I've got a flyback transformer circuit, that was designed by a firm and they tended to over-engineer things (this is in a commercial project).

I re-laid out their circuit and on the primary end of the flyback circuit they have both an RC snubber and a TVS Diode protecting a MOSFET from transient voltages and for inductive loads when the MOSFET closes.

I've actually torture-tested this circuit and removed both the snubber and TVS diode at the same time and no magic smoke pours out of the FET as I am applying the maximum current with a PWM pulse to the flyback transformer (ie. it is receiving current, right below the saturation point and the peak duty cycle -- ie. transformer torture test).

Is there a reason you would want both, do they seem superfluous?

I could see a case where perhaps the RC snubber is protecting the MOSFET, and even though the flyback circuit can run for hours you are prematurely damaging the MOSFET by removing the snubber?

I really don't see the case so much for the TVS Diode... I have no idea what a transient voltage would look like in a transformer failure, but if the transformer fails the board is pretty much trashed anyway from a product perspective.

  • \$\begingroup\$ RC snubbers are usually used to limit EMI, not protect the MOSFET. Your idea of over-engineering may be suspect.. \$\endgroup\$ – Trevor_G Sep 12 '17 at 16:01
  • \$\begingroup\$ That's a good point, I looked at this two weeks ago I forgot about that as a benefit but did note it as I went through. I have some EMI probes I could at least see if the noise factor adjusts much (relative, I can't make a very good absolute measurement without a farraday cage...). \$\endgroup\$ – Leroy105 Sep 12 '17 at 16:09
  • 1
    \$\begingroup\$ Doesn't the term flyback mean anything to you? That's it's operating mode and any leakage inductance on the primary means you get flyback at the primary that must be quenched by one means or another. Sometimes the self capacitance of the mosfet can keep this limited but, you are definitely risking the longevity of the supply. \$\endgroup\$ – Andy aka Sep 12 '17 at 18:13
  • \$\begingroup\$ Well yeah... but to what extent, this a VERY CHEAP commercial product... other commercial builds (from a $500MM company), I've seen the RC snubber omitted and simply a TVS to shunt transient surges when the FET closes. I agree there may (probably is) some leakage when the FET closes, but how do you quantity the limited lifespan of the MOSFET...? I can torture test it for 1,000,000,000 cycles -- if it passes, omit both? I kind of expected a smoking FET when I pulled the RC snubber... but no magic smoke. \$\endgroup\$ – Leroy105 Sep 12 '17 at 19:07

Without a schematic, it is difficult to judge. A \$RC\$ snubber is different than an \$RC\$ damper. What is possible is to have the parallel combination of a damper and a TVS as shown below:

enter image description here

The \$RC\$ network is there to damp the resonance of the leakage inductance and the capacitance lumped at the drain node. Sometimes, the leakage contribution is so energetic that when the TVS series diode blocks, the ringing on the drain can go below ground and forward-bias the MOSFET body diode. Not a big deal with a discrete MOSFET but it can inject carriers into the substrate if we talk about a switcher featuring a lateral FET. By dissipating energy, the \$RC\$ network will damp the resonance and keep the leakage ringing under control. Reverse-bias of CMOS-based switchers are a cause of destruction and must be avoided.

A TVS is a) an expensive device (compared to a classical \$RCD\$ peak clamp and b) it radiates a significant amount of energy as the leakage inductance reset time is very small. For all of these reasons, TVS are rarely used in commercial products. However, again, if you have a switcher with an integrated FET, as its body diode is lightly doped, it cannot accept accidental avalanche and a hard clamp is truly needed, hence the TVS presence. Please keep in mind that beside delivering power at nominal input, these power supplies (at least for quality models) have to go through numerous input surges tests and must survive them. Sometimes, protection components that look superfluous are there to pass these tests and keep silent for most of the time. I would not remove them : )


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.