The simple switchers are great. Just pay close attention to the switching loop and keeping it tightly routed, otherwise you can create EMI issues for yourself.
I’d personally go with something like a 7805 drop in replacement buck converter like I linked above for simplicity, but I can’t remember if they go up to 3A.
It depends on your priorities but I usually would. It adds components/complexity/cost but saves you from silly mistakes. If it takes $1 and 10 minutes to add but will save you from wasting $20 and an hour on the 10% chance you do something absentminded later, I’d say it’s worth it. It’s always about tailoring the solution to the problem and there’s no one-true way of doing anything. You can implement it super cheap or make it expensive and overkill.
The simplest is a 12V or 16V TVS directly on the input which also gives you some protection against being plugged into the wrong voltage power supply and will forward bias if a wrong supply is plugged in. I’d use something like an SMCJ16A usually. That’ll work well saving you with smaller current-limited power supplies (3-4A rated) but not against all situations. For running it off a battery where the fault currents would be higher, I’d add something like a 2A schottky diode in series with the input before the TVS, I usually use an SS14 but there are tons of equivalents.
There are methods using a FET that can work well, as well. There are standalone IC solutions that can work really well be start to get expensive, etc. It’s just risk vs reward. For a one-off that I’m wiring myself, I’d keep it simple/cheap and keep up the momentum. For my designs that I make a couple k of at a time and that get wired up by technicians at the company that I sell them to? That gets a slightly more ‘robust’ set of protection components…
As with everything, you sure can and there’s no wrong answer. The key questions are why do that? What does it take to do that, time wise? How much more does it cost? Are there any downsides?
Honestly, for your first design, I’d say to keep the momentum high, keep the design as simple as possible and make sure you’re continuously making tangible progress. You’ll be way better off getting a cheap/simple controller together quickly and testing that to find out what you need next than spending months doing a kitchen-sink build that ends up needing revision anyway.
I mean this in the kindest way possible: They will get it right, 99.9% of the time. It’s YOU that will get it wrong. I guarantee it, especially if you keep adding complexity. No amount of time spent looking at it, 3rd party review etc. leads to a perfect board every time, so you have to hope for the best (no issues) but plan for the worst (replacing components, cutting/rewiring traces).
If you want to get into it and learn to replace components with a high skill ceiling, go for it!
If you don’t want to get into anything too complex but are ok with basic stuff then keep the parts you’re using large/simple and plan for a revision if you get something major wrong.
If you don’t want to do any of that, plan for a 2nd revision to get manufactured and just assume it’ll happen. Another reason to keep the design cheap/simple at first and evolve it, perhaps.
I really can’t stress this enough: MINIMUM VIABLE… Plan to add stuff later, or add it in the lightest touch way possible. If you have to make another little adapter board to add the 3rd thermocouple or airspeed sensor or add a powered USB hub to power the webcam then all of that is a better approach, especially if this is a ‘just for you’ design and not something you’re going to sell.
