I am fairly new to circuit design. I was wondering if I have correctly implemented the TVS diode (D8), flyback diode (D6) and the snubber circuit (R20 and C41) for this buck converter. I apologize in advance if the circuit design is messy/offensively incorrect! The only component downstream from this buck converter is a gate driver.


  • 2
    \$\begingroup\$ The circuit is messy and offensive. I like the power flow from left to right. The whole circuit should be basically mirrored. Start with IC1. I like the grounds pointing down when possible (which is your case. And definitely not pointing up). Inputs are best on the left. Outputs to the right. \$\endgroup\$
    – MOSFET
    Commented May 23 at 2:17

2 Answers 2


First, D8 isn't a TVS diode. and if it is, it's the wrong symbol. NBD, rookie mistake. Also, this isn't where you would typically put a TVS.

Flyback diode (D6) is in correct location and orientation. Snubber (C41 and R20) is also in the correct location.

D8 is redundant and doesn't really make sense.

Also, you are probobly going to want some more input capacitance and a small ceramic chip capacitor to decouple high-frequency power supply ripple that will cause the controller to behave erratically. Everything else seems fine as a first pass reviewing this highly offensive circuit.

Pro tip: If you have little experience drafting the schematic on your own, just start by copying the application circuit in datasheet; tweak from there. In first-order, it's going to be drafted perfectly because the app circuits are drafted by seasoned professionals. This makes everything from a communication standpoint clearer for all parties involved. With more experience, you will be able to draft the circuit from just the snippet circuit on page 1 of the datasheet. Not just for this chip, but all the chips. They all follow this paradigm, give or take.

  • \$\begingroup\$ Thank you for your fast reply. I did use the default diode symbol for KiCAD, I apologize! Is the TVS diode necessary, or would it be redundant in any location? If so, would it be placed on the SW node of the buck converter rather than after the inductor? Also, the "typical application" schematic I found in the datasheet does not include a snubber circuit, TVS diode, or flyback diode. Perhaps I am overthinking and do not need such protection? \$\endgroup\$
    – NickRand
    Commented May 23 at 2:55
  • \$\begingroup\$ @NickRand I've never seen a TVS on a switched-node in any power supply design - You typically use TVS diode for protecting sensitive inputs, and CMOS pins including MOSFET gates on some high-power designs. Or other pins that can't tolerate hv transients, especially if they connect to "the outside world" AKA off the board. The switched-node is self driven and low-impedance, it can support its own transients. For the app circuit those parts aren't needed - you get the flyback diode for free from the internal low-side switch. The snubber is needed for tight emi reqs or if you effed up the layout. \$\endgroup\$
    – MOSFET
    Commented May 23 at 3:04
  • \$\begingroup\$ Ahh, the entire purpose of implementing the flyback diode was to counter the problem I read about previously with inductive kickback when power is interrupted, i.e. shutting off the system. Perhaps decoupling capacitance would help with this in any case, allowing the power through the circuit to ramp down gradually. I was unaware that the buck converter was able to solve this regardless of whether or not this is an issue. Thank you very much for your help and your patience! \$\endgroup\$
    – NickRand
    Commented May 23 at 3:25
  • \$\begingroup\$ @NickRand "when power is interrupted" that's a different problem. What we are talking about is during steady-state operation. Related. But not the same. See pg 8 in the datasheet. the diode is already in the package (in the LS-FET). The decoupling cap doesn't help this in lieu of the diode - it serves a different prepose - it stabilizes the power rail so the IC can operate as designed. The flyback effect is intrinsic to the design of all buck converters. It's not a transient effect akin to turning off a solenoid. \$\endgroup\$
    – MOSFET
    Commented May 23 at 3:36

Always start with the datasheet:

The TPS62136 and TPS621361 are high efficiency and easy to use synchronous step-down DC-DC converters [ed: emph]

The application diagrams show no catch diode, nor RC snubber; D6, and likely R20-C41, are superfluous.

A catch diode sometimes shows up in synchronous designs, when the low side switch isn't always driven synchronously, or to mitigate body diode recovery due to excessive LS-to-HS dead time; but mostly it's a hack, and dealing with the underlying problem is more effective (i.e. making it fully synchronous, or tightening the dead time). In a normal functioning synchronous design, the diode does nothing useful, and simply increases capacitance on the switch node, and therefore increases switching loss.

You could add the R+C to mitigate emissions somewhat; but mind this comes at the cost of efficiency, and is better solved in general by careful layout and appropriate component choice. It should be an unusual/exceptional case where both the snubber is needed, and provides enough value to be useful -- certain automotive EMC tests perhaps (near / E-field test?).

There may be notes about the capacitor choice as well, but their types will need to be given. To summarize: type 2 (X5R, X7R, Y5P, etc.) ceramic capacitors exhibit a reduction in value under DC (voltage) bias. The amount of reduction is independent of the voltage rating, so it is not enough to simply put in, say, 50 or even 100V rated capacitors. You must look for the characteristic data of the parts, and ascertain whether they are sufficient for the application. The main outcome will be: for 10 and 22uF parts, even at 3.3V, you will likely need 1206 chip size at least, and preferably 1210, and of a grade no worse than X5R to X8S.

No layout is given, but do pay close attention to the suggested layout in the application section. Layout is critical: every mm of trace length contributes a good 0.5nH or so of stray inductance. May not sound like much, but it adds up quickly -- pin and component body lengths count just as well as PCB traces do -- and at the edge rates in regulators such as these, it doesn't take much, maybe even 5 or 10 nH, to cause dubious behavior -- or outright malfunction or damage.

  • \$\begingroup\$ Thank you for your response. The "typical application" schematic indeed has no snubber circuit, TVS diode, or flyback diode. As this is my first circuit design, I am likely overthinking in terms of protection. The gate driver, which this buck converter is powering, is connected VIA wire (app. 5 inch) to four external PCBs which house their respective MOSFETs (one per PCB). Perhaps I should focus on adding flyback protection at the output of this gate driver instead? As for your notes on capacitor selection, I am using Murata's "SimSurfing" tool for selection. Will this suffice? \$\endgroup\$
    – NickRand
    Commented May 23 at 3:06
  • \$\begingroup\$ SimSurfing shows DC bias, yes. You may also want to ask a question about your application generally, as lone MOSFETs sounds like a layout nightmare; if you do, please provide details of what they're powered by, what they're connected to or switching into, how fast/often (edge rate and frequency), and the layout (including PCB diagrams or wiring photos). \$\endgroup\$ Commented May 23 at 12:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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