I'm designing an industrial board that will be powered with a +24V DC. On the PCB there are several fuses to protect each branch against over-current/short-circuit. Example:


simulate this circuit – Schematic created using CircuitLab

Of course the resistors simulates the actual loads: MCU, relays, LEDs, ... The 24V is applied through screw terminals.

I need to protect the circuits ("loads" in this examples) from ESD. As far as I know the ESD protection should be placed very close to the input connector.

So the obviously solution is to add an unidirectional TVS at the input:


simulate this circuit

Usually I put a fuse before the TVS so it can also be effective on persistent over-voltages or inversion of polarity. But here I don't think it's worth to add another fuse!

Is a single TVS enough for ESD protection? Is there a better (= safer, more reliable) circuit to use in this scenario?

By the way, I don't have a real "ground", so I need to close the discharge to the common voltage (0V).

  • \$\begingroup\$ If you don't have a ground (i.e. an earth) then that doesn't mean the ESD discharge magically uses your 0 volts. That wouldn't be correct unless your 0 volts is earthed. In all unearthed scenarios the ESD discharge flows through your exposed terminals via parasitic capacitance to real ground (earth). \$\endgroup\$ – Andy aka Jan 21 at 15:14
  • 1
    \$\begingroup\$ re: no earth -- make an "FE" terminal on your board, and if it goes into an insulated box with only an unshielded cable connecting the box to the outside world, that terminal simply won't get hooked up. \$\endgroup\$ – Pete W Jan 21 at 20:35
  • \$\begingroup\$ Generally, fuses is mostly a thing for high current electronics. If you add a 1A fuse to protect a PCB drawing some 100mA, it will achieve absolutely nothing. 1A short circuit current will fry the whole board long before the fuse blows. That's enough current to burn off traces, bond wires, everything. \$\endgroup\$ – Lundin Feb 2 at 7:21
  • \$\begingroup\$ @Lundin, where did I write that a put a 1A fuse to protect something that draws 100 mA? \$\endgroup\$ – Mark Feb 2 at 7:46
  • \$\begingroup\$ You didn't. But in case your device draws 950mA the fuse would be equally pointless, for other reasons. \$\endgroup\$ – Lundin Feb 2 at 7:58

here's the general pattern.

Ideally, AC-couple the ground plane to FE or chassis. Also think through the consequences of how this interacts with all the non-power signal connections, and every conceivable way of the field wiring being done wrong (often multiple mistakes at once due to wires being swapped or shifted). A common mode choke may be called for too, for EMI, depending on details - not shown here.

enter image description here

  • \$\begingroup\$ I used something similar in the past when the currents were lower (< 1A). Now I have to handle over 8A... it's just a matter of sizing the components correctly? And of course, thermal considerations. But can I still use a series Schottky and choke? \$\endgroup\$ – Mark Jan 22 at 5:42
  • 1
    \$\begingroup\$ "just a matter of sizing the components correctly" -- Mostly. All the series components first of all. The damping R as shown might become impractical. You would have to consider the dI/dt, which depends on what the whole thing does. If it is primarily on the inrush, an active limiter of some kind is something to explore. // The ESD passives have the same function though. Eat on the order of 100A burst for a few ns, while the diodes turn on. \$\endgroup\$ – Pete W Jan 22 at 14:00
  • \$\begingroup\$ please would you mind to explain why is useful to put a bidirectional TVS followed by a unidirectional one? \$\endgroup\$ – Mark Feb 2 at 7:47
  • 1
    \$\begingroup\$ @Mark -- it was recommended by some app notes, for 2 reasons. First, the bidirectional TVS on the outside of the series diode survives the common accident of connecting 24v power backwards, whereas a unidirectional would conduct in forward mode and would either burn out or trip 24V power supply limit, turning off other devices. Second, in a burst the outside TVS clips the V waveform to about +/- 100V ( on ns timescale) and about 45V on longer timescales. The the series diode then rectifies that to a train of + pulses. The second TVS in combination with capacitance cleans that up. \$\endgroup\$ – Pete W Feb 2 at 14:25

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.