical conduction mode I think that would be good. Although I'm also trying to figure out how to do that effectively since my load is very highly variable.

Question: What schematic, and/or PCB layout tricks can I do to prevent, or at least drastically mitigate even more, the impact these transients can have on the driver circuit feeding the FET gate? I’ve flooded the switching node and common planes as much as I could on the PCB, tried separating high current grounds from analog/digital grounds, and added some filtering to the DC/DC brick. What can you recommend for other protection/transient suppression schemes in order to protect my controller and MOSFET? Further, remember that these are repeated transients at 165kHz (+/- 75kHz) so the protection would have to operate (almost) continuously. You can ignore the magenta waveforms in my test results. Any suggestions and brainstorming is very welcome.

 

How much power do you expect from the converter?

You have apparently removed the RCD snubber across the input winding of the inductor. What is your expectation of disposition of the the energy stored in the leakage inductance? With no snubber the FET will avalanche - there simply is no other place for the energy to go (not strictly true; small amounts will be "eaten" by capacitances). This is permissible if both the current and repetitive avalanche energy are within the ratings of the the FET.

Layout is very critical with switchers. Simply pouring copper will not necessarily help. Minimizing loop areas and unwanted inductances is critical.

Probing switchers is not easy. With ordinary passive probes the ground leads in resonance with the tip capacitance will produce all sorts of artifacts. Tip probe grounders are essential. With low voltages an active probe can be used with a ground lead, which is a lot more convenient, but you can destroy an expensive active probe instantly if you put it on a voltage that is too high for it.

 

- 50W
- The reason I got rid of the primary snubber was due to previous heavy problems with the snubber overheating dramatically. Imagine 90'C on a 6.5W resistor. I know the snubber's role is to solely mitigate the ringing only but it appears it is also eating up energy from the resonance of the XFMR magnetizing inductance and Coss while in DCM. At least as far as I can tell - why else would it be overheating the way it was?
- I've read about layout tips online and am aware of this. I tried my best to minimize the loop areas even though some of the components are geometrically large. This is my first switching converter project so I'm mostly relying on my judgement and how it relates to the literature I've read.
- I am choosing to use a differential probe (Micsig DP10013) for measuring everything on the transformer primary, and its respective driver. It's being fed from a -48VDC source with a floating common on the driver due to an EMI filter. Would you still recommend a probe ground clip to be used on the return lead of a differential probe?