So by now you’ve probably connected power backwards to a board and let the magic smoke out. So here you are wanting to protect your design. Let’s go over some methods of getting this done and comparing them! This is meant as an introduction and to assist in circuit topology selection. This guide isn’t going to give information on component selection; hopefully that can be found on Google or ChatGPT. If not I may have to come back for another round.
The first solution is the classic “do nothing option”. Every method of keeping your board safe from reverse polarity being applied to it’s power terminals involves placing a semiconductor in your power path. This is unavoidable as protecting downstream components from a voltage means not conducting it while normal correctly applied voltage needs to allow for current flow. All semiconductors will introduce some loss of power (as heat) when conducting current (in this case, when your board is functioning normally).
Next up we’ll go with the simplest solution, a series Schottky diode. The diode will be placed between your input power net (i.e. from a connector) and your load (i.e. the rest of your PCBA). I’m too lazy to find some photos of this, but can be easily Googled. The concept of operation is fairly simple, if the input voltage is applied in reverse the diode will become reversed bias and will block the current and during normal operation will conduct. However, diodes have this little thing called a forward voltage drop. Let’s say your load draws 1 A at 5 V (5 W) and we assume a diode forward voltage drop of 0.7 V (depends on the diode you pick here and a bunch of other things, this value is a reasonable ballpark) this means you’d need an input voltage of 5.7 V and your diode is consuming (P=I*V=1*0.7) 0.7 W. This would mean your reverse polarity protection circuit is consuming 15% of the power that your load is consuming. This is a significant amount! This circuit can be implemented with a single component so it’s simplicity is it’s primary selling point.
Next is the series PCH. This circuit is a PCH FET in series with your input power path, a resistor pulldown on the gate, and a Zener diode on the gate to source of the FET. PCH FETs have much lower losses than a series diode, and this circuit is only three components. Running some more calculations, I’ll ballpark a PCH RDSon to be 100m ohm (once again this varies based on a bunch of things and this is a reasonable ballpark). Same load a before the FET conduction losses would be (P=(I^2)*R=(1^2)*0.1) 0.1 W which comes in at 2% of the power the load would be consuming. Much better than the series Schottky, but slightly more complex to design since it’s got a few more components.
Finally we have the classic series NMOS. This circuit is a series NCH FET in the power path. This circuit requires a charge pump to saturate the FET, however, implementing this is done by many ICs so you don’t need to do this yourself! Further, some ICs, especially for lower power applications simply put the NCH FET inside the chip itself. Realistic NCH FETs for this application could have RDSon is between 20m ohm all the way down to 0.1m ohm so much less than a PCH. If you crave even lower ESR you can simply run a few of these FETs in parallel to reduce losses even more! When following a datasheet’s instructions designing one of these circuits can be surprisingly doable. Adding an IC to the circuit will increase cost, but it is often worth it. Component count will end up about the same or more than the series PCH circuit.
There are other options, but these are most common. Feel free to add more to this document if you find something nice! All of these options can be simulated in LTSpice or Simplis for a deeper understanding (though getting a model for a specific IC can be annoying).